<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:media="http://search.yahoo.com/mrss/"><channel><title>Vulnerability Management | FirewallSync</title><link>https://firewallsync.com/tags/vulnerability-management/</link><description>Practical cybersecurity writing for engineering teams that ship fast.</description><language>en-us</language><image><url>https://firewallsync.com/icon-512.png</url><title>FirewallSync</title><link>https://firewallsync.com/</link></image><lastBuildDate>Sat, 21 Nov 2026 04:30:00 +0000</lastBuildDate><atom:link href="https://firewallsync.com/tags/vulnerability-management/index.xml" rel="self" type="application/rss+xml"/><item><title>CSAF VEX Remediations and Threats: What Trivy Gets Wrong</title><link>https://firewallsync.com/posts/csaf-vex-remediations-threats-category-product-ids-trivy/</link><guid>https://firewallsync.com/posts/csaf-vex-remediations-threats-category-product-ids-trivy/</guid><pubDate>Sat, 21 Nov 2026 04:30:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>CSAF 2.0 puts the fix, the workaround, the restart cost and the reason a product is not affected in two arrays: remediations and threats. Here is what each field means, how to resolve them per product, and three places where Trivy v0.76.2 and gocsaf v3.7.0 diverge from the spec.</description><enclosure url="https://firewallsync.com/posts/csaf-vex-remediations-threats-category-product-ids-trivy/cover_hu12283328291714519248.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/csaf-vex-remediations-threats-category-product-ids-trivy/cover_hu12283328291714519248.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/csaf-vex-remediations-threats-category-product-ids-trivy/cover_hu12283328291714519248.jpg" alt="A set of chrome combination wrenches in a black rack, lying on rusty metal in a sunlit workshop" width="1200" height="630"></p><p><strong>Short answer:</strong> In a CSAF 2.0 VEX document, <code>product_status</code> says <em>whether</em> each product is affected. Two arrays under each vulnerability say <em>what to do about it</em>: <strong><code>remediations</code></strong> (category, details, date, URL, entitlements, restart cost) and <strong><code>threats</code></strong> (impact, exploit status, target set). Both apply to products only through their own <code>product_ids</code> or <code>group_ids</code>, so you have to resolve them per product. Trivy v0.76.2 does not: it ignores remediations, takes the first <code>impact</code> threat for every product it filters, and, through the gocsaf v3.7.0 library, fails to decode any file that uses <code>product_groups</code>.</p>
<p>If you consume CSAF VEX from a vendor such as Red Hat or from an internal team, these two arrays are where the patch-or-wait decision lives. This article reads the <a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/prose/csaf-v2-editor-draft.md">CSAF 2.0 specification</a> and <a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/json_schema/csaf_json_schema.json">JSON schema</a>, <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/csaf.go">Trivy v0.76.2</a> and <a href="https://github.com/gocsaf/csaf/blob/v3.7.0/csaf/advisory.go">gocsaf v3.7.0</a>, the version pinned in Trivy&rsquo;s go.mod. For resolving product IDs to purls first, see <a href="/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/">CSAF VEX product_tree</a>; this piece picks up where that one stops.</p>
<h2 id="remediations-the-fields">Remediations: the fields</h2>
<p>Every remediation item has two mandatory properties, <code>category</code> and <code>details</code>. The spec adds: &ldquo;At least one of the optional elements Group IDs (<code>group_ids</code>) and Product IDs (<code>product_ids</code>) MUST be present to state for which products or product groups this remediation is applicable.&rdquo; A remediation with neither fails mandatory test 6.1.29, <em>Remediation without Product Reference</em>.</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Type</th>
          <th>What it tells you</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>category</code></td>
          <td>enum (required)</td>
          <td>What kind of action this is (table below)</td>
      </tr>
      <tr>
          <td><code>details</code></td>
          <td>string (required)</td>
          <td>&ldquo;a thorough human-readable discussion of the remediation&rdquo;</td>
      </tr>
      <tr>
          <td><code>product_ids</code> / <code>group_ids</code></td>
          <td>ID lists</td>
          <td>Which products the remediation covers; at least one must be present</td>
      </tr>
      <tr>
          <td><code>date</code></td>
          <td>date-time</td>
          <td>&ldquo;the date from which the remediation is available&rdquo;</td>
      </tr>
      <tr>
          <td><code>url</code></td>
          <td>URI</td>
          <td>&ldquo;the URL where to obtain the remediation&rdquo;</td>
      </tr>
      <tr>
          <td><code>entitlements</code></td>
          <td>string list</td>
          <td>&ldquo;vendor-defined constraints for obtaining fixed software or hardware&rdquo;, such as a support contract</td>
      </tr>
      <tr>
          <td><code>restart_required</code></td>
          <td>object</td>
          <td><code>category</code> (required) and <code>details</code>: what must restart for the remediation to take effect</td>
      </tr>
  </tbody>
</table>
<h3 id="the-five-remediation-categories">The five remediation categories</h3>
<table>
  <thead>
      <tr>
          <th><code>category</code></th>
          <th>Spec meaning (CSAF 2.0, 3.2.3.12.1)</th>
          <th>Ticket action</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>vendor_fix</code></td>
          <td>&ldquo;an official fix that is issued by the original author of the affected product. Unless otherwise noted, it is assumed that this fix fully resolves the vulnerability.&rdquo;</td>
          <td>Schedule the upgrade; use <code>url</code> and <code>date</code></td>
      </tr>
      <tr>
          <td><code>workaround</code></td>
          <td>a configuration or deployment scenario &ldquo;that can be used to avoid exposure to the vulnerability&rdquo;</td>
          <td>Apply now, keep the ticket open for the fix</td>
      </tr>
      <tr>
          <td><code>mitigation</code></td>
          <td>&ldquo;helps to reduce the risk of the vulnerability but that does not resolve the vulnerability on the affected product&rdquo;</td>
          <td>Compensating control; the finding stays open</td>
      </tr>
      <tr>
          <td><code>none_available</code></td>
          <td>&ldquo;there is currently no fix or other remediation available&rdquo;</td>
          <td>Track; the spec suggests the issuer include an expected date</td>
      </tr>
      <tr>
          <td><code>no_fix_planned</code></td>
          <td>no fix &ldquo;and it is not planned to provide one at any time&rdquo;, often for end-of-life products</td>
          <td>Replace, isolate or accept the risk</td>
      </tr>
  </tbody>
</table>
<p>Two consistency rules matter for parsers. <code>vendor_fix</code> &ldquo;contradicts with the categories <code>none_available</code> and <code>no_fix_planned</code> for the same product&rdquo;, so a document should never carry both for one product. And the difference between <code>workaround</code> and <code>mitigation</code> is the difference between &ldquo;avoids exposure&rdquo; and &ldquo;reduces risk&rdquo;: do not close a finding on a mitigation.</p>
<h3 id="restart_required">restart_required</h3>
<p>This is the field operations teams want and scanners rarely show. CSAF 2.0 defines nine values: <code>none</code>, <code>vulnerable_component</code>, <code>service</code>, <code>parent</code>, <code>dependencies</code>, <code>connected</code>, <code>machine</code>, <code>zone</code> and <code>system</code>. They escalate from &ldquo;No restart required&rdquo; through <code>machine</code> (&ldquo;This is the value which SHOULD be used if an OS needs to be restarted&rdquo;) to <code>system</code>, a restart of &ldquo;the whole system&rdquo;, which &ldquo;MAY include multiple security zones&rdquo;. If you build change tickets automatically, map <code>machine</code>, <code>zone</code> and <code>system</code> to a maintenance window rather than a rolling restart.</p>
<h2 id="threats-impact-exploit_status-target_set">Threats: impact, exploit_status, target_set</h2>
<p>Threat items also need <code>category</code> and <code>details</code>, take optional <code>date</code>, <code>product_ids</code> and <code>group_ids</code>, and describe information that &ldquo;can change as the vulnerability ages&rdquo;. There are three categories:</p>
<ul>
<li><strong><code>impact</code></strong>: an assessment of the impact if the vulnerability is exploited &ldquo;or a description why it cannot be exploited&rdquo;.</li>
<li><strong><code>exploit_status</code></strong>: &ldquo;the degree to which an exploit for the vulnerability is known&rdquo;, which the spec says can mirror CVSS exploitability or use a contextual status such as &ldquo;Weaponized&rdquo;.</li>
<li><strong><code>target_set</code></strong>: &ldquo;the currently known victim population&rdquo;, for example platform, product type, user segment or geography.</li>
</ul>
<p>For threats, <code>date</code> is &ldquo;the date when the assessment was done or the threat appeared&rdquo;, so an old <code>exploit_status</code> is a reason to check <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV</a> and EPSS rather than trust it.</p>
<h3 id="what-the-vex-profile-requires">What the VEX profile requires</h3>
<p>The <code>csaf_vex</code> profile (section 4.5) ties both arrays to product status:</p>
<ul>
<li>Every product in <code>known_not_affected</code> needs an impact statement, &ldquo;as machine readable flag in <code>/vulnerabilities[]/flags</code> or as human readable justification in <code>/vulnerabilities[]/threats</code>&rdquo;. For a threat, &ldquo;the <code>category</code> value for such a statement MUST be <code>impact</code>&rdquo;.</li>
<li>Every product in <code>known_affected</code> needs &ldquo;additional product specific information&rdquo; in <code>remediations</code> &ldquo;as an action statement&rdquo;. <code>no_fix_planned</code> and <code>none_available</code> are allowed.</li>
<li>Group IDs may be used, but &ldquo;it MUST be ensured that for each Product ID the required information according to its product status as stated in the two points above is available&rdquo;.</li>
</ul>
<p>That last sentence is the important one: the requirement is per product. A vulnerability can, and in multi-product advisories usually does, carry several impact threats with different <code>product_ids</code>.</p>
<h2 id="resolving-remediations-and-threats-per-product">Resolving remediations and threats per product</h2>
<p>The file below is an illustrative CSAF 2.0 VEX document: one CVE (a placeholder ID, CVE-2099-0001), two container images, one product group. <code>example-worker</code> is affected and has a vendor fix; <code>example-api</code> is not affected, and its impact threat says why. It validates with zero errors against the CSAF 2.0 JSON schema (Python <code>jsonschema</code>, with the FIRST CVSS schemas the CSAF repository ships under <code>csaf_2.0/referenced_schema</code>).</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span><span style="color:#e6db74">&#34;product_groups&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> [
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;group_id&#34;</span>: <span style="color:#e6db74">&#34;CSAFGID-ALL&#34;</span>, <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-API&#34;</span>, <span style="color:#e6db74">&#34;CSAFPID-WORKER&#34;</span>]}
</span></span><span style="display:flex;"><span>]<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span><span style="color:#960050;background-color:#1e0010">...</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;product_status&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> {
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;known_affected&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-WORKER&#34;</span>],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;known_not_affected&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-API&#34;</span>]
</span></span><span style="display:flex;"><span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;threats&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> [
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;impact&#34;</span>, <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-WORKER&#34;</span>],
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;example-worker parses uploaded XML with libexample; a crafted file can crash the worker.&#34;</span>},
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;impact&#34;</span>, <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-API&#34;</span>],
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;example-api ships libexample but never calls its XML parser.&#34;</span>},
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;exploit_status&#34;</span>, <span style="color:#f92672">&#34;group_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFGID-ALL&#34;</span>],
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;No public exploit known to Example Corp.&#34;</span>, <span style="color:#f92672">&#34;date&#34;</span>: <span style="color:#e6db74">&#34;2026-11-21T00:00:00Z&#34;</span>}
</span></span><span style="display:flex;"><span>]<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;remediations&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> [
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;vendor_fix&#34;</span>, <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFPID-WORKER&#34;</span>],
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;Upgrade to example-worker 1.8.1.&#34;</span>, <span style="color:#f92672">&#34;date&#34;</span>: <span style="color:#e6db74">&#34;2026-11-21T00:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;url&#34;</span>: <span style="color:#e6db74">&#34;https://example.com/releases/worker-1.8.1&#34;</span>,
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;restart_required&#34;</span>: {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;service&#34;</span>, <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;Restart the worker deployment.&#34;</span>}},
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;mitigation&#34;</span>, <span style="color:#f92672">&#34;group_ids&#34;</span>: [<span style="color:#e6db74">&#34;CSAFGID-ALL&#34;</span>],
</span></span><span style="display:flex;"><span>   <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;Reject XML uploads at the gateway until upgraded.&#34;</span>}
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>This Python 3 script (standard library only) expands <code>group_ids</code> and prints, for each product and status, the threats and remediations that actually apply to it:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">expand</span>(item, groups):
</span></span><span style="display:flex;"><span>    ids <span style="color:#f92672">=</span> set(item<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_ids&#34;</span>, []))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> g <span style="color:#f92672">in</span> item<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;group_ids&#34;</span>, []):
</span></span><span style="display:flex;"><span>        ids <span style="color:#f92672">|=</span> set(groups<span style="color:#f92672">.</span>get(g, []))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> ids
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>doc <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>groups <span style="color:#f92672">=</span> {g[<span style="color:#e6db74">&#34;group_id&#34;</span>]: g[<span style="color:#e6db74">&#34;product_ids&#34;</span>]
</span></span><span style="display:flex;"><span>          <span style="color:#66d9ef">for</span> g <span style="color:#f92672">in</span> doc[<span style="color:#e6db74">&#34;product_tree&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_groups&#34;</span>, [])}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> vuln <span style="color:#f92672">in</span> doc[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    status <span style="color:#f92672">=</span> vuln<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_status&#34;</span>, {})
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> state <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;known_affected&#34;</span>, <span style="color:#e6db74">&#34;known_not_affected&#34;</span>, <span style="color:#e6db74">&#34;fixed&#34;</span>, <span style="color:#e6db74">&#34;under_investigation&#34;</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> pid <span style="color:#f92672">in</span> status<span style="color:#f92672">.</span>get(state, []):
</span></span><span style="display:flex;"><span>            rem <span style="color:#f92672">=</span> [r <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> vuln<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;remediations&#34;</span>, []) <span style="color:#66d9ef">if</span> pid <span style="color:#f92672">in</span> expand(r, groups)]
</span></span><span style="display:flex;"><span>            thr <span style="color:#f92672">=</span> [t <span style="color:#66d9ef">for</span> t <span style="color:#f92672">in</span> vuln<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;threats&#34;</span>, []) <span style="color:#66d9ef">if</span> pid <span style="color:#f92672">in</span> expand(t, groups)]
</span></span><span style="display:flex;"><span>            print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>vuln<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;cve&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>pid<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>state<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> t <span style="color:#f92672">in</span> thr:
</span></span><span style="display:flex;"><span>                print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;  threat  </span><span style="color:#e6db74">{</span>t[<span style="color:#e6db74">&#39;category&#39;</span>]<span style="color:#e6db74">:</span><span style="color:#e6db74">&lt;15</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>t[<span style="color:#e6db74">&#39;details&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> rem:
</span></span><span style="display:flex;"><span>                restart <span style="color:#f92672">=</span> r<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;restart_required&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;category&#34;</span>, <span style="color:#e6db74">&#34;-&#34;</span>)
</span></span><span style="display:flex;"><span>                print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;  remedy  </span><span style="color:#e6db74">{</span>r[<span style="color:#e6db74">&#39;category&#39;</span>]<span style="color:#e6db74">:</span><span style="color:#e6db74">&lt;15</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> date=</span><span style="color:#e6db74">{</span>r<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;date&#39;</span>, <span style="color:#e6db74">&#39;-&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74"> restart=</span><span style="color:#e6db74">{</span>restart<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>Run against the full example file, it prints:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2099-0001 CSAFPID-WORKER known_affected
</span></span><span style="display:flex;"><span>  threat  impact          example-worker parses uploaded XML with libexample; a crafted file can crash the worker.
</span></span><span style="display:flex;"><span>  threat  exploit_status  No public exploit known to Example Corp.
</span></span><span style="display:flex;"><span>  remedy  vendor_fix      date=2026-11-21T00:00:00Z restart=service
</span></span><span style="display:flex;"><span>  remedy  mitigation      date=- restart=-
</span></span><span style="display:flex;"><span>CVE-2099-0001 CSAFPID-API known_not_affected
</span></span><span style="display:flex;"><span>  threat  impact          example-api ships libexample but never calls its XML parser.
</span></span><span style="display:flex;"><span>  threat  exploit_status  No public exploit known to Example Corp.
</span></span><span style="display:flex;"><span>  remedy  mitigation      date=- restart=-
</span></span></code></pre></div><p>The script reads only <code>product_ids</code> and <code>group_ids</code> on plain products. Products defined through <code>relationships</code> need the full-product-name resolution described in the <a href="/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/">product_tree article</a> first. Treat it as a starting point, not a complete CSAF consumer.</p>
<h2 id="where-trivy-v0762-diverges">Where Trivy v0.76.2 diverges</h2>
<p>Trivy is a common first consumer of CSAF VEX files. Reading its source at v0.76.2, and running the same decoding path against the files above, shows three gaps. We could not run Trivy itself in our build environment; instead we compiled gocsaf v3.7.0&rsquo;s <code>csaf</code> package from its source files, unmodified, and called it the way Trivy&rsquo;s <code>decodeCSAF</code> does (<code>json.NewDecoder(r).Decode(&amp;adv)</code>), together with a Go port of Trivy&rsquo;s <code>statement()</code> function.</p>
<h3 id="1-files-with-product_groups-do-not-decode">1. Files with product_groups do not decode</h3>
<p>The CSAF 2.0 schema defines <code>product_tree.product_groups</code> as <code>&quot;type&quot;: &quot;array&quot;</code> of group objects. gocsaf v3.7.0 declares the field as <code>ProductGroups *ProductGroups</code>, where <code>ProductGroups</code> is a struct with a <code>product_group_ids</code> list. Decoding the schema-valid example fails:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>decode error: json: cannot unmarshal array into Go struct field ProductTree.product_tree.product_groups of type csaf.ProductGroups
</span></span></code></pre></div><p>In Trivy, <code>decodeVEX</code> collects that error, <code>NewDocument</code> returns it as &ldquo;unable to load VEX from file&rdquo;, and the VEX filter wraps it as a &ldquo;VEX error&rdquo;. None of the file&rsquo;s statements are applied. The struct is unchanged in gocsaf v3.7.1 and on the main branch as of November 21, 2026, and Trivy&rsquo;s main branch still pins v3.7.0. Until that changes, a workaround is to expand groups into <code>product_ids</code> before handing the file to Trivy; the same document with groups expanded decoded cleanly.</p>
<h3 id="2-one-impact-threat-for-every-filtered-product">2. One impact threat for every filtered product</h3>
<p>When Trivy filters a finding, it attaches a statement taken by <code>statement()</code>, which returns the <code>details</code> of the first threat whose category is <code>impact</code>, and does not look at that threat&rsquo;s <code>product_ids</code>. With groups expanded, our example decodes, and Trivy would filter <code>example-api</code> (known_not_affected) with this statement:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>example-worker parses uploaded XML with libexample; a crafted file can crash the worker.
</span></span></code></pre></div><p>That is the impact text for the <em>affected</em> product, recorded as the justification for suppressing the unaffected one. The suppression itself is correct (it comes from <code>product_status</code>), but anyone auditing modified findings reads the wrong reason. Order your impact threats with this in mind if Trivy is your consumer, and do not use Trivy&rsquo;s statement field as your audit record for multi-product files.</p>
<h3 id="3-remediations-are-never-read">3. Remediations are never read</h3>
<p>Trivy&rsquo;s matcher builds its status map from <code>known_not_affected</code> and <code>fixed</code> only. <code>known_affected</code> findings pass through unchanged, and nothing in <code>pkg/vex/csaf.go</code> reads <code>remediations</code>. The vendor fix URL, the availability <code>date</code>, <code>entitlements</code> and <code>restart_required</code> never reach the scan output. If you want them in tickets, parse them separately, as above, and join on CVE and product.</p>
<h2 id="csaf-21-adds-two-remediation-categories">CSAF 2.1 adds two remediation categories</h2>
<p>The OASIS CSAF TC&rsquo;s repository now carries CSAF 2.1 prose headed &ldquo;Committee Specification 01&rdquo;, dated 9 November 2026; it is not yet an OASIS Standard. Its schema has seven remediation categories, adding:</p>
<ul>
<li><strong><code>fix_planned</code></strong>: &ldquo;there is a fix for the vulnerability planned but not yet ready&rdquo;. This takes over the &ldquo;fix is being developed&rdquo; use that CSAF 2.0 put under <code>none_available</code>.</li>
<li><strong><code>optional_patch</code></strong>: a patch from the original author whose &ldquo;application is not necessary, but might be desired by the user, e.g. to calm a security scanner&rdquo;. The spec adds: &ldquo;Unless otherwise noted, it is assumed that this does not change the state regarding the vulnerability.&rdquo;</li>
</ul>
<p><code>optional_patch</code> is the category a vendor would attach to a <code>known_not_affected</code> product that still ships a flagged library version. Do not treat it as a <code>vendor_fix</code>.</p>
<p>Tooling has not caught up. gocsaf v3.7.0 accepts only <code>csaf_version</code> &ldquo;2.0&rdquo; and the five 2.0 remediation categories; in our test, a file declaring 2.1 failed with <code>2.1 not in [2.0]</code>, and a 2.0 file with a <code>fix_planned</code> remediation failed with <code>fix_planned not in [mitigation|no_fix_planned|none_available|vendor_fix|workaround]</code>. Both also fail the CSAF 2.0 JSON schema, as expected.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Applying a remediation to every product in the vulnerability.</strong> Remediations and threats are scoped by their own <code>product_ids</code> and <code>group_ids</code>, not by <code>product_status</code>.</li>
<li><strong>Closing a finding on <code>mitigation</code> or <code>workaround</code>.</strong> Only <code>vendor_fix</code> is assumed to resolve the flaw, and only once you have applied it.</li>
<li><strong>Forgetting group expansion.</strong> A product named only through a group ID still has its remediation. A parser that reads only <code>product_ids</code> misses it.</li>
<li><strong>Reading an old exploit_status as current.</strong> Check its <code>date</code> against fresher exploitation signals.</li>
</ul>
<h2 id="where-this-fits">Where this fits</h2>
<p>Remediations and threats answer the last questions in the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">CVE-record-to-ticket workflow</a>: once VEX tells you a product is affected, what is the fix, when is it available and what will it cost to apply. For the meaning of the VEX statuses themselves, see <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explained</a>; for the equivalent OpenVEX fields, see <a href="/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/">OpenVEX statement fields</a>.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>The Trivy findings come from reading v0.76.2 source and running gocsaf v3.7.0&rsquo;s decoding code, not from running the Trivy binary. Behaviour in later Trivy releases may differ.</li>
<li>The example document is illustrative; CVE-2099-0001 and Example Corp are placeholders.</li>
<li>CSAF 2.1 text is quoted from the TC&rsquo;s Committee Specification 01 draft in the oasis-tcs/csaf repository and may change before it becomes an OASIS Standard.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Remediation and threat fields, required properties, five categories, restart_required values, field meanings</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/prose/csaf-v2-editor-draft.md">CSAF 2.0 prose, 3.2.3.12 and 3.2.3.14</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>VEX profile impact and action statement rules; tests 6.1.27.9, 6.1.27.10, 6.1.29</td>
          <td>Same document, sections 4.5 and 6.1</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>product_groups is an array in the schema; example validates</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/json_schema/csaf_json_schema.json">CSAF 2.0 JSON schema</a>, run with Python jsonschema</td>
          <td>Verified (run)</td>
      </tr>
      <tr>
          <td>Trivy acts on known_not_affected and fixed; statement() takes first impact threat; decodeCSAF uses json.Decoder</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/csaf.go">Trivy v0.76.2 pkg/vex/csaf.go</a>, <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/document.go">document.go</a></td>
          <td>Verified (source)</td>
      </tr>
      <tr>
          <td>Trivy pins gocsaf v3.7.0; ProductGroups is a struct; version and category enums</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/go.mod">Trivy go.mod</a>, <a href="https://github.com/gocsaf/csaf/blob/v3.7.0/csaf/advisory.go">gocsaf v3.7.0 advisory.go</a></td>
          <td>Verified (source, run)</td>
      </tr>
      <tr>
          <td>CSAF 2.1 CS01 dated 9 November 2026; fix_planned and optional_patch text</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.1/prose/share/csaf-v2.1-draft.md">oasis-tcs/csaf csaf_2.1 prose</a></td>
          <td>Qualified (Committee Specification, not OASIS Standard)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>OpenVEX Statement Fields: What Trivy and go-vex Ignore</title><link>https://firewallsync.com/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/</link><guid>https://firewallsync.com/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/</guid><pubDate>Fri, 20 Nov 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>AppSec &amp; APIs</category><description>OpenVEX v0.2.0 defines 13 statement fields, but Trivy v0.76.2 acts on two of them: status and justification, from the matching statement with the newest timestamp. Here is what happens to status_notes, action_statement, last_updated, version and supplier, read from the go-vex v0.2.9 and Trivy source.</description><enclosure url="https://firewallsync.com/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/cover_hu9992264959083435414.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/cover_hu9992264959083435414.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/openvex-statement-fields-status-notes-action-statement-version-trivy/cover_hu9992264959083435414.jpg" alt="Black-framed reading glasses resting on a printed document, the text blurred beneath the lenses" width="1200" height="630"></p><p><strong>Short answer:</strong> OpenVEX v0.2.0 gives a statement 13 fields, from <code>status</code> and <code>justification</code> to <code>status_notes</code>, <code>action_statement</code>, <code>last_updated</code> and a statement-level <code>version</code>. Trivy v0.76.2, through the go-vex v0.2.9 library, acts on two of them once a statement matches: <strong><code>status</code></strong> and <strong><code>justification</code></strong>. It picks the winning statement by <strong><code>timestamp</code></strong> alone. Everything else is either decoded and ignored, or, for statement <code>version</code> and <code>supplier</code>, not decoded at all and lost when a go-vex tool rewrites the file.</p>
<p>That matters if you write VEX for your own images and expect a scanner to honour it. A field you set to change the outcome may have no effect, and a field you rely on as an audit trail may vanish on the next <code>vexctl add</code>.</p>
<p>This article reads the <a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX v0.2.0 specification</a>, its <a href="https://github.com/openvex/spec/blob/main/openvex_json_schema.json">JSON schema</a>, <a href="https://github.com/openvex/go-vex/tree/v0.2.9">go-vex v0.2.9</a> (the newest tag on November 20, 2026, and the version pinned in Trivy&rsquo;s go.mod) and <a href="https://github.com/aquasecurity/trivy/tree/v0.76.2">Trivy v0.76.2</a>. For what the statuses mean, start with <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explained</a>; for how products and purls match, see <a href="/posts/openvex-products-subcomponents-purl-matching-trivy/">OpenVEX products and subcomponents</a>.</p>
<h2 id="field-by-field-spec-schema-go-vex-trivy">Field by field: spec, schema, go-vex, Trivy</h2>
<table>
  <thead>
      <tr>
          <th>Statement field</th>
          <th>Spec v0.2.0</th>
          <th>JSON schema</th>
          <th>go-vex v0.2.9</th>
          <th>Trivy v0.76.2</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>vulnerability</code></td>
          <td>required</td>
          <td>required</td>
          <td>decoded</td>
          <td>matched against the finding ID</td>
      </tr>
      <tr>
          <td><code>products</code> / <code>subcomponents</code></td>
          <td>required for a complete statement, can be inherited</td>
          <td>optional</td>
          <td>decoded</td>
          <td>matched against the scanned purls</td>
      </tr>
      <tr>
          <td><code>status</code></td>
          <td>required</td>
          <td>required</td>
          <td>decoded, validated by <code>Validate()</code></td>
          <td><strong>acts on it</strong>: suppresses <code>not_affected</code> and <code>fixed</code></td>
      </tr>
      <tr>
          <td><code>justification</code></td>
          <td><code>not_affected</code> needs it or <code>impact_statement</code></td>
          <td>same rule (<code>anyOf</code>)</td>
          <td>decoded</td>
          <td><strong>copied</strong> into the modified finding&rsquo;s <code>Statement</code></td>
      </tr>
      <tr>
          <td><code>timestamp</code></td>
          <td>optional, inherits from the document</td>
          <td>optional</td>
          <td>decoded, <strong>used to sort</strong></td>
          <td>the newest matching statement wins</td>
      </tr>
      <tr>
          <td><code>last_updated</code></td>
          <td>optional</td>
          <td>optional</td>
          <td>decoded, not compared</td>
          <td>ignored</td>
      </tr>
      <tr>
          <td><code>status_notes</code></td>
          <td>MAY</td>
          <td>optional</td>
          <td>decoded</td>
          <td>ignored</td>
      </tr>
      <tr>
          <td><code>impact_statement</code></td>
          <td>alternative to <code>justification</code></td>
          <td>alternative</td>
          <td>decoded</td>
          <td>ignored</td>
      </tr>
      <tr>
          <td><code>action_statement</code></td>
          <td>MUST for <code>affected</code></td>
          <td>required for <code>affected</code></td>
          <td>decoded; <code>Validate()</code> enforces it</td>
          <td>ignored, not validated</td>
      </tr>
      <tr>
          <td><code>action_statement_timestamp</code></td>
          <td>optional</td>
          <td>optional</td>
          <td>decoded</td>
          <td>ignored</td>
      </tr>
      <tr>
          <td><code>version</code> (statement)</td>
          <td>optional, &ldquo;Defaults to zero&rdquo;</td>
          <td>integer, <code>minimum: 1</code></td>
          <td><strong>no struct field: dropped</strong></td>
          <td>not seen</td>
      </tr>
      <tr>
          <td><code>supplier</code> (statement)</td>
          <td>optional</td>
          <td>(not listed)</td>
          <td><strong>no struct field: dropped</strong></td>
          <td>not seen</td>
      </tr>
  </tbody>
</table>
<p>&ldquo;Ignored&rdquo; means the field takes no part in whether Trivy suppresses a finding or in what it records about the suppression. The data is still in your file; it is just not consumed.</p>
<h2 id="how-trivy-decides">How Trivy decides</h2>
<p>Trivy&rsquo;s OpenVEX handler is short. In <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/openvex.go"><code>pkg/vex/openvex.go</code></a>, <code>NotAffected</code> asks go-vex for the matching statements, then:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-go" data-lang="go"><span style="display:flex;"><span><span style="color:#75715e">// Take the latest statement for a given vulnerability and product
</span></span></span><span style="display:flex;"><span><span style="color:#75715e">// as a sequence of statements can be overridden by the newer one.
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span><span style="color:#a6e22e">stmt</span> <span style="color:#f92672">:=</span> <span style="color:#a6e22e">stmts</span>[len(<span style="color:#a6e22e">stmts</span>)<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>]
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> <span style="color:#a6e22e">stmt</span>.<span style="color:#a6e22e">Status</span> <span style="color:#f92672">==</span> <span style="color:#a6e22e">openvex</span>.<span style="color:#a6e22e">StatusNotAffected</span> <span style="color:#f92672">||</span> <span style="color:#a6e22e">stmt</span>.<span style="color:#a6e22e">Status</span> <span style="color:#f92672">==</span> <span style="color:#a6e22e">openvex</span>.<span style="color:#a6e22e">StatusFixed</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#a6e22e">modifiedFindings</span> <span style="color:#f92672">:=</span> <span style="color:#a6e22e">types</span>.<span style="color:#a6e22e">NewModifiedFinding</span>(<span style="color:#a6e22e">vuln</span>, <span style="color:#a6e22e">findingStatus</span>(<span style="color:#a6e22e">stmt</span>.<span style="color:#a6e22e">Status</span>), string(<span style="color:#a6e22e">stmt</span>.<span style="color:#a6e22e">Justification</span>), <span style="color:#a6e22e">v</span>.<span style="color:#a6e22e">source</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#a6e22e">modifiedFindings</span>, <span style="color:#66d9ef">true</span>
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>Two consequences follow directly from that code:</p>
<ol>
<li><strong>Only <code>status</code> changes the result.</strong> <code>affected</code> and <code>under_investigation</code> leave the finding in place, whatever <code>action_statement</code> or <code>status_notes</code> say.</li>
<li><strong>Only <code>justification</code> is carried into the output.</strong> The third argument to <code>NewModifiedFinding</code> becomes the <code>Statement</code> field of the finding listed under <code>ExperimentalModifiedFindings</code> in Trivy&rsquo;s JSON report. A <code>not_affected</code> statement that explains itself only with <code>impact_statement</code> is valid, yet Trivy records an empty <code>Statement</code> for it. If an auditor reads Trivy&rsquo;s report rather than your VEX file, the reasoning is gone.</li>
</ol>
<h2 id="which-statement-wins-timestamp-not-last_updated-or-version">Which statement wins: timestamp, not last_updated or version</h2>
<p>The list Trivy indexes into is sorted by go-vex&rsquo;s <code>SortStatements</code> in <a href="https://github.com/openvex/go-vex/blob/v0.2.9/pkg/vex/statement.go"><code>pkg/vex/statement.go</code></a>. For statements about the same vulnerability it compares <code>Timestamp</code>, and a statement with no timestamp of its own uses the document&rsquo;s:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-go" data-lang="go"><span style="display:flex;"><span><span style="color:#a6e22e">iTime</span> <span style="color:#f92672">:=</span> <span style="color:#a6e22e">stmts</span>[<span style="color:#a6e22e">i</span>].<span style="color:#a6e22e">Timestamp</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> <span style="color:#a6e22e">iTime</span> <span style="color:#f92672">==</span> <span style="color:#66d9ef">nil</span> <span style="color:#f92672">||</span> <span style="color:#a6e22e">iTime</span>.<span style="color:#a6e22e">IsZero</span>() {
</span></span><span style="display:flex;"><span>    <span style="color:#a6e22e">iTime</span> = <span style="color:#f92672">&amp;</span><span style="color:#a6e22e">documentTimestamp</span>
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span><span style="color:#75715e">// ...
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span><span style="color:#66d9ef">return</span> <span style="color:#a6e22e">iTime</span>.<span style="color:#a6e22e">Before</span>(<span style="color:#f92672">*</span><span style="color:#a6e22e">jTime</span>)
</span></span></code></pre></div><p><code>last_updated</code> is never compared, and statement <code>version</code> does not exist in the struct. The sort is stable, so two statements with equal timestamps stay in file order and the later one wins.</p>
<p>This follows the spec, which defines <code>timestamp</code> as &ldquo;the time at which the information expressed in the Statement was known to be true&rdquo; and says a statement &ldquo;is useless without a timestamp as it cannot be related to others talking about the same subject.&rdquo; The trap is editing a statement in place. If you change an old <code>not_affected</code> statement&rsquo;s text and bump its <code>last_updated</code> and <code>version</code>, it keeps its old <code>timestamp</code> and still loses to any newer statement for the same product and CVE.</p>
<p>The spec&rsquo;s own update example shows the safe pattern: keep the old statement unchanged with its original timestamp, and add a new statement that carries the new status and a newer timestamp (its own, or the bumped document timestamp it inherits). <code>vexctl add</code> does this for you, as covered in our <a href="/posts/vexctl-create-add-merge-attest-openvex-documents/">vexctl guide</a>.</p>
<h2 id="statement-version-and-supplier-are-dropped-on-rewrite">Statement version and supplier are dropped on rewrite</h2>
<p>The spec lists <code>version</code> on the statement as an &ldquo;Optional integer representing the statement&rsquo;s version number. Defaults to zero, required when incremented&rdquo;, and <code>supplier</code> as the &ldquo;Supplier of the product or subcomponent.&rdquo; The go-vex v0.2.9 <code>Statement</code> struct has fields for <code>@id</code>, <code>vulnerability</code>, <code>timestamp</code>, <code>last_updated</code>, <code>products</code>, <code>status</code>, <code>status_notes</code>, <code>justification</code>, <code>impact_statement</code>, <code>action_statement</code> and <code>action_statement_timestamp</code>, and nothing else. Go&rsquo;s <code>encoding/json</code> discards keys it has no field for.</p>
<p>We checked the effect with a stdlib-only Go 1.24.7 program whose <code>Statement</code> struct copies go-vex v0.2.9&rsquo;s field set and JSON tags (with <code>products</code> and <code>vulnerability</code> held as raw JSON). Input:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>:<span style="color:#ae81ff">2</span>,<span style="color:#f92672">&#34;supplier&#34;</span>:<span style="color:#e6db74">&#34;Example Corp&#34;</span>,<span style="color:#f92672">&#34;vulnerability&#34;</span>:{<span style="color:#f92672">&#34;name&#34;</span>:<span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},
</span></span><span style="display:flex;"><span> <span style="color:#f92672">&#34;timestamp&#34;</span>:<span style="color:#e6db74">&#34;2026-11-01T10:00:00Z&#34;</span>,<span style="color:#f92672">&#34;last_updated&#34;</span>:<span style="color:#e6db74">&#34;2026-11-20T08:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span> <span style="color:#f92672">&#34;products&#34;</span>:[{<span style="color:#f92672">&#34;@id&#34;</span>:<span style="color:#e6db74">&#34;pkg:oci/app@sha256:abc&#34;</span>}],<span style="color:#f92672">&#34;status&#34;</span>:<span style="color:#e6db74">&#34;affected&#34;</span>,
</span></span><span style="display:flex;"><span> <span style="color:#f92672">&#34;status_notes&#34;</span>:<span style="color:#e6db74">&#34;Reachable from the upload handler&#34;</span>,<span style="color:#f92672">&#34;action_statement&#34;</span>:<span style="color:#e6db74">&#34;Upgrade git to 2.45.1&#34;</span>,
</span></span><span style="display:flex;"><span> <span style="color:#f92672">&#34;action_statement_timestamp&#34;</span>:<span style="color:#e6db74">&#34;2026-11-20T08:00:00Z&#34;</span>}
</span></span></code></pre></div><p>Output after <code>json.Unmarshal</code> and <code>json.Marshal</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;vulnerability&#34;</span>:{<span style="color:#f92672">&#34;name&#34;</span>:<span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},<span style="color:#f92672">&#34;timestamp&#34;</span>:<span style="color:#e6db74">&#34;2026-11-01T10:00:00Z&#34;</span>,<span style="color:#f92672">&#34;last_updated&#34;</span>:<span style="color:#e6db74">&#34;2026-11-20T08:00:00Z&#34;</span>,<span style="color:#f92672">&#34;products&#34;</span>:[{<span style="color:#f92672">&#34;@id&#34;</span>:<span style="color:#e6db74">&#34;pkg:oci/app@sha256:abc&#34;</span>}],<span style="color:#f92672">&#34;status&#34;</span>:<span style="color:#e6db74">&#34;affected&#34;</span>,<span style="color:#f92672">&#34;status_notes&#34;</span>:<span style="color:#e6db74">&#34;Reachable from the upload handler&#34;</span>,<span style="color:#f92672">&#34;action_statement&#34;</span>:<span style="color:#e6db74">&#34;Upgrade git to 2.45.1&#34;</span>,<span style="color:#f92672">&#34;action_statement_timestamp&#34;</span>:<span style="color:#e6db74">&#34;2026-11-20T08:00:00Z&#34;</span>}
</span></span></code></pre></div><p><code>version</code> and <code>supplier</code> are gone. Any tool that loads a document through go-vex and writes it back, including <code>vexctl add</code> and <code>vexctl merge</code>, will do the same. If your process uses statement <code>version</code> as a change counter, keep that history somewhere go-vex does not touch, such as the document <code>version</code> (which go-vex does keep) or your Git history.</p>
<p>The spec and schema also disagree slightly here: the spec says statement <code>version</code> defaults to zero, while the schema sets <code>minimum: 1</code>. A statement with an explicit <code>&quot;version&quot;: 0</code> fails schema validation.</p>
<p>The openvex/go-vex main branch has an unreleased validator (commit 42a83ee, November 19, 2026) whose field table carries the same warning for both fields: &ldquo;go-vex does not decode it; it is dropped when the document is rewritten.&rdquo; The vexctl main branch already wires it into a <code>vexctl validate</code> command; the latest vexctl release, v0.4.4, does not include it.</p>
<h2 id="action_statement-required-by-the-spec-not-checked-by-trivy">action_statement: required by the spec, not checked by Trivy</h2>
<p>The spec says an <code>affected</code> statement &ldquo;MUST include a statement that SHOULD describe actions to remediate or mitigate the vulnerability,&rdquo; and the JSON schema makes <code>action_statement</code> required when <code>status</code> is <code>affected</code>. go-vex enforces this in <code>Statement.Validate()</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-go" data-lang="go"><span style="display:flex;"><span><span style="color:#75715e">// action statement is now required
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span><span style="color:#66d9ef">if</span> <span style="color:#a6e22e">v</span> <span style="color:#f92672">:=</span> <span style="color:#a6e22e">stmt</span>.<span style="color:#a6e22e">ActionStatement</span>; <span style="color:#a6e22e">v</span> <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;&#34;</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#a6e22e">fmt</span>.<span style="color:#a6e22e">Errorf</span>(<span style="color:#e6db74">&#34;action statement must be set when using status %q&#34;</span>, <span style="color:#a6e22e">s</span>)
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>Trivy never calls it. Its <code>decodeOpenVEX</code> in <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/document.go"><code>pkg/vex/document.go</code></a> runs <code>json.NewDecoder(r).Decode(&amp;openVEX)</code> and only checks that <code>@context</code> is not empty. go-vex&rsquo;s own <code>Parse</code> also just unmarshals. A non-conforming <code>affected</code> statement loads without complaint, and since Trivy does not suppress <code>affected</code> findings, the scan output looks identical either way.</p>
<p>On the producer side, vexctl (v0.4.4 and main) sets the <code>--action-statement</code> flag&rsquo;s default to the go-vex constant <code>NoActionStatementMsg</code>, which is the string <code>&quot;No action statement provided&quot;</code>. An <code>affected</code> statement created without <code>-a</code> therefore passes validation with a placeholder that tells a reader nothing. Search your VEX files for that string before publishing them.</p>
<h2 id="a-pre-publish-check-you-can-run">A pre-publish check you can run</h2>
<p>The script below flags the cases above. It is plain Python 3 with no dependencies; we ran it with Python 3 against the sample document that follows. It checks fields, not purl matching, so pair it with schema validation.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Flag OpenVEX v0.2.0 statement fields that Trivy v0.76.2 / go-vex v0.2.9 will not act on.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> datetime
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">ts</span>(v):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> datetime<span style="color:#f92672">.</span>fromisoformat(v<span style="color:#f92672">.</span>replace(<span style="color:#e6db74">&#34;Z&#34;</span>, <span style="color:#e6db74">&#34;+00:00&#34;</span>)) <span style="color:#66d9ef">if</span> v <span style="color:#66d9ef">else</span> <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>doc <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>doc_ts <span style="color:#f92672">=</span> doc<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;timestamp&#34;</span>)
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> i, s <span style="color:#f92672">in</span> enumerate(doc<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;statements&#34;</span>, [])):
</span></span><span style="display:flex;"><span>    where <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;statements[</span><span style="color:#e6db74">{</span>i<span style="color:#e6db74">}</span><span style="color:#e6db74">] </span><span style="color:#e6db74">{</span>s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;vulnerability&#39;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;name&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;status&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>    st <span style="color:#f92672">=</span> s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;status&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;version&#34;</span> <span style="color:#f92672">in</span> s:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: statement &#39;version&#39; is dropped when go-vex tools rewrite the file&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;supplier&#34;</span> <span style="color:#f92672">in</span> s:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: statement &#39;supplier&#39; is dropped when go-vex tools rewrite the file&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> st <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;affected&#34;</span> <span style="color:#f92672">and</span> <span style="color:#f92672">not</span> s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;action_statement&#34;</span>):
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: affected without action_statement (spec MUST; Trivy loads it anyway)&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> st <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;affected&#34;</span> <span style="color:#f92672">and</span> s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;action_statement&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;No action statement provided&#34;</span>:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: action_statement is the vexctl placeholder&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> st <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;not_affected&#34;</span> <span style="color:#f92672">and</span> <span style="color:#f92672">not</span> s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;justification&#34;</span>):
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: not_affected with impact_statement only; Trivy reports an empty Statement&#34;</span>)
</span></span><span style="display:flex;"><span>    eff <span style="color:#f92672">=</span> ts(s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;timestamp&#34;</span>) <span style="color:#f92672">or</span> doc_ts)
</span></span><span style="display:flex;"><span>    lu <span style="color:#f92672">=</span> ts(s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;last_updated&#34;</span>))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> lu <span style="color:#f92672">and</span> eff <span style="color:#f92672">and</span> lu <span style="color:#f92672">&gt;</span> eff:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>where<span style="color:#e6db74">}</span><span style="color:#e6db74">: last_updated </span><span style="color:#e6db74">{</span>s[<span style="color:#e6db74">&#39;last_updated&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> is newer than the timestamp used for ordering (</span><span style="color:#e6db74">{</span>eff<span style="color:#f92672">.</span>isoformat()<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span></code></pre></div><p>Sample document (illustrative names and digest). The author edited the second statement on November 19, bumping <code>last_updated</code> and <code>version</code>, and expects <code>not_affected</code> to win:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@context&#34;</span>: <span style="color:#e6db74">&#34;https://openvex.dev/ns/v0.2.0&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;https://example.com/vex/app-2026-11-20&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;author&#34;</span>: <span style="color:#e6db74">&#34;Example Corp Product Security&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-11-20T08:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#ae81ff">3</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;statements&#34;</span>: [
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [{<span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/app@sha256:abc&#34;</span>}],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-11-01T10:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;action_statement&#34;</span>: <span style="color:#e6db74">&#34;No action statement provided&#34;</span>
</span></span><span style="display:flex;"><span>    },
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#ae81ff">2</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [{<span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/app@sha256:abc&#34;</span>}],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-10-15T10:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;last_updated&#34;</span>: <span style="color:#e6db74">&#34;2026-11-19T12:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;impact_statement&#34;</span>: <span style="color:#e6db74">&#34;The vulnerable clone path is never called&#34;</span>
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>  ]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>Output:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>statements[0] CVE-2024-32002 affected: action_statement is the vexctl placeholder
</span></span><span style="display:flex;"><span>statements[1] CVE-2024-32002 not_affected: statement &#39;version&#39; is dropped when go-vex tools rewrite the file
</span></span><span style="display:flex;"><span>statements[1] CVE-2024-32002 not_affected: not_affected with impact_statement only; Trivy reports an empty Statement
</span></span><span style="display:flex;"><span>statements[1] CVE-2024-32002 not_affected: last_updated 2026-11-19T12:00:00Z is newer than the timestamp used for ordering (2026-10-15T10:00:00+00:00)
</span></span></code></pre></div><p>By go-vex&rsquo;s sort, the <code>affected</code> statement (timestamp November 1) is newer than the <code>not_affected</code> one (October 15), so Trivy keeps reporting the CVE. The fix is a new <code>not_affected</code> statement with a current timestamp and a <code>justification</code> label, not an edit to the old one.</p>
<h2 id="what-to-do">What to do</h2>
<ul>
<li><strong>Supersede, do not edit.</strong> Add a new statement with a newer <code>timestamp</code> for every status change. Treat <code>last_updated</code> as a note for humans.</li>
<li><strong>Always set <code>justification</code> on <code>not_affected</code>.</strong> Use <code>impact_statement</code> as extra detail, not as the only explanation, if a scanner report is what auditors see.</li>
<li><strong>Write real action statements.</strong> Reject <code>&quot;No action statement provided&quot;</code> in CI. Trivy will not use the text, but customers and your own triage queue will.</li>
<li><strong>Keep statement <code>version</code> and <code>supplier</code> out of go-vex round trips,</strong> or accept that they will be lost. Put supplier data on the product or subcomponent purl instead.</li>
<li><strong>Validate against the schema in CI.</strong> Neither Trivy nor go-vex&rsquo;s <code>Parse</code> will reject a non-conforming file for you. The <code>openvex_json_schema.json</code> file catches the missing <code>action_statement</code> and a statement <code>version</code> of 0.</li>
</ul>
<p>VEX is one input to the patch decision, alongside KEV, EPSS and SSVC. Our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a> shows where a <code>not_affected</code> statement should close a ticket, and where it should not.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>This describes Trivy v0.76.2 and go-vex v0.2.9 as of November 20, 2026, read from source; we did not run Trivy itself. Other consumers (Grype, Docker Scout, vendor platforms) may read <code>status_notes</code> or <code>action_statement</code>, or sort on different fields. The go-vex validator and <code>vexctl validate</code> mentioned above are unreleased and may change before a tagged release. The Go round-trip test used a copy of go-vex&rsquo;s struct definition, not the library itself.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Statement field list, MUST/MAY wording, timestamp inheritance, update example</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX spec v0.2.0</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td><code>action_statement</code> required for <code>affected</code>; statement <code>version</code> minimum 1</td>
          <td><a href="https://github.com/openvex/spec/blob/main/openvex_json_schema.json">openvex_json_schema.json</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Statement struct fields; <code>SortStatements</code> compares timestamps only; <code>Validate()</code> rules</td>
          <td><a href="https://github.com/openvex/go-vex/blob/v0.2.9/pkg/vex/statement.go">go-vex v0.2.9 statement.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Trivy takes the last matching statement, suppresses only <code>not_affected</code>/<code>fixed</code>, passes <code>justification</code> as Statement</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/openvex.go">Trivy v0.76.2 openvex.go</a>, <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/types/finding.go">finding.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Trivy decodes OpenVEX without validation; pins go-vex v0.2.9</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/document.go">Trivy v0.76.2 document.go</a>, <a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/go.mod">go.mod</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>vexctl&rsquo;s default action statement placeholder</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/internal/cmd/options.go">vexctl options.go</a>, <a href="https://github.com/openvex/go-vex/blob/v0.2.9/pkg/vex/vex.go">go-vex vex.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Unreleased validator warns that statement <code>version</code> and <code>supplier</code> are dropped</td>
          <td><a href="https://github.com/openvex/go-vex/blob/main/pkg/validate/schema.go">go-vex main pkg/validate/schema.go</a></td>
          <td>Qualified (unreleased)</td>
      </tr>
      <tr>
          <td><code>version</code> and <code>supplier</code> dropped on a JSON round trip</td>
          <td>Our Go 1.24.7 test with a copy of the go-vex struct</td>
          <td>Verified (reproduction, not the library)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Public PoC CVEs: Pull All 47,926 SSVC poc Records</title><link>https://firewallsync.com/posts/public-poc-cves-cisa-adp-ssvc-exploitation-poc-epss/</link><guid>https://firewallsync.com/posts/public-poc-cves-cisa-adp-ssvc-exploitation-poc-epss/</guid><pubDate>Thu, 19 Nov 2026 04:20:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>CISA-ADP marks 47,926 CVE records with SSVC Exploitation poc, meaning a public proof of concept or a well-known exploitation method. We pulled them from the cvelistV5 corpus, joined EPSS and KEV, and show what the label does and does not tell you, with a script that writes the full list to CSV.</description><enclosure url="https://firewallsync.com/posts/public-poc-cves-cisa-adp-ssvc-exploitation-poc-epss/cover_hu9432398450888018006.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/public-poc-cves-cisa-adp-ssvc-exploitation-poc-epss/cover_hu9432398450888018006.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/public-poc-cves-cisa-adp-ssvc-exploitation-poc-epss/cover_hu9432398450888018006.jpg" alt="A brass key hanging on a ring from the keyhole of a white door, above a black iron handle" width="1200" height="630"></p><p><strong>Short answer:</strong> CISA&rsquo;s Vulnrichment program writes an SSVC <code>Exploitation</code> value into the CISA-ADP container of the CVE records it scores. The value <code>poc</code> means a public proof of concept or a well-known exploitation method exists. In the cvelistV5 snapshot of 2026-11-18 00:00 UTC, <strong>47,926</strong> records have it. That makes it the largest machine-readable list of &ldquo;public exploit available&rdquo; CVEs you can download for free, but it is a capability label, not a priority queue: only 2.6% of those CVEs score 0.1 or higher in EPSS.</p>
<p>This article shows how to pull the list, what the counts look like next to KEV and EPSS, and how to cut it down to something a team can act on. For what each SSVC field means and how CISA&rsquo;s decision tree turns them into Track, Track*, Attend or Act, see our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC guide</a>. For where this fits in a patch decision from CVE record to ticket, see the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>.</p>
<h2 id="what-poc-means-in-the-specs-words">What poc means, in the spec&rsquo;s words</h2>
<p>The SSVC project at Carnegie Mellon SEI defines the <code>Exploitation</code> decision point in <a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/exploitation.py"><code>exploitation.py</code></a>. It has two versions, and the meaning of the middle value narrowed between them.</p>
<table>
  <thead>
      <tr>
          <th>Version</th>
          <th>Value</th>
          <th>Definition (verbatim)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>1.0.0</td>
          <td>PoC</td>
          <td>&ldquo;One of the following cases is true: (1) private evidence of exploitation is attested but not shared; (2) widespread hearsay attests to exploitation; (3) typical public PoC in places such as Metasploit or ExploitDB; or (4) the vulnerability has a well-known method of exploitation.&rdquo;</td>
      </tr>
      <tr>
          <td>1.1.0 (latest)</td>
          <td>Public PoC</td>
          <td>&ldquo;One of the following is true: (1) Typical public PoC exists in sources such as Metasploit or websites like ExploitDB; or (2) the vulnerability has a well-known method of exploitation.&rdquo;</td>
      </tr>
      <tr>
          <td>both</td>
          <td>Active</td>
          <td>&ldquo;Shared, observable, reliable evidence that the exploit is being used in the wild by real attackers; there is credible public reporting.&rdquo;</td>
      </tr>
      <tr>
          <td>both</td>
          <td>None</td>
          <td>&ldquo;There is no evidence of active exploitation and no public proof of concept (PoC) of how to exploit the vulnerability.&rdquo;</td>
      </tr>
  </tbody>
</table>
<p>CISA&rsquo;s <a href="https://github.com/cisagov/vulnrichment">Vulnrichment README</a> describes a <code>poc</code> value as &ldquo;indicating there was a public proof-of-concept available at the time of analysis&rdquo;. Two practical points follow:</p>
<ul>
<li><strong>&ldquo;Well-known method of exploitation&rdquo; counts.</strong> A <code>poc</code> label does not guarantee a working exploit sits on GitHub or Exploit-DB. Some records get it because the bug class is trivially exploitable.</li>
<li><strong>&ldquo;At the time of analysis.&rdquo;</strong> The SSVC block carries a <code>timestamp</code>. A <code>none</code> scored a year ago says nothing about the PoC that appeared last week.</li>
</ul>
<p>The CVE record does not say which <code>Exploitation</code> version CISA applied. Every CISA-ADP SSVC block in our snapshot carries <code>&quot;version&quot;: &quot;2.0.3&quot;</code>, which is a version string for the SSVC block, not for the decision point.</p>
<h2 id="where-it-sits-in-the-record">Where it sits in the record</h2>
<p>The value lives in the CISA-ADP container, as one of three single-key objects in <code>options</code>. This is the example from CISA&rsquo;s own README (CVE-2024-25522):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span><span style="color:#e6db74">&#34;options&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> [
</span></span><span style="display:flex;"><span>  { <span style="color:#f92672">&#34;Exploitation&#34;</span>: <span style="color:#e6db74">&#34;poc&#34;</span> },
</span></span><span style="display:flex;"><span>  { <span style="color:#f92672">&#34;Automatable&#34;</span>: <span style="color:#e6db74">&#34;yes&#34;</span> },
</span></span><span style="display:flex;"><span>  { <span style="color:#f92672">&#34;Technical Impact&#34;</span>: <span style="color:#e6db74">&#34;total&#34;</span> }
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>The path is <code>containers.adp[]</code> where <code>providerMetadata.shortName</code> is <code>CISA-ADP</code>, then <code>metrics[].other</code> where <code>type</code> is <code>ssvc</code>, then <code>content.options</code>. The <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">CVE ADP containers guide</a> covers the container itself.</p>
<h2 id="the-numbers-poc-next-to-kev-and-epss">The numbers: poc next to KEV and EPSS</h2>
<p>We counted every CISA-ADP SSVC block in the cvelistV5 release <code>2026-11-18_all_CVEs_at_midnight.zip.zip</code>, then joined the KEV catalog (version 2026.11.18, 1,786 entries) and the EPSS scores for 2026-11-18 (model <code>v2026.06.15</code>). Every SSVC record had an EPSS score.</p>
<table>
  <thead>
      <tr>
          <th>Exploitation</th>
          <th style="text-align: right">Records</th>
          <th style="text-align: right">Share</th>
          <th style="text-align: right">In KEV</th>
          <th style="text-align: right">Median EPSS</th>
          <th style="text-align: right">EPSS ≥ 0.1</th>
          <th style="text-align: right">EPSS ≥ 0.5</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>none</td>
          <td style="text-align: right">173,136</td>
          <td style="text-align: right">77.7%</td>
          <td style="text-align: right">3</td>
          <td style="text-align: right">0.00361</td>
          <td style="text-align: right">2,026</td>
          <td style="text-align: right">598</td>
      </tr>
      <tr>
          <td>poc</td>
          <td style="text-align: right">47,926</td>
          <td style="text-align: right">21.5%</td>
          <td style="text-align: right">1</td>
          <td style="text-align: right">0.00478</td>
          <td style="text-align: right">1,231</td>
          <td style="text-align: right">308</td>
      </tr>
      <tr>
          <td>active</td>
          <td style="text-align: right">1,787</td>
          <td style="text-align: right">0.8%</td>
          <td style="text-align: right">1,782</td>
          <td style="text-align: right">0.53535</td>
          <td style="text-align: right">1,348</td>
          <td style="text-align: right">916</td>
      </tr>
      <tr>
          <td><strong>Total</strong></td>
          <td style="text-align: right"><strong>222,849</strong></td>
          <td style="text-align: right"></td>
          <td style="text-align: right"><strong>1,786</strong></td>
          <td style="text-align: right"></td>
          <td style="text-align: right"></td>
          <td style="text-align: right"></td>
      </tr>
  </tbody>
</table>
<p>Three things stand out.</p>
<p><strong>1. poc is not a queue for KEV.</strong> You might expect a pool of poc CVEs waiting to be added to KEV. The current data cannot show that, because CISA moves a record to <code>active</code> when it joins the catalog. 1,782 of the 1,786 KEV entries read <code>active</code>. The other four (CVE-2024-10441, CVE-2026-40217, CVE-2026-90817 and CVE-2026-90436) all have a KEV <code>dateAdded</code> of 2026-11-18, the same day as the snapshot, so their CISA-ADP blocks had not caught up yet. The single poc record in KEV is one of them. To measure how many poc CVEs later become KEV entries you need two points in time; our <a href="/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/">CVE exploitation alerts</a> post diffs snapshots that way.</p>
<p><strong>2. active and KEV are not identical.</strong> Five records read <code>active</code> but are not in the current KEV catalog: CVE-2023-24932, CVE-2022-21894, CVE-2021-30120, CVE-2021-30119 and CVE-2021-21983. Treat the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed</a> as the authority for &ldquo;known exploited&rdquo; and the ADP value as a second opinion.</p>
<p><strong>3. Most poc CVEs score low in EPSS.</strong> The poc median (0.00478) is only slightly above the none median (0.00361). EPSS estimates the probability of exploitation activity in the next 30 days, and a published PoC raises that only somewhat on its own. Read our <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS feed guide</a> before comparing these numbers to a percentile threshold.</p>
<p>There is also a coverage gap to keep in mind. Of the 17,558 CVEs with an EPSS score of 0.1 or higher on 2026-11-18, 12,953 have no CISA-ADP SSVC block at all. CISA scores records as they are published and has not scored the back catalog, so an absent <code>poc</code> does not mean &ldquo;no PoC&rdquo;.</p>
<h2 id="slice-poc-by-the-other-two-decision-points">Slice poc by the other two decision points</h2>
<p><code>Exploitation</code> is one of three values CISA writes. Splitting the poc set by the other two gives smaller groups with different EPSS profiles:</p>
<table>
  <thead>
      <tr>
          <th>Automatable</th>
          <th>Technical Impact</th>
          <th style="text-align: right">poc records</th>
          <th style="text-align: right">Median EPSS</th>
          <th style="text-align: right">EPSS ≥ 0.1</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>yes</td>
          <td>total</td>
          <td style="text-align: right">5,328</td>
          <td style="text-align: right">0.00891</td>
          <td style="text-align: right">394</td>
      </tr>
      <tr>
          <td>no</td>
          <td>total</td>
          <td style="text-align: right">9,701</td>
          <td style="text-align: right">0.00628</td>
          <td style="text-align: right">377</td>
      </tr>
      <tr>
          <td>yes</td>
          <td>partial</td>
          <td style="text-align: right">9,887</td>
          <td style="text-align: right">0.00516</td>
          <td style="text-align: right">165</td>
      </tr>
      <tr>
          <td>no</td>
          <td>partial</td>
          <td style="text-align: right">23,010</td>
          <td style="text-align: right">0.00385</td>
          <td style="text-align: right">295</td>
      </tr>
  </tbody>
</table>
<p>The poc / yes / total slice (11.1% of poc records) has about twice the median EPSS of the full poc set. It is also the combination CISA&rsquo;s own decision tree treats most seriously short of active exploitation: our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment guide</a> shows it at Attend for high mission and well-being impact.</p>
<p>The highest-scoring poc CVEs are not in KEV. CVE-2023-27372 (EPSS 0.99682) and CVE-2025-1974 (0.99400) are both poc / yes / total and absent from catalog version 2026.11.18. In the 2026-11-18 data, 308 poc CVEs score 0.5 or higher. That small set is where poc earns its keep: a public PoC, a high EPSS score and no KEV entry yet.</p>
<h2 id="pull-the-list-yourself">Pull the list yourself</h2>
<p>The script below reads the cvelistV5 daily release zip and an EPSS daily file, and writes every poc CVE to CSV. It uses only the Python standard library.</p>
<p>Inputs:</p>
<ul>
<li><strong>CVE corpus:</strong> the <code>&lt;date&gt;_all_CVEs_at_midnight.zip.zip</code> asset from a <code>cve_&lt;date&gt;_0000Z</code> release of <a href="https://github.com/CVEProject/cvelistV5/releases">CVEProject/cvelistV5</a> (about 660 MB; it holds an inner <code>cves.zip</code>). Some days the asset is missing; take the previous day&rsquo;s.</li>
<li><strong>EPSS:</strong> a daily <code>epss_scores-YYYY-MM-DD.csv.gz</code> from the <a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a> repository. Its first line is a <code>#model_version</code> comment, which the script skips.</li>
</ul>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;List CVEs that CISA-ADP marks SSVC Exploitation &#39;poc&#39;, joined to EPSS.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">usage: public_poc.py &lt;cvelistV5 *_all_CVEs_at_midnight.zip.zip&gt; &lt;epss_scores-YYYY-MM-DD.csv.gz&gt; &gt; poc.csv&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> gzip<span style="color:#f92672">,</span> io<span style="color:#f92672">,</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> zipfile
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cisa_ssvc</span>(record):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> record<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;containers&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []) <span style="color:#f92672">or</span> []:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;providerMetadata&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;shortName&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> metric <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []) <span style="color:#f92672">or</span> []:
</span></span><span style="display:flex;"><span>            other <span style="color:#f92672">=</span> metric<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;other&#34;</span>, {})
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> other<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;type&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;ssvc&#34;</span>:
</span></span><span style="display:flex;"><span>                content <span style="color:#f92672">=</span> other<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;content&#34;</span>, {})
</span></span><span style="display:flex;"><span>                opts <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">for</span> d <span style="color:#f92672">in</span> content<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;options&#34;</span>, []):
</span></span><span style="display:flex;"><span>                    opts<span style="color:#f92672">.</span>update(d)  <span style="color:#75715e"># options is a list of one-key objects</span>
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">return</span> content<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;timestamp&#34;</span>), opts
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#66d9ef">None</span>, <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>epss <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> gzip<span style="color:#f92672">.</span>open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>], <span style="color:#e6db74">&#34;rt&#34;</span>) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>    next(f)  <span style="color:#75715e"># first line is &#34;#model_version:...,score_date:...&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> row <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(f):
</span></span><span style="display:flex;"><span>        epss[row[<span style="color:#e6db74">&#34;cve&#34;</span>]] <span style="color:#f92672">=</span> (row[<span style="color:#e6db74">&#34;epss&#34;</span>], row[<span style="color:#e6db74">&#34;percentile&#34;</span>])
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>outer <span style="color:#f92672">=</span> zipfile<span style="color:#f92672">.</span>ZipFile(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>])
</span></span><span style="display:flex;"><span>inner <span style="color:#f92672">=</span> zipfile<span style="color:#f92672">.</span>ZipFile(io<span style="color:#f92672">.</span>BytesIO(outer<span style="color:#f92672">.</span>read(outer<span style="color:#f92672">.</span>namelist()[<span style="color:#ae81ff">0</span>])))
</span></span><span style="display:flex;"><span>out <span style="color:#f92672">=</span> csv<span style="color:#f92672">.</span>writer(sys<span style="color:#f92672">.</span>stdout)
</span></span><span style="display:flex;"><span>out<span style="color:#f92672">.</span>writerow([<span style="color:#e6db74">&#34;cve&#34;</span>, <span style="color:#e6db74">&#34;datePublished&#34;</span>, <span style="color:#e6db74">&#34;ssvc_timestamp&#34;</span>, <span style="color:#e6db74">&#34;automatable&#34;</span>,
</span></span><span style="display:flex;"><span>              <span style="color:#e6db74">&#34;technical_impact&#34;</span>, <span style="color:#e6db74">&#34;epss&#34;</span>, <span style="color:#e6db74">&#34;epss_percentile&#34;</span>])
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> name <span style="color:#f92672">in</span> inner<span style="color:#f92672">.</span>namelist():
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;/CVE-&#34;</span> <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> name <span style="color:#f92672">or</span> <span style="color:#f92672">not</span> name<span style="color:#f92672">.</span>endswith(<span style="color:#e6db74">&#34;.json&#34;</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>loads(inner<span style="color:#f92672">.</span>read(name))
</span></span><span style="display:flex;"><span>    ts, opts <span style="color:#f92672">=</span> cisa_ssvc(rec)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> opts <span style="color:#f92672">or</span> opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Exploitation&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;poc&#34;</span>:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    cve <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>]
</span></span><span style="display:flex;"><span>    e, p <span style="color:#f92672">=</span> epss<span style="color:#f92672">.</span>get(cve, (<span style="color:#e6db74">&#34;&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>))
</span></span><span style="display:flex;"><span>    out<span style="color:#f92672">.</span>writerow([cve, rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;datePublished&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>), ts,
</span></span><span style="display:flex;"><span>                  opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Automatable&#34;</span>), opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Technical Impact&#34;</span>), e, p])
</span></span></code></pre></div><p>We ran it with Python 3 against the 2026-11-18 release and EPSS file: it took about 37 seconds and wrote 47,926 rows. It loads the inner zip into memory, so give it at least 2.5 GB of RAM (peak resident memory was about 2.2 GB in our run). To keep the list current between full runs, apply cvelistV5&rsquo;s hourly changes as described in our <a href="/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/">delta.json and deltaLog.json guide</a>.</p>
<p>Then filter. For example, poc / yes / total with EPSS at or above 0.1 that is not in KEV:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> json
</span></span><span style="display:flex;"><span>kev <span style="color:#f92672">=</span> {v[<span style="color:#e6db74">&#34;cveID&#34;</span>] <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> json<span style="color:#f92672">.</span>load(open(<span style="color:#e6db74">&#34;known_exploited_vulnerabilities.json&#34;</span>))[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>rows <span style="color:#f92672">=</span> [r <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(open(<span style="color:#e6db74">&#34;poc.csv&#34;</span>))
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> r[<span style="color:#e6db74">&#34;automatable&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;yes&#34;</span> <span style="color:#f92672">and</span> r[<span style="color:#e6db74">&#34;technical_impact&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;total&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#f92672">and</span> float(r[<span style="color:#e6db74">&#34;epss&#34;</span>] <span style="color:#f92672">or</span> <span style="color:#ae81ff">0</span>) <span style="color:#f92672">&gt;=</span> <span style="color:#ae81ff">0.1</span> <span style="color:#f92672">and</span> r[<span style="color:#e6db74">&#34;cve&#34;</span>] <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> kev]
</span></span><span style="display:flex;"><span>print(len(rows))
</span></span></code></pre></div><p>On the 2026-11-18 data this prints 394, since no poc / yes / total record is in KEV. The 0.1 cut-off is an example, not a recommendation; pick a threshold you can staff.</p>
<h2 id="how-to-use-poc-in-a-patch-decision">How to use poc in a patch decision</h2>
<ul>
<li><strong>Never as the only gate.</strong> 47,926 CVEs is a backlog, not a priority list. Match the list against your inventory first; the share that applies to you will be far smaller.</li>
<li><strong>Rank inside it.</strong> Within matched poc CVEs, sort by Automatable and Technical Impact, then EPSS, then your exposure (internet-facing first).</li>
<li><strong>KEV beats poc.</strong> A KEV entry means confirmed exploitation and, for US federal agencies, a due date. Check the KEV feed before the poc list.</li>
<li><strong>Watch for upgrades.</strong> A move from <code>none</code> to <code>poc</code>, or from <code>poc</code> to <code>active</code>, is the event worth an alert. Store the value you last saw and compare.</li>
<li><strong>Check the timestamp.</strong> A <code>none</code> with an old SSVC <code>timestamp</code> on a CVE that is now in the news may simply be stale.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>The counts describe one snapshot: cvelistV5 at 2026-11-18 00:00 UTC, KEV version 2026.11.18 and EPSS scores of 2026-11-18. They move daily.</li>
<li>A current snapshot cannot tell you how often poc CVEs become KEV entries, because CISA rewrites the value to active. That needs historical snapshots.</li>
<li>CISA-ADP covers records CISA has scored, mostly those published or updated since 2024. Older CVEs and many 2022-2023 ones have no SSVC block.</li>
<li>We did not check whether the PoCs behind individual poc labels work, or where they are published.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Exploitation definitions, versions 1.0.0 and 1.1.0</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/exploitation.py">CERTCC/SSVC exploitation.py</a></td>
          <td>Verified (verbatim)</td>
      </tr>
      <tr>
          <td>poc means a public PoC &ldquo;at the time of analysis&rdquo;; CVE-2024-25522 example</td>
          <td><a href="https://github.com/cisagov/vulnrichment">cisagov/vulnrichment README</a></td>
          <td>Verified (verbatim)</td>
      </tr>
      <tr>
          <td>222,849 SSVC records; 47,926 poc, 1,787 active, 173,136 none</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/releases">cvelistV5 release, 2026-11-18 midnight</a>, counted with the script above</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>KEV: 1,786 entries; 1,782 active; four same-day additions; five active not in KEV</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a>, catalog 2026.11.18, joined to the corpus</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>EPSS medians and threshold counts</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a>, 2026-11-18 file, model v2026.06.15</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>EPSS estimates exploitation activity over the next 30 days</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores README</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>KEV dueDate: 3, 14 or 21 Days After dateAdded, 2021-2026</title><link>https://firewallsync.com/posts/kev-duedate-days-after-dateadded-3-14-21-day-deadlines/</link><guid>https://firewallsync.com/posts/kev-duedate-days-after-dateadded-3-14-21-day-deadlines/</guid><pubDate>Wed, 18 Nov 2026 04:58:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>We measured dueDate minus dateAdded for all 1,782 entries in CISA's KEV catalog (version 2026.11.17) and replayed the cisagov/kev-data history to find every dueDate CISA edited after publication. The default gap was 14 days in 2021, 21 days from March 2022, and since mid-2026 is 3 or 14 days. Every forensicTriage Yes entry is due in 3 days, but 23 No entries are too.</description><enclosure url="https://firewallsync.com/posts/kev-duedate-days-after-dateadded-3-14-21-day-deadlines/cover_hu5441541435602915000.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/kev-duedate-days-after-dateadded-3-14-21-day-deadlines/cover_hu5441541435602915000.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/kev-duedate-days-after-dateadded-3-14-21-day-deadlines/cover_hu5441541435602915000.jpg" alt="A wall calendar page for December 2020 in four languages, with week numbers down the left and weekend dates in grey and red" width="1200" height="630"></p><p><strong>Short answer:</strong> the KEV catalog&rsquo;s <code>dueDate</code> is a calendar date, set per entry, and the gap to <code>dateAdded</code> has changed three times. In the current catalog (version 2026.11.17, 1,782 entries) every entry added since June 10, 2026 is due <strong>3 or 14 days</strong> after it was added. Before that, from March 2022 until early March 2026, the usual gap was <strong>21 days</strong>. If you build SLAs from KEV, read <code>dueDate</code> from every catalog sync rather than hard-coding a number of days, because CISA also edits due dates after publishing them.</p>
<p>This article measures the field. For what each KEV field means, see our <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed field guide</a>. For what BOD 26-04 requires of federal agencies, see <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">CISA BOD 26-04 explained</a>. For where KEV fits in a patch decision, see the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>.</p>
<h2 id="what-the-field-is">What the field is</h2>
<p>The catalog schema in <a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a> describes <code>dueDate</code> as &ldquo;The date the required action is due in the format YYYY-MM-DD&rdquo;. It is required on every entry. A second field, <code>forensicTriage</code>, is described as &ldquo;&lsquo;Yes&rsquo; if this vulnerability requires forensic triage per BOD 26-04; &lsquo;No&rsquo; if forensic triage is not required per BOD 26-04&rdquo;. That field first appears in the catalog published on September 1, 2026, when it was filled in for all entries then listed (38 Yes, 1,649 No).</p>
<p>Neither field states the number of days. To get it, subtract <code>dateAdded</code> from <code>dueDate</code>.</p>
<h2 id="gap-by-year-added">Gap by year added</h2>
<p>We computed <code>dueDate − dateAdded</code> in days for all 1,782 entries in catalog version 2026.11.17 (released 2026-11-17T17:53Z). The three most common gaps per year:</p>
<table>
  <thead>
      <tr>
          <th>Year added</th>
          <th>Entries</th>
          <th>Most common gaps (days: entries)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2021</td>
          <td>311</td>
          <td>181: 189 · 14: 111 · 182: 11</td>
      </tr>
      <tr>
          <td>2022</td>
          <td>555</td>
          <td>21: 403 · 14: 88 · 181: 49</td>
      </tr>
      <tr>
          <td>2023</td>
          <td>187</td>
          <td>21: 177 · 4: 7 · 7: 1</td>
      </tr>
      <tr>
          <td>2024</td>
          <td>186</td>
          <td>21: 175 · 7: 7 · 12: 2</td>
      </tr>
      <tr>
          <td>2025</td>
          <td>245</td>
          <td>21: 226 · 7: 12 · 1: 6</td>
      </tr>
      <tr>
          <td>2026</td>
          <td>298</td>
          <td>3: 159 · 14: 92 · 21: 44</td>
      </tr>
  </tbody>
</table>
<p>Three regimes are visible:</p>
<ol>
<li><strong>2021 to early 2022: two weeks or six months.</strong> <a href="https://www.cisa.gov/binding-operational-directive-22-01">BOD 22-01</a> told agencies to remediate &ldquo;within 6 months for vulnerabilities with a Common Vulnerabilities and Exposures (CVE) ID assigned prior to 2021&rdquo; and &ldquo;within two weeks for all other vulnerabilities&rdquo;, and added: &ldquo;These default timelines may be adjusted in the case of grave risk to the Federal Enterprise.&rdquo; The 2021 data follows that split: 179 of the 2021 entries with pre-2021 CVE IDs have a 181-day gap and 11 have 182; 110 entries with 2021 CVE IDs have 14 days.</li>
<li><strong>March 2022 to March 2026: 21 days.</strong> The first 21-day entry was added on 2022-03-03 and the last on 2026-03-05. The last six-month entry was added on 2022-03-07 and the last 2022 entry with 14 days on 2022-06-08. We did not find a CISA statement explaining the switch from two weeks to 21 days, so we report it only as what the data shows.</li>
<li><strong>2026: 3 and 14 days.</strong> The first 3-day entry was added on 2026-01-27 (Fortinet, CVE-2026-24858), months before BOD 26-04 was published on June 10, 2026. The first 14-day entry in this regime was added on 2026-03-09. Since June 10, 2026 no other gap appears.</li>
</ol>
<p>Short one-off gaps exist outside those patterns, for example six 1-day entries added in 2025 and two 2-day Cisco entries (CVE-2022-20775 and CVE-2026-20127) added on 2026-02-25.</p>
<h2 id="since-bod-26-04-3-or-14-days-and-what-forensictriage-tells-you">Since BOD 26-04: 3 or 14 days, and what forensicTriage tells you</h2>
<p>For the 165 entries added on or after June 10, 2026:</p>
<table>
  <thead>
      <tr>
          <th>Gap</th>
          <th>forensicTriage</th>
          <th>Entries</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>3 days</td>
          <td>Yes</td>
          <td>105</td>
      </tr>
      <tr>
          <td>3 days</td>
          <td>No</td>
          <td>23</td>
      </tr>
      <tr>
          <td>14 days</td>
          <td>No</td>
          <td>37</td>
      </tr>
  </tbody>
</table>
<p>Every Yes entry is due in 3 days. But a No does not mean 14 days: 23 No entries are due in 3 days too, including CVE-2015-5477 (BIND) from the October 8, 2026 batch our BOD 26-04 article examined.</p>
<p>We also joined these 165 entries to their CVE records in <a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> and read the CISA-ADP SSVC &ldquo;Technical Impact&rdquo; option:</p>
<table>
  <thead>
      <tr>
          <th>Gap</th>
          <th>forensicTriage</th>
          <th>Technical Impact total</th>
          <th>Technical Impact partial</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>3 days</td>
          <td>Yes</td>
          <td>104</td>
          <td>1</td>
      </tr>
      <tr>
          <td>3 days</td>
          <td>No</td>
          <td>13</td>
          <td>10</td>
      </tr>
      <tr>
          <td>14 days</td>
          <td>No</td>
          <td>19</td>
          <td>18</td>
      </tr>
  </tbody>
</table>
<p>Under BOD 26-04 the agency timeline depends on KEV status, public exposure, automatability and technical impact for the specific asset, and the catalog publishes one date per CVE. Because 19 &ldquo;total&rdquo; entries got 14 days and 10 &ldquo;partial&rdquo; entries got 3, you cannot rebuild the catalog&rsquo;s <code>dueDate</code> from the SSVC values in the CVE record. Treat <code>dueDate</code> as CISA&rsquo;s published decision for that CVE and copy it as is. (The SSVC fields themselves are explained in <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">our Vulnrichment SSVC guide</a>.)</p>
<h3 id="weekends-count">Weekends count</h3>
<p>The gap is in calendar days. Of the 165 due dates since June 10, 2026, 26 fall on a Saturday and 23 on a Sunday, and 44 on a Friday. An internal SLA of &ldquo;3 business days from KEV listing&rdquo; will often land after the catalog date.</p>
<h2 id="cisa-edits-due-dates-after-publishing">CISA edits due dates after publishing</h2>
<p>The <a href="https://github.com/cisagov/kev-data">kev-data</a> repository keeps every catalog version since January 27, 2025 (345 commits to the JSON file when we checked). Replaying them and comparing each entry&rsquo;s <code>dateAdded</code>, <code>dueDate</code> and <code>forensicTriage</code> between consecutive versions found these changes, apart from the one-time September 1, 2026 backfill of <code>forensicTriage</code>:</p>
<table>
  <thead>
      <tr>
          <th>Catalog version</th>
          <th>CVE</th>
          <th>Change</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2025.02.18</td>
          <td>CVE-2025-0108 (PAN-OS)</td>
          <td>dueDate 2025-04-08 → 2025-03-11</td>
      </tr>
      <tr>
          <td>2025.04.29</td>
          <td>CVE-2025-3928 (Commvault)</td>
          <td>dueDate 2025-05-17 → 2025-05-19</td>
      </tr>
      <tr>
          <td>2025.10.06</td>
          <td>CVE-2025-61882</td>
          <td>dateAdded and dueDate each moved one day earlier</td>
      </tr>
      <tr>
          <td>2025.10.15</td>
          <td>CVE-2025-54253</td>
          <td>dateAdded and dueDate each moved one day earlier</td>
      </tr>
      <tr>
          <td>2025.12.09</td>
          <td>CVE-2025-55182 (React Server Components)</td>
          <td>dueDate 2025-12-26 → 2025-12-12</td>
      </tr>
      <tr>
          <td>2025.12.17</td>
          <td>15 entries added 2021-11-03</td>
          <td>dueDate corrected to 2022-05-03 (previously earlier than dateAdded)</td>
      </tr>
      <tr>
          <td>2026.05.20</td>
          <td>CVE-2010-0249 (Internet Explorer)</td>
          <td>dateAdded 2026-06-03 → 2026-05-20, dueDate unchanged</td>
      </tr>
      <tr>
          <td>2026.08.17</td>
          <td>CVE-2025-62593</td>
          <td>dateAdded and dueDate each moved one day earlier</td>
      </tr>
      <tr>
          <td>2026.10.08</td>
          <td>CVE-2026-86950 (Apple)</td>
          <td>forensicTriage Yes → No, dueDate 2026-10-02 → 2026-10-13</td>
      </tr>
      <tr>
          <td>2026.10.08</td>
          <td>CVE-2022-0995 (Linux kernel)</td>
          <td>forensicTriage No → Yes, dueDate 2026-09-09 → 2026-08-29</td>
      </tr>
      <tr>
          <td>2026.10.08</td>
          <td>CVE-2026-88779 (NetScaler), CVE-2023-22894 (Strapi)</td>
          <td>forensicTriage flipped, dueDate unchanged</td>
      </tr>
      <tr>
          <td>2026.10.13</td>
          <td>CVE-2026-12569 (PTC Windchill)</td>
          <td>forensicTriage No → Yes, dueDate unchanged</td>
      </tr>
  </tbody>
</table>
<p>That is 23 <code>dueDate</code> changes in total. Two cases matter most for automation:</p>
<ul>
<li><strong>Shortened deadlines.</strong> CVE-2025-0108 lost four weeks and CVE-2025-55182 lost two. A ticket created from the first catalog version would show the wrong date.</li>
<li><strong>Retroactive deadlines.</strong> CVE-2022-0995&rsquo;s new dueDate (2026-08-29) was published on October 8, 2026, so the entry went straight to overdue. If your pipeline only reads new entries, it never sees this.</li>
</ul>
<p>Our <a href="/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/">KEV time-to-listing analysis</a> covers the other end of the timeline: how long CVEs wait before <code>dateAdded</code>.</p>
<h2 id="run-it-yourself">Run it yourself</h2>
<p>Tested with Python 3 (standard library only) against <code>known_exploited_vulnerabilities.json</code> downloaded from <code>raw.githubusercontent.com/cisagov/kev-data/main/</code> on November 18, 2026.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> collections
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> date
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>kev <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">&#34;catalogVersion&#34;</span>, kev[<span style="color:#e6db74">&#34;catalogVersion&#34;</span>], <span style="color:#e6db74">&#34;count&#34;</span>, kev[<span style="color:#e6db74">&#34;count&#34;</span>])
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">days</span>(v):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (date<span style="color:#f92672">.</span>fromisoformat(v[<span style="color:#e6db74">&#34;dueDate&#34;</span>]) <span style="color:#f92672">-</span> date<span style="color:#f92672">.</span>fromisoformat(v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>]))<span style="color:#f92672">.</span>days
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>by_year <span style="color:#f92672">=</span> collections<span style="color:#f92672">.</span>defaultdict(collections<span style="color:#f92672">.</span>Counter)
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> kev[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    by_year[v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>][:<span style="color:#ae81ff">4</span>]][days(v)] <span style="color:#f92672">+=</span> <span style="color:#ae81ff">1</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> year <span style="color:#f92672">in</span> sorted(by_year):
</span></span><span style="display:flex;"><span>    c <span style="color:#f92672">=</span> by_year[year]
</span></span><span style="display:flex;"><span>    print(year, sum(c<span style="color:#f92672">.</span>values()), c<span style="color:#f92672">.</span>most_common(<span style="color:#ae81ff">4</span>))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Entries added from a given date: interval vs forensicTriage</span>
</span></span><span style="display:flex;"><span>since <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>] <span style="color:#66d9ef">if</span> len(sys<span style="color:#f92672">.</span>argv) <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">2</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;2026-06-10&#34;</span>
</span></span><span style="display:flex;"><span>pairs <span style="color:#f92672">=</span> collections<span style="color:#f92672">.</span>Counter(
</span></span><span style="display:flex;"><span>    (days(v), v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;forensicTriage&#34;</span>))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> kev[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>] <span style="color:#66d9ef">if</span> v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>] <span style="color:#f92672">&gt;=</span> since)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">&#34;since&#34;</span>, since, sorted(pairs<span style="color:#f92672">.</span>items()))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Open items: due today or later, and those due on a weekend</span>
</span></span><span style="display:flex;"><span>today <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>today()
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> kev[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    due <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>fromisoformat(v[<span style="color:#e6db74">&#34;dueDate&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> due <span style="color:#f92672">&gt;=</span> today:
</span></span><span style="display:flex;"><span>        flag <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34; (weekend)&#34;</span> <span style="color:#66d9ef">if</span> due<span style="color:#f92672">.</span>weekday() <span style="color:#f92672">&gt;=</span> <span style="color:#ae81ff">5</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>
</span></span><span style="display:flex;"><span>        print(v[<span style="color:#e6db74">&#34;cveID&#34;</span>], v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>], v[<span style="color:#e6db74">&#34;dueDate&#34;</span>], days(v), v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;forensicTriage&#34;</span>), flag)
</span></span></code></pre></div><p>Run it as <code>python3 kev_due.py known_exploited_vulnerabilities.json</code>. On November 18, 2026 it printed seven entries still open, all with 14-day gaps, the latest CVE-2026-115731 due 2026-12-01.</p>
<p>To find edited due dates, clone <code>cisagov/kev-data</code> (about 3 MB), list commits with <code>git log --reverse -- known_exploited_vulnerabilities.json</code>, load each version with <code>git show &lt;commit&gt;:known_exploited_vulnerabilities.json</code>, and compare each CVE&rsquo;s <code>(dateAdded, dueDate, forensicTriage)</code> with the previous version.</p>
<h2 id="what-to-do-with-this">What to do with this</h2>
<ul>
<li><strong>Copy <code>dueDate</code> into the ticket and update it on every sync.</strong> Key the comparison on <code>cveID</code> across the whole catalog, not only on entries added since your last run.</li>
<li><strong>Alert on changes, not just additions.</strong> A shortened or past <code>dueDate</code> on an existing entry deserves the same page as a new entry.</li>
<li><strong>Do not derive the deadline</strong> from <code>forensicTriage</code>, SSVC Technical Impact or CVSS. In the current data none of them reproduces the catalog date.</li>
<li><strong>Count calendar days.</strong> Map the catalog date to your SLA directly instead of adding business days.</li>
<li><strong>If you are not a federal agency,</strong> the catalog date is a strong urgency signal, not your legal deadline. Our <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV comparison</a> covers how to weigh it.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>Counts come from catalog version 2026.11.17 and the kev-data history from January 27, 2025 onward. Edits made before that date are not visible in the repository.</li>
<li><code>dueDate</code> in the catalog is one date per CVE. Federal agencies&rsquo; asset-level timelines under BOD 26-04 depend on exposure and other decision points the catalog does not carry; we did not re-verify the directive&rsquo;s Table 1 for this article.</li>
<li>We describe what the data shows about the 21-day period and the 2026 change; we found no CISA statement explaining either.</li>
<li>SSVC values were read from cvelistV5 records on November 18, 2026; CISA-ADP can update them after a CVE is added to KEV.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>dueDate and forensicTriage field descriptions</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities_schema.json">kev-data schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Gap counts by year and since June 10, 2026</td>
          <td><a href="https://github.com/cisagov/kev-data">KEV catalog JSON, version 2026.11.17</a> (script above)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>forensicTriage first published 2026-09-01 (38 Yes, 1,649 No)</td>
          <td>kev-data git history</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 22-01 six-month / two-week defaults and grave-risk adjustment</td>
          <td><a href="https://www.cisa.gov/binding-operational-directive-22-01">CISA BOD 22-01</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC Technical Impact per entry</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> CISA-ADP containers</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>23 dueDate edits since January 2025</td>
          <td>kev-data git history replay</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 publication date June 10, 2026</td>
          <td><a href="https://www.cisa.gov/news-events/directives/bod-26-04-prioritizing-security-updates-based-risk">CISA BOD 26-04</a>, via our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">earlier article</a></td>
          <td>Qualified (page not re-fetched today)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>KEV in CVE Records: Audit of All 1,781 CISA-ADP kev Blocks</title><link>https://firewallsync.com/posts/kev-in-cve-records-cisa-adp-kev-block-full-catalog-audit/</link><guid>https://firewallsync.com/posts/kev-in-cve-records-cisa-adp-kev-block-full-catalog-audit/</guid><pubDate>Tue, 17 Nov 2026 04:55:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>We compared every entry in CISA's KEV catalog (version 2026.11.13, 1,781 CVEs) with the CISA-ADP kev block in each CVE record, and scanned all 426,227 records in the November 17, 2026 cvelistV5 snapshot for kev blocks that point at CVEs no longer in the catalog. The record copy matches the catalog for 1,777 CVEs; this article lists the four that don't, what happens to a record when CISA removes a CVE from KEV, and a script to run the check yourself.</description><enclosure url="https://firewallsync.com/posts/kev-in-cve-records-cisa-adp-kev-block-full-catalog-audit/cover_hu446428385090800819.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/kev-in-cve-records-cisa-adp-kev-block-full-catalog-audit/cover_hu446428385090800819.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/kev-in-cve-records-cisa-adp-kev-block-full-catalog-audit/cover_hu446428385090800819.jpg" alt="A wall of old wooden card-catalog drawers with brass label holders and handwritten labels, receding into the dark" width="1200" height="630"></p><p>If your vulnerability pipeline reads CVE JSON 5 records instead of the KEV catalog, you are trusting a copy. CISA&rsquo;s Vulnrichment program writes a <code>kev</code> block into the CISA-ADP container of each CVE record that is on the <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">Known Exploited Vulnerabilities catalog</a>. We checked that copy against the whole catalog on November 17, 2026.</p>
<p><strong>The short answer:</strong> the copy is close to complete. 1,777 of 1,781 KEV CVEs have a kev block whose <code>dateAdded</code> matches the catalog. Two have no block at all, two show a different date, and no record anywhere in the CVE corpus carries a kev block for a CVE that is not on the catalog, including the three CVEs CISA has removed since January 2025.</p>
<p>This piece goes beyond our <a href="/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/">30-day exploitation-alert diff</a>, which measured how quickly new KEV entries reach records. Here the question is whether the record copy can be trusted at full-catalog scale, in both directions.</p>
<h2 id="where-the-kev-block-lives">Where the kev block lives</h2>
<p>The kev block sits in the CISA-ADP container&rsquo;s <code>metrics</code> array, next to the SSVC block that the <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC guide</a> explains. CISA&rsquo;s <a href="https://github.com/cisagov/vulnrichment">Vulnrichment README</a> states the rule: &ldquo;For those CVEs that are on the KEV, the CISA ADP will add a KEV block &hellip; For those that aren&rsquo;t, no update will occur.&rdquo;</p>
<p>Here is the CISA-ADP <code>metrics</code> content of CVE-2026-82078 (PaperCut NG/MF) as it appeared in cvelistV5 on November 17, 2026, trimmed to the two <code>other</code> entries:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>[
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;other&#34;</span>: {<span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;ssvc&#34;</span>, <span style="color:#f92672">&#34;content&#34;</span>: {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-08-28T00:00:00+00:00&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;options&#34;</span>: [{<span style="color:#f92672">&#34;Exploitation&#34;</span>: <span style="color:#e6db74">&#34;active&#34;</span>}, {<span style="color:#f92672">&#34;Automatable&#34;</span>: <span style="color:#e6db74">&#34;no&#34;</span>}, {<span style="color:#f92672">&#34;Technical Impact&#34;</span>: <span style="color:#e6db74">&#34;total&#34;</span>}],
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;role&#34;</span>: <span style="color:#e6db74">&#34;CISA Coordinator&#34;</span>, <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;2.0.3&#34;</span>, <span style="color:#f92672">&#34;id&#34;</span>: <span style="color:#e6db74">&#34;CVE-2026-82078&#34;</span>}}},
</span></span><span style="display:flex;"><span>  {<span style="color:#f92672">&#34;other&#34;</span>: {<span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;kev&#34;</span>, <span style="color:#f92672">&#34;content&#34;</span>: {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;dateAdded&#34;</span>: <span style="color:#e6db74">&#34;2026-08-28&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;reference&#34;</span>: <span style="color:#e6db74">&#34;https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-82078&#34;</span>}}}
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>The block has two fields: <code>dateAdded</code> and a <code>reference</code> URL. Everything else KEV publishes (<code>dueDate</code>, <code>requiredAction</code>, <code>knownRansomwareCampaignUse</code>, <code>notes</code>, <code>cwes</code>) is only in the catalog. Our <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed field guide</a> covers those fields.</p>
<h2 id="what-we-compared">What we compared</h2>
<table>
  <thead>
      <tr>
          <th>Input</th>
          <th>Version used</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV catalog</td>
          <td><code>known_exploited_vulnerabilities.json</code> from <a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a>, catalogVersion 2026.11.13 (dateReleased 2026-11-13T20:24:18Z), 1,781 entries, dateAdded range 2021-11-03 to 2026-11-13</td>
      </tr>
      <tr>
          <td>CVE records for KEV CVEs</td>
          <td>Each record fetched from the <code>main</code> branch of <a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> on 2026-11-17 (branch head <code>a683955</code>)</td>
      </tr>
      <tr>
          <td>Full CVE corpus</td>
          <td><code>2026-11-17_all_CVEs_at_midnight.zip.zip</code> from the cvelistV5 release <code>cve_2026-11-17_0000Z</code>: 426,227 records</td>
      </tr>
      <tr>
          <td>Removal history</td>
          <td>Git history of <code>known_exploited_vulnerabilities.json</code> in cisagov/kev-data, 344 commits from 2025-01-27</td>
      </tr>
  </tbody>
</table>
<p>Two checks ran. <strong>Forward:</strong> for every catalog CVE, does the record have a CISA-ADP kev block, and does its <code>dateAdded</code> equal the catalog&rsquo;s? <strong>Reverse:</strong> does any record in the full corpus carry a CISA-ADP kev block for a CVE that is not in the catalog?</p>
<h2 id="results">Results</h2>
<table>
  <thead>
      <tr>
          <th>Check</th>
          <th>Count</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV catalog entries</td>
          <td>1,781</td>
      </tr>
      <tr>
          <td>Records with a CISA-ADP kev block and matching dateAdded</td>
          <td>1,777</td>
      </tr>
      <tr>
          <td>Records with a kev block but a different dateAdded</td>
          <td>2</td>
      </tr>
      <tr>
          <td>Records with a CISA-ADP container but no kev block</td>
          <td>2</td>
      </tr>
      <tr>
          <td>KEV records with no CISA-ADP container</td>
          <td>0</td>
      </tr>
      <tr>
          <td>KEV records not in PUBLISHED state</td>
          <td>0</td>
      </tr>
      <tr>
          <td>Records in the full corpus with a kev block for a non-KEV CVE</td>
          <td>0</td>
      </tr>
  </tbody>
</table>
<p>The 1,781 KEV CVEs span every catalog year: 311 added in 2021, 555 in 2022, 187 in 2023, 186 in 2024, 245 in 2025 and 297 so far in 2026. The kev blocks therefore cover the backfilled early catalog as well as recent additions.</p>
<p>Every one of the 1,781 records also carried a CISA-ADP SSVC block with Exploitation set to <code>active</code>, including the two without a kev block.</p>
<h3 id="two-kev-cves-with-no-kev-block">Two KEV CVEs with no kev block</h3>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Catalog entry</th>
          <th>KEV dateAdded</th>
          <th>CISA-ADP container dateUpdated</th>
          <th>SSVC Exploitation</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-48356</td>
          <td>Adobe Commerce and Magento</td>
          <td>2026-10-27</td>
          <td>2026-10-28T03:55:44Z</td>
          <td>active</td>
      </tr>
      <tr>
          <td>CVE-2026-111497</td>
          <td>Google Brotli</td>
          <td>2026-10-27</td>
          <td>2026-10-28T03:55:48Z</td>
          <td>active</td>
      </tr>
  </tbody>
</table>
<p>Both were already missing a kev block in the November 16 snapshot used in our <a href="/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/">exploitation-alert diff</a>. Their CISA-ADP containers were updated the day after the KEV addition (the SSVC change arrived) but the kev block did not. We do not know why; CISA has not published an explanation that we could find. A pipeline that treats &ldquo;kev block present&rdquo; as &ldquo;on KEV&rdquo; would have missed both for three weeks.</p>
<h3 id="two-kev-blocks-with-a-different-dateadded">Two kev blocks with a different dateAdded</h3>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Catalog entry</th>
          <th>Catalog dateAdded</th>
          <th>Record kev dateAdded</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-81578</td>
          <td>PaperCut NG/MF</td>
          <td>2026-08-31</td>
          <td>2026-08-28</td>
      </tr>
      <tr>
          <td>CVE-2026-82078</td>
          <td>PaperCut NG/MF</td>
          <td>2026-08-31</td>
          <td>2026-08-28</td>
      </tr>
  </tbody>
</table>
<p>The kev-data git history shows both CVEs first appearing in catalog version 2026.08.31, committed 2026-08-31, with dateAdded 2026-08-31, and the date never changing afterwards. The record says three days earlier, matching the timestamp on the record&rsquo;s SSVC block. If you compute time-to-KEV or BOD due dates from the record, these two come out three days early. Use the catalog date; our <a href="/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/">KEV time-to-listing analysis</a> does.</p>
<h2 id="what-happens-to-the-record-when-a-cve-leaves-kev">What happens to the record when a CVE leaves KEV</h2>
<p>CISA occasionally removes entries. The kev-data history since January 27, 2025 shows three CVEs that were in the catalog and are not in it now, plus one that briefly disappeared and came back:</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Catalog entry</th>
          <th>In catalog</th>
          <th>Removed from catalog</th>
          <th>Record today</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2025-4664</td>
          <td>Google Chromium</td>
          <td>from version 2025.05.15</td>
          <td>version 2025.06.09</td>
          <td>no kev block; SSVC Exploitation <code>none</code></td>
      </tr>
      <tr>
          <td>CVE-2025-6264</td>
          <td>Rapid7 Velociraptor</td>
          <td>from version 2025.10.14</td>
          <td>version 2025.10.22</td>
          <td>no kev block; SSVC Exploitation <code>none</code></td>
      </tr>
      <tr>
          <td>CVE-2026-69836</td>
          <td>Microsoft Entra ID</td>
          <td>added 2026-08-21 17:28 UTC commit</td>
          <td>2026-08-21 18:42 UTC commit</td>
          <td>no kev block; SSVC Exploitation <code>none</code></td>
      </tr>
      <tr>
          <td>CVE-2026-1281</td>
          <td>Ivanti EPMM</td>
          <td>added 2026-01-29 18:15 UTC commit</td>
          <td>dropped 19:14 UTC, back 20:41 UTC the same day</td>
          <td>kev block, dateAdded 2026-01-29</td>
      </tr>
  </tbody>
</table>
<p>These are what the data shows; the commits do not record why an entry was removed, and we did not find a CISA statement for each. The record behaviour is consistent, though: when an entry left the catalog, the record lost its kev block and its SSVC Exploitation value was set to <code>none</code>. A pipeline that only ever adds &ldquo;exploited&rdquo; flags and never clears them will keep all three marked as exploited.</p>
<p>The reverse scan backs this up at corpus scale: none of the 426,227 records in the November 17 snapshot carries a CISA-ADP kev block for a CVE outside the catalog.</p>
<h2 id="run-the-audit-yourself">Run the audit yourself</h2>
<p>Tested with Python 3 on November 17, 2026. Fetch the catalog and the records, then run the script.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>curl -sSO https://raw.githubusercontent.com/cisagov/kev-data/main/known_exploited_vulnerabilities.json
</span></span><span style="display:flex;"><span>mkdir -p records
</span></span><span style="display:flex;"><span>python3 - <span style="color:#e6db74">&lt;&lt;&#39;EOF&#39; &gt; curl.cfg
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">import json
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">for v in json.load(open(&#34;known_exploited_vulnerabilities.json&#34;))[&#34;vulnerabilities&#34;]:
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    cve = v[&#34;cveID&#34;]; year, num = cve.split(&#34;-&#34;)[1:]
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    path = f&#34;cves/{year}/{int(num)//1000}xxx/{cve}.json&#34;
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    print(f&#39;url=&#34;https://raw.githubusercontent.com/CVEProject/cvelistV5/main/{path}&#34;&#39;)
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    print(f&#39;output=&#34;records/{cve}.json&#34;&#39;)
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">EOF</span>
</span></span><span style="display:flex;"><span>curl -sS --parallel --parallel-max <span style="color:#ae81ff">16</span> -K curl.cfg
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e"># audit_kev_blocks.py: compare the KEV catalog with the CISA-ADP &#34;kev&#34; block in each CVE record.</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Usage: python3 audit_kev_blocks.py known_exploited_vulnerabilities.json records/</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> pathlib
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>catalog <span style="color:#f92672">=</span> {v[<span style="color:#e6db74">&#34;cveID&#34;</span>]: v <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>records <span style="color:#f92672">=</span> pathlib<span style="color:#f92672">.</span>Path(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cisa_adp_metrics</span>(record):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> record<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;containers&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;providerMetadata&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;shortName&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> {m[<span style="color:#e6db74">&#34;other&#34;</span>][<span style="color:#e6db74">&#34;type&#34;</span>]: m[<span style="color:#e6db74">&#34;other&#34;</span>][<span style="color:#e6db74">&#34;content&#34;</span>]
</span></span><span style="display:flex;"><span>                    <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []) <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;other&#34;</span> <span style="color:#f92672">in</span> m}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>missing, date_mismatch, ok <span style="color:#f92672">=</span> [], [], <span style="color:#ae81ff">0</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> cve, entry <span style="color:#f92672">in</span> sorted(catalog<span style="color:#f92672">.</span>items()):
</span></span><span style="display:flex;"><span>    record <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>loads((records <span style="color:#f92672">/</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74">.json&#34;</span>)<span style="color:#f92672">.</span>read_text())
</span></span><span style="display:flex;"><span>    metrics <span style="color:#f92672">=</span> cisa_adp_metrics(record) <span style="color:#f92672">or</span> {}
</span></span><span style="display:flex;"><span>    kev <span style="color:#f92672">=</span> metrics<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;kev&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> kev <span style="color:#f92672">is</span> <span style="color:#66d9ef">None</span>:
</span></span><span style="display:flex;"><span>        missing<span style="color:#f92672">.</span>append((cve, entry[<span style="color:#e6db74">&#34;dateAdded&#34;</span>]))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">elif</span> kev[<span style="color:#e6db74">&#34;dateAdded&#34;</span>] <span style="color:#f92672">!=</span> entry[<span style="color:#e6db74">&#34;dateAdded&#34;</span>]:
</span></span><span style="display:flex;"><span>        date_mismatch<span style="color:#f92672">.</span>append((cve, entry[<span style="color:#e6db74">&#34;dateAdded&#34;</span>], kev[<span style="color:#e6db74">&#34;dateAdded&#34;</span>]))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>        ok <span style="color:#f92672">+=</span> <span style="color:#ae81ff">1</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;catalog entries: </span><span style="color:#e6db74">{</span>len(catalog)<span style="color:#e6db74">}</span><span style="color:#e6db74">  matching kev block: </span><span style="color:#e6db74">{</span>ok<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">&#34;no kev block:&#34;</span>, missing)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">&#34;dateAdded differs (catalog, record):&#34;</span>, date_mismatch)
</span></span></code></pre></div><p>Output on November 17, 2026:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>catalog entries: 1781  matching kev block: 1777
</span></span><span style="display:flex;"><span>no kev block: [(&#39;CVE-2026-111497&#39;, &#39;2026-10-27&#39;), (&#39;CVE-2026-48356&#39;, &#39;2026-10-27&#39;)]
</span></span><span style="display:flex;"><span>dateAdded differs (catalog, record): [(&#39;CVE-2026-81578&#39;, &#39;2026-08-31&#39;, &#39;2026-08-28&#39;), (&#39;CVE-2026-82078&#39;, &#39;2026-08-31&#39;, &#39;2026-08-28&#39;)]
</span></span></code></pre></div><p>The script assumes every catalog CVE has a published record (true on this date). For the reverse check, stream the midnight release zip with Python&rsquo;s <code>zipfile</code>, skip any record whose bytes do not contain <code>&quot;kev&quot;</code>, and collect CISA-ADP kev blocks whose CVE is not in the catalog; a full pass took about 20 seconds here. To keep the check current without refetching everything, drive it from cvelistV5 <code>delta.json</code> as described in our <a href="/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/">incremental CVE sync guide</a>.</p>
<h2 id="what-to-do-with-this">What to do with this</h2>
<ol>
<li><strong>Keep the catalog as the source of truth.</strong> It is published first, carries the fields the record lacks, and its <code>dateAdded</code> is the one BOD due dates are computed from.</li>
<li><strong>Use the kev block as a cross-check, not a trigger.</strong> Alert when the two disagree: a catalog CVE with no kev block, a kev block with a different date, or a kev block for a non-catalog CVE.</li>
<li><strong>Handle removals.</strong> If an exploited flag in your system came from KEV or from SSVC <code>active</code>, clear or re-review it when the CVE leaves the catalog or Exploitation goes back to <code>none</code>.</li>
<li><strong>Fall back to SSVC when the kev block is missing.</strong> Both missing-block CVEs had Exploitation <code>active</code>, so a rule of &ldquo;kev block OR SSVC active OR in catalog&rdquo; would not have missed them.</li>
</ol>
<p>These steps slot into the KEV and exploitation stages of our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>One snapshot: the catalog as of version 2026.11.13 and records as of November 17, 2026. The two missing blocks may be added later.</li>
<li>The removal history covers only what the cisagov/kev-data repository records, which starts on 2025-01-27. Earlier removals, if any, are not counted.</li>
<li>We report what the data shows about removals, not why CISA removed each entry.</li>
<li>The forward check fetched records from the moving <code>main</code> branch over about 11 seconds; the reverse check used the fixed midnight release.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>1,781 catalog entries, version 2026.11.13</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a> <code>known_exploited_vulnerabilities.json</code></td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>1,777 matching, 2 missing, 2 date mismatches</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> records, script above</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>0 kev blocks for non-catalog CVEs across 426,227 records</td>
          <td>cvelistV5 release <code>cve_2026-11-17_0000Z</code></td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>CISA ADP adds a kev block only for KEV CVEs</td>
          <td><a href="https://github.com/cisagov/vulnrichment">cisagov/vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Three removals and one remove-and-re-add since 2025-01-27</td>
          <td>cisagov/kev-data git history</td>
          <td>Verified (computed); reasons not stated</td>
      </tr>
      <tr>
          <td>Removed CVEs have no kev block and SSVC Exploitation none</td>
          <td>cvelistV5 records, November 17, 2026</td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE Exploitation Alerts: Diff CISA-ADP KEV and SSVC Data</title><link>https://firewallsync.com/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/</link><guid>https://firewallsync.com/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/</guid><pubDate>Mon, 16 Nov 2026 04:56:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>A CVE record's CISA-ADP container gains a KEV block and an SSVC Exploitation value of active when exploitation is confirmed. We diffed 31,312 changed records between the October 17 and November 16, 2026 cvelistV5 snapshots to see how those signals move, then built a small watcher that alerts on them.</description><enclosure url="https://firewallsync.com/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/cover_hu7826863987297709055.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/cover_hu7826863987297709055.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-exploitation-alerts-cisa-adp-kev-ssvc-cvelistv5-diff/cover_hu7826863987297709055.jpg" alt="A person standing on a frozen, rocky shore holds a lit flare above their head, trailing a cloud of red smoke against a white sky" width="1200" height="630"></p><p>If you want to know the moment a CVE in your inventory becomes exploited, there are two machine-readable places to look: CISA&rsquo;s KEV JSON feed and the CISA-ADP container inside the CVE record itself. The record carries an SSVC <strong>Exploitation</strong> value (<code>none</code>, <code>poc</code> or <code>active</code>) and, for KEV CVEs, a <code>kev</code> block. Diffing those two fields between record versions gives you an exploitation alert that rides on the same cvelistV5 sync you may already run.</p>
<p>We measured how those fields actually moved over 30 days, from the cvelistV5 snapshot tagged <code>cve_2026-10-17_0000Z</code> to <code>cve_2026-11-16_0000Z</code>. The short answer: in this window, <code>active</code> never appeared on a CVE that was not already in KEV, and the <code>kev</code> block followed the KEV catalog by about a day, with two misses. So the KEV feed stays the primary trigger; the CVE-record diff is the second signal, and it is the one that catches <code>poc</code> changes and rejections.</p>
<p>This builds on the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>, where exploitation is the first gate on the way from CVE record to ticket.</p>
<h2 id="where-the-exploitation-signals-live-in-a-cve-record">Where the exploitation signals live in a CVE record</h2>
<p>CISA publishes its enrichment through an Authorized Data Publisher container (see <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">CVE ADP containers</a> for how ADP containers relate to the CNA&rsquo;s). In a CVE JSON 5 record, look for <code>containers.adp[]</code> with <code>providerMetadata.shortName</code> equal to <code>CISA-ADP</code>. Its <code>metrics[]</code> array holds <code>other</code> objects, two of which matter here. From the current record for GitLab CVE-2026-97229:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;ssvc&#34;</span>, <span style="color:#f92672">&#34;content&#34;</span>: {<span style="color:#f92672">&#34;id&#34;</span>: <span style="color:#e6db74">&#34;CVE-2026-97229&#34;</span>, <span style="color:#f92672">&#34;role&#34;</span>: <span style="color:#e6db74">&#34;CISA Coordinator&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;options&#34;</span>: [{<span style="color:#f92672">&#34;Exploitation&#34;</span>: <span style="color:#e6db74">&#34;active&#34;</span>}, {<span style="color:#f92672">&#34;Automatable&#34;</span>: <span style="color:#e6db74">&#34;...&#34;</span>}, {<span style="color:#f92672">&#34;Technical Impact&#34;</span>: <span style="color:#e6db74">&#34;...&#34;</span>}],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;2.0.3&#34;</span>, <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;...&#34;</span>}}
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;kev&#34;</span>, <span style="color:#f92672">&#34;content&#34;</span>: {<span style="color:#f92672">&#34;dateAdded&#34;</span>: <span style="color:#e6db74">&#34;2026-10-29&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;reference&#34;</span>: <span style="color:#e6db74">&#34;https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-97229&#34;</span>}}
</span></span></code></pre></div><p>(Automatable, Technical Impact and the timestamp are elided; the <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment explainer</a> covers all three SSVC values.)</p>
<p>The <a href="https://github.com/cisagov/vulnrichment/blob/develop/README.md">CISA Vulnrichment README</a> states the rule for the second block: &ldquo;For those CVEs that are on the KEV, the CISA ADP will add a KEV block. For those that aren&rsquo;t, no update will occur.&rdquo; The SSVC values come from the CISA Coordinator decision tree; the SSVC project&rsquo;s <a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/exploitation.py">Exploitation decision point</a> defines <code>active</code> as &ldquo;Shared, observable, reliable evidence that the exploit is being used in the wild by real attackers; there is credible public reporting&rdquo;, and <code>none</code> as no evidence of active exploitation and no public proof of concept.</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Values</th>
          <th>What a change means</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>ssvc</code> Exploitation</td>
          <td><code>none</code>, <code>poc</code>, <code>active</code></td>
          <td>CISA&rsquo;s current assessment of exploitation state</td>
      </tr>
      <tr>
          <td><code>kev</code> dateAdded</td>
          <td>a date</td>
          <td>The CVE is on the KEV catalog; date matches the catalog</td>
      </tr>
      <tr>
          <td><code>cveMetadata.state</code></td>
          <td><code>PUBLISHED</code>, <code>REJECTED</code></td>
          <td>A rejected record keeps no ADP containers</td>
      </tr>
  </tbody>
</table>
<h2 id="how-we-measured">How we measured</h2>
<p>We cloned <a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> without blobs and ran <code>git diff --name-status</code> between the two tags. The <code>cve_2026-10-17_0000Z</code> tag points at a commit made 2026-10-16 23:29 UTC; <code>cve_2026-11-16_0000Z</code> at 2026-11-16 00:15 UTC. That produced 31,312 changed CVE files: 16,690 added and 14,622 modified. We fetched both versions of each modified record and the new version of each added record from raw.githubusercontent.com at the two commits, then compared the CISA-ADP values. KEV comparisons used the <a href="https://github.com/cisagov/kev-data">kev-data</a> JSON as of November 16, 2026.</p>
<p>A two-point diff only sees the start and end state, so a record that went <code>none</code> to <code>poc</code> to <code>active</code> inside the window counts once, as <code>none</code> to <code>active</code>.</p>
<h2 id="what-moved-in-30-days">What moved in 30 days</h2>
<table>
  <thead>
      <tr>
          <th>Change in CISA-ADP Exploitation (modified records)</th>
          <th>Records</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>no SSVC block → <code>none</code></td>
          <td>2,389</td>
      </tr>
      <tr>
          <td>no SSVC block → <code>poc</code></td>
          <td>282</td>
      </tr>
      <tr>
          <td><code>none</code> → <code>active</code></td>
          <td>14</td>
      </tr>
      <tr>
          <td><code>none</code> → <code>poc</code></td>
          <td>10</td>
      </tr>
      <tr>
          <td><code>poc</code> → <code>none</code></td>
          <td>5</td>
      </tr>
      <tr>
          <td>no SSVC block → <code>active</code></td>
          <td>2</td>
      </tr>
      <tr>
          <td><code>none</code> or <code>poc</code> → no SSVC block</td>
          <td>19</td>
      </tr>
  </tbody>
</table>
<p>All 19 losses were records that changed from <code>PUBLISHED</code> to <code>REJECTED</code>; none of the 672 rejected records in the changed set carries an ADP container. (29 modified records were rejected in the window in total; the other 10 had no CISA-ADP SSVC to lose.) The <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">CVE record states</a> post covers what a rejection means for triage.</p>
<p>Of the 16,690 records first published in the window, 12,071 arrived with Exploitation <code>none</code>, 1,940 with <code>poc</code>, 18 with <code>active</code> and 2,661 with no CISA-ADP SSVC block yet.</p>
<p>Two findings matter for alerting:</p>
<ol>
<li><strong><code>active</code> tracked KEV exactly.</strong> All 34 records that gained <code>active</code> (16) or were published with it (18) were on the KEV catalog. In this window, alerting on <code>active</code> would not have told you about any CVE that the KEV feed did not.</li>
<li><strong>Exploitation can go down.</strong> Five records went from <code>poc</code> to <code>none</code>. Store the current value and alert on increases; do not assume the field only ratchets upward.</li>
</ol>
<h2 id="how-closely-the-kev-block-followed-the-catalog">How closely the kev block followed the catalog</h2>
<p>CISA added 33 CVEs to KEV with a <code>dateAdded</code> between October 17 and November 15, 2026. By the November 16 snapshot:</p>
<ul>
<li>31 had a CISA-ADP <code>kev</code> block whose <code>dateAdded</code> matched the catalog.</li>
<li>All 33 had SSVC Exploitation <code>active</code>.</li>
<li>Two had <code>active</code> but <strong>no</strong> <code>kev</code> block: Adobe CVE-2026-48356 and Google CVE-2026-111497, both added to KEV on October 27. Their CISA-ADP containers were last updated on October 28 with only the SSVC entry.</li>
</ul>
<p>For the 31 with a block, the CISA-ADP container&rsquo;s <code>dateUpdated</code> fell one calendar day after the KEV <code>dateAdded</code> for 28 CVEs, two days later for 2 (the PHP pair CVE-2026-109618 and CVE-2026-109619) and four days later for 1 (Ivanti CVE-2026-118527). Most updates landed between 03:55 and 04:57 UTC. Because <code>dateUpdated</code> is the container&rsquo;s <em>last</em> update, it is an upper bound on when the block appeared, not an exact timestamp. The KEV catalog&rsquo;s own <code>dateAdded</code> is a date with no time.</p>
<p>The practical reading: the record-level <code>kev</code> block is a copy that typically lags the catalog by about a day and occasionally does not arrive. If you need same-day KEV alerts, read the catalog. The <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed field guide</a> covers matching it against a scanner export.</p>
<h2 id="a-watcher-that-diffs-cisa-adp-signals">A watcher that diffs CISA-ADP signals</h2>
<p>The script below reads cvelistV5&rsquo;s <code>delta.json</code> (the change list described in the <a href="/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/">delta.json and deltaLog.json guide</a>), fetches each new or updated record, extracts the CISA-ADP signals and compares them with a local state file. It alerts on a new <code>kev</code> block, an increase in Exploitation to <code>poc</code> or <code>active</code>, and a change to <code>REJECTED</code>. Python 3 standard library only.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Alert when a CVE record&#39;s CISA-ADP container gains a KEV block or a new SSVC Exploitation value.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage:
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  python3 adp_watch.py state.json                 # read cvelistV5 delta.json, fetch changed records
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  python3 adp_watch.py state.json rec1.json ...   # offline: compare local record files
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>DELTA <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/delta.json&#34;</span>
</span></span><span style="display:flex;"><span>RANK <span style="color:#f92672">=</span> {<span style="color:#66d9ef">None</span>: <span style="color:#ae81ff">0</span>, <span style="color:#e6db74">&#34;none&#34;</span>: <span style="color:#ae81ff">1</span>, <span style="color:#e6db74">&#34;poc&#34;</span>: <span style="color:#ae81ff">2</span>, <span style="color:#e6db74">&#34;active&#34;</span>: <span style="color:#ae81ff">3</span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">signals</span>(rec):
</span></span><span style="display:flex;"><span>    s <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;state&#34;</span>: rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;state&#34;</span>], <span style="color:#e6db74">&#34;exploitation&#34;</span>: <span style="color:#66d9ef">None</span>, <span style="color:#e6db74">&#34;kev&#34;</span>: <span style="color:#66d9ef">None</span>}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> rec<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;containers&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []) <span style="color:#f92672">or</span> []:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;providerMetadata&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;shortName&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []) <span style="color:#f92672">or</span> []:
</span></span><span style="display:flex;"><span>            other <span style="color:#f92672">=</span> m<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;other&#34;</span>) <span style="color:#f92672">or</span> {}
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> other<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;type&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;ssvc&#34;</span>:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">for</span> opt <span style="color:#f92672">in</span> other[<span style="color:#e6db74">&#34;content&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;options&#34;</span>, []):
</span></span><span style="display:flex;"><span>                    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;Exploitation&#34;</span> <span style="color:#f92672">in</span> opt:
</span></span><span style="display:flex;"><span>                        s[<span style="color:#e6db74">&#34;exploitation&#34;</span>] <span style="color:#f92672">=</span> opt[<span style="color:#e6db74">&#34;Exploitation&#34;</span>]
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">elif</span> other<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;type&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;kev&#34;</span>:
</span></span><span style="display:flex;"><span>                s[<span style="color:#e6db74">&#34;kev&#34;</span>] <span style="color:#f92672">=</span> other[<span style="color:#e6db74">&#34;content&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;dateAdded&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> s
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">compare</span>(cve, old, new):
</span></span><span style="display:flex;"><span>    alerts <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> new[<span style="color:#e6db74">&#34;state&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;REJECTED&#34;</span> <span style="color:#f92672">and</span> (old <span style="color:#f92672">or</span> {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;state&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;REJECTED&#34;</span>:
</span></span><span style="display:flex;"><span>        alerts<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74">: record REJECTED, ADP data removed&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> new[<span style="color:#e6db74">&#34;kev&#34;</span>] <span style="color:#f92672">and</span> <span style="color:#f92672">not</span> (old <span style="color:#f92672">or</span> {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;kev&#34;</span>):
</span></span><span style="display:flex;"><span>        alerts<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74">: KEV block added (dateAdded </span><span style="color:#e6db74">{</span>new[<span style="color:#e6db74">&#39;kev&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span><span style="display:flex;"><span>    o <span style="color:#f92672">=</span> (old <span style="color:#f92672">or</span> {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;exploitation&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> RANK<span style="color:#f92672">.</span>get(new[<span style="color:#e6db74">&#34;exploitation&#34;</span>], <span style="color:#ae81ff">0</span>) <span style="color:#f92672">&gt;</span> RANK<span style="color:#f92672">.</span>get(o, <span style="color:#ae81ff">0</span>) <span style="color:#f92672">and</span> new[<span style="color:#e6db74">&#34;exploitation&#34;</span>] <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;poc&#34;</span>, <span style="color:#e6db74">&#34;active&#34;</span>):
</span></span><span style="display:flex;"><span>        alerts<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74">: SSVC Exploitation </span><span style="color:#e6db74">{</span>o<span style="color:#e6db74">}</span><span style="color:#e6db74"> -&gt; </span><span style="color:#e6db74">{</span>new[<span style="color:#e6db74">&#39;exploitation&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> alerts
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">get</span>(url):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(url, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">30</span>) <span style="color:#66d9ef">as</span> r:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> json<span style="color:#f92672">.</span>load(r)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>():
</span></span><span style="display:flex;"><span>    state_path, files <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>:]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>        state <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(state_path))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">except</span> <span style="color:#a6e22e">FileNotFoundError</span>:
</span></span><span style="display:flex;"><span>        state <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> files:
</span></span><span style="display:flex;"><span>        records <span style="color:#f92672">=</span> [json<span style="color:#f92672">.</span>load(open(f)) <span style="color:#66d9ef">for</span> f <span style="color:#f92672">in</span> files]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>        delta <span style="color:#f92672">=</span> get(DELTA)
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;delta fetchTime </span><span style="color:#e6db74">{</span>delta[<span style="color:#e6db74">&#39;fetchTime&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>delta[<span style="color:#e6db74">&#39;numberOfChanges&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> changes&#34;</span>)
</span></span><span style="display:flex;"><span>        records <span style="color:#f92672">=</span> [get(e[<span style="color:#e6db74">&#34;githubLink&#34;</span>]) <span style="color:#66d9ef">for</span> k <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;new&#34;</span>, <span style="color:#e6db74">&#34;updated&#34;</span>) <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> delta<span style="color:#f92672">.</span>get(k, [])]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> rec <span style="color:#f92672">in</span> records:
</span></span><span style="display:flex;"><span>        cve <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>]
</span></span><span style="display:flex;"><span>        new <span style="color:#f92672">=</span> signals(rec)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> a <span style="color:#f92672">in</span> compare(cve, state<span style="color:#f92672">.</span>get(cve), new):
</span></span><span style="display:flex;"><span>            print(<span style="color:#e6db74">&#34;ALERT&#34;</span>, a)
</span></span><span style="display:flex;"><span>        state[cve] <span style="color:#f92672">=</span> new
</span></span><span style="display:flex;"><span>    json<span style="color:#f92672">.</span>dump(state, open(state_path, <span style="color:#e6db74">&#34;w&#34;</span>), indent<span style="color:#f92672">=</span><span style="color:#ae81ff">1</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main()
</span></span></code></pre></div><p><strong>How we ran it (Python 3.13, November 16, 2026):</strong> feeding the October 17 version and then the November 16 version of CVE-2026-97229 printed:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>ALERT CVE-2026-97229: KEV block added (dateAdded 2026-10-29)
</span></span><span style="display:flex;"><span>ALERT CVE-2026-97229: SSVC Exploitation none -&gt; active
</span></span></code></pre></div><p>The same two-step test on CVE-2023-54404 printed <code>record REJECTED, ADP data removed</code>. A live run against <code>delta.json</code> (fetchTime 2026-11-16T03:48:36Z, one change) fetched the record and wrote state without alerting.</p>
<p>Limits of the script as written: it fetches records one at a time, so a large delta is slow; <code>delta.json</code> only holds the latest batch, so a missed run loses changes unless you replay <code>deltaLog.json</code> as the delta guide describes; and a CVE seen for the first time with <code>poc</code> or <code>active</code> alerts immediately, which is what you want for new records but noisy on the first run. Filter to the CVEs in your inventory before alerting, and seed the state file from a full snapshot before going live. Treat it as a starting point, not production code.</p>
<h2 id="wiring-it-into-a-patch-decision">Wiring it into a patch decision</h2>
<ul>
<li><strong>KEV feed first.</strong> Poll the KEV JSON and alert on new <code>cveID</code> values in your inventory. In our window it was never behind the CVE record.</li>
<li><strong>CISA-ADP diff second.</strong> Use it for <code>poc</code> changes, SSVC values on CVEs that will never reach KEV, and rejections that should close tickets.</li>
<li><strong>Re-score on change.</strong> An Exploitation change moves the SSVC outcome; feed it back into the decision step of the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">prioritization workflow</a> rather than paging on every <code>poc</code>.</li>
<li><strong>Watch the ransomware flag separately.</strong> KEV&rsquo;s <code>knownRansomwareCampaignUse</code> changes in the catalog only; see <a href="/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/">KEV ransomware flag flips</a>.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>One 30-day window. The finding that <code>active</code> matched KEV exactly is an observation for October 17 to November 16, 2026, not a documented CISA rule.</li>
<li>Two-point diff: intermediate states inside the window are not visible.</li>
<li>Lag figures use the CISA-ADP container&rsquo;s last <code>dateUpdated</code>, which can be later than the moment the <code>kev</code> block first appeared.</li>
<li>The two missing <code>kev</code> blocks were missing as of the November 16 snapshot; they may be added later.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CISA-ADP adds a KEV block only for KEV CVEs</td>
          <td><a href="https://github.com/cisagov/vulnrichment/blob/develop/README.md">cisagov/vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC Exploitation definitions</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/exploitation.py">CERTCC/SSVC exploitation.py</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>31,312 changed records; transition counts; 34 active, all in KEV</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> tags cve_2026-10-17_0000Z and cve_2026-11-16_0000Z, our script</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>33 KEV adds Oct 17 to Nov 15; 31 with kev block; lag 1/2/4 days</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a>, CVE records at cve_2026-11-16_0000Z</td>
          <td>Verified (computed); lag qualified as upper bound</td>
      </tr>
      <tr>
          <td>CVE-2026-48356 and CVE-2026-111497 lack a kev block</td>
          <td>CVE records at cve_2026-11-16_0000Z</td>
          <td>Verified as of that snapshot</td>
      </tr>
      <tr>
          <td>Watcher output</td>
          <td>Run on November 16, 2026</td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE problemTypes and cweId: Getting CWE From CVE JSON</title><link>https://firewallsync.com/posts/cve-problemtypes-cweid-cwe-from-cve-json/</link><guid>https://firewallsync.com/posts/cve-problemtypes-cweid-cwe-from-cve-json/</guid><pubDate>Sun, 15 Nov 2026 04:50:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>The CWE for a CVE lives in problemTypes[].descriptions[].cweId, but a count across the November 15, 2026 cvelistV5 snapshot shows 24% of 2026 CVE records have no CNA cweId. Where the gaps come from, what CISA-ADP fills in (often CWE-284), and a parser that handles all four cases.</description><enclosure url="https://firewallsync.com/posts/cve-problemtypes-cweid-cwe-from-cve-json/cover_hu11811165115856360047.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-problemtypes-cweid-cwe-from-cve-json/cover_hu11811165115856360047.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-problemtypes-cweid-cwe-from-cve-json/cover_hu11811165115856360047.jpg" alt="Rows of wooden library card catalogue drawers with metal label holders, receding into soft focus" width="1200" height="630"></p><p><strong>Short answer:</strong> the CWE for a CVE is in <code>containers.cna.problemTypes[].descriptions[].cweId</code>. That field is optional, and a lot of records do not have it. Of the 97,709 CVE records published in 2026 in the cvelistV5 snapshot of November 15, 2026, <strong>24.1% have no CNA <code>cweId</code></strong>: 13,044 carry only text and 10,476 have no <code>problemTypes</code> at all. CISA-ADP fills 15,650 of those gaps, most often with CWE-284, a mapping CWE itself discourages. If you build CWE metrics from CVE JSON, read three places in a fixed order and keep track of which one answered.</p>
<p>This is a field-level guide for engineers building weakness dashboards, SAST/CVE correlation or CWE Top 25-style reports from raw CVE data. Where it fits in the larger triage flow is covered in the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>. How ADP containers work in general is in <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">CVE ADP containers: CISA-ADP, CVE Program and supplier ADPs</a>.</p>
<h2 id="the-problemtypes-structure">The problemTypes structure</h2>
<p>The <a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema</a> (v5.2.0 in the repository) defines <code>problemTypes</code> as an array with at least one item. Each item has a required <code>descriptions</code> array, and each description object looks like this:</p>
<table>
  <thead>
      <tr>
          <th>Property</th>
          <th>Required</th>
          <th>Schema constraint</th>
          <th>What it holds</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>lang</code></td>
          <td>yes</td>
          <td>language tag</td>
          <td><code>en</code>, <code>en-US</code></td>
      </tr>
      <tr>
          <td><code>description</code></td>
          <td>yes</td>
          <td>string, 1-4096 chars</td>
          <td>&ldquo;Text description of problemType, or title from CWE or OWASP.&rdquo;</td>
      </tr>
      <tr>
          <td><code>cweId</code></td>
          <td>no</td>
          <td><code>^CWE-[1-9][0-9]*$</code>, 5-9 chars</td>
          <td>&ldquo;CWE ID of the CWE that best describes this problemType entry.&rdquo;</td>
      </tr>
      <tr>
          <td><code>type</code></td>
          <td>no</td>
          <td>free string, 1-128 chars</td>
          <td>&ldquo;Problemtype source, text, OWASP, CWE, etc.,&rdquo;</td>
      </tr>
      <tr>
          <td><code>references</code></td>
          <td>no</td>
          <td>references array</td>
          <td>supporting links</td>
      </tr>
  </tbody>
</table>
<p>A well-formed CNA entry:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span><span style="color:#e6db74">&#34;problemTypes&#34;</span><span style="color:#960050;background-color:#1e0010">:</span> [
</span></span><span style="display:flex;"><span>  { <span style="color:#f92672">&#34;descriptions&#34;</span>: [
</span></span><span style="display:flex;"><span>      { <span style="color:#f92672">&#34;lang&#34;</span>: <span style="color:#e6db74">&#34;en&#34;</span>, <span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;CWE&#34;</span>, <span style="color:#f92672">&#34;cweId&#34;</span>: <span style="color:#e6db74">&#34;CWE-352&#34;</span>,
</span></span><span style="display:flex;"><span>        <span style="color:#f92672">&#34;description&#34;</span>: <span style="color:#e6db74">&#34;CWE-352 Cross-Site Request Forgery (CSRF)&#34;</span> } ] }
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>Two schema details cause most parser bugs:</p>
<ul>
<li><strong><code>problemTypes</code> is not required for a CNA.</strong> The <code>cnaPublishedContainer</code> required list is <code>providerMetadata</code>, <code>descriptions</code>, <code>affected</code> and <code>references</code>. The schema&rsquo;s description text says the field &ldquo;Must contain: At least one entry&rdquo;, but validation will not enforce that.</li>
<li><strong><code>type</code> is free text.</strong> In the full corpus the most common values are <code>CWE</code> (198,469 descriptions) and <code>text</code> (183,625), then a missing <code>type</code> (19,161), <code>Impact</code> (2,778) and lower-case <code>cwe</code> (1,586), plus a tail of one-off values such as &ldquo;RCE (Remote Code Execution)&rdquo;. <code>type: &quot;CWE&quot;</code> does not guarantee a <code>cweId</code> is present.</li>
</ul>
<p>The program rules match the schema&rsquo;s looseness. <a href="https://www.cve.org/Resources/Roles/Cnas/CNA_Rules_v4.1.0.pdf">CNA Operational Rules 4.1.0</a> (effective May 14, 2025), rule 5.1.7, say a CNA &ldquo;MUST identify the type of Vulnerability&rdquo; and that the record &ldquo;SHOULD use the Common Weakness Enumeration (CWE)&rdquo;. A text label satisfies the MUST.</p>
<h2 id="how-often-the-cna-supplies-a-cweid">How often the CNA supplies a cweId</h2>
<p>We streamed the <code>2026-11-15_all_CVEs_at_midnight</code> release zip from cvelistV5 and counted every record in state <code>PUBLISHED</code>, grouped by the year of <code>cveMetadata.datePublished</code>. Each record lands in exactly one CNA bucket.</p>
<table>
  <thead>
      <tr>
          <th>Publication year</th>
          <th>Published records</th>
          <th>CNA cweId</th>
          <th>CNA text only</th>
          <th>No problemTypes</th>
          <th>No cweId anywhere (CNA or CISA-ADP)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2024</td>
          <td>39,924</td>
          <td>25,623 (64.2%)</td>
          <td>9,054 (22.7%)</td>
          <td>5,247 (13.1%)</td>
          <td>5,366 (13.4%)</td>
      </tr>
      <tr>
          <td>2025</td>
          <td>48,150</td>
          <td>34,764 (72.2%)</td>
          <td>6,982 (14.5%)</td>
          <td>6,404 (13.3%)</td>
          <td>5,626 (11.7%)</td>
      </tr>
      <tr>
          <td>2026 (to Nov 15)</td>
          <td>97,709</td>
          <td>74,189 (75.9%)</td>
          <td>13,044 (13.3%)</td>
          <td>10,476 (10.7%)</td>
          <td>7,870 (8.1%)</td>
      </tr>
  </tbody>
</table>
<p>CNA coverage is improving year on year, but roughly one in four 2026 records still needs a fallback. Most records with a CNA <code>cweId</code> have exactly one: across all years, 157,531 of 177,478 (88.8%) list a single distinct CWE and 19,947 list two or more. Treat <code>cweId</code> as a set, not a scalar.</p>
<h2 id="where-the-gaps-come-from">Where the gaps come from</h2>
<p>The gaps are not random. They follow individual CNAs&rsquo; publishing habits, which matters because a dashboard filtered to one vendor can show zero CWE data for that vendor without anything being wrong with your parser.</p>
<table>
  <thead>
      <tr>
          <th>CNA (<code>assignerShortName</code>)</th>
          <th>2026 records</th>
          <th>Pattern</th>
          <th>Example</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>Linux</code></td>
          <td>8,262</td>
          <td>no <code>problemTypes</code> in any 2026 record</td>
          <td>CVE-2026-74687</td>
      </tr>
      <tr>
          <td><code>oracle</code></td>
          <td>4,154</td>
          <td>all text only (risk sentence such as &ldquo;Easily exploitable vulnerability allows &hellip;&rdquo;)</td>
          <td>CVE-2026-82997</td>
      </tr>
      <tr>
          <td><code>mitre</code></td>
          <td>3,773</td>
          <td>3,027 text-only, description <code>n/a</code></td>
          <td>CVE-2026-104586</td>
      </tr>
      <tr>
          <td><code>WPScan</code></td>
          <td>2,314</td>
          <td>CWE named in text with <code>type: &quot;CWE&quot;</code>, no <code>cweId</code> (2,312 records)</td>
          <td>CVE-2026-82126</td>
      </tr>
      <tr>
          <td><code>apple</code></td>
          <td>1,196</td>
          <td>all text only</td>
          <td>CVE-2026-86917</td>
      </tr>
      <tr>
          <td><code>Chrome</code></td>
          <td>3,972</td>
          <td>705 text-only (labels such as &ldquo;Inappropriate implementation&rdquo;)</td>
          <td>CVE-2026-106484</td>
      </tr>
      <tr>
          <td><code>mozilla</code></td>
          <td>968</td>
          <td>no <code>problemTypes</code> in any 2026 record</td>
          <td></td>
      </tr>
  </tbody>
</table>
<p>The WPScan pattern deserves its own handling. Its entries look like this:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{ <span style="color:#f92672">&#34;description&#34;</span>: <span style="color:#e6db74">&#34;CWE-284 Improper Access Control&#34;</span>, <span style="color:#f92672">&#34;lang&#34;</span>: <span style="color:#e6db74">&#34;en&#34;</span>, <span style="color:#f92672">&#34;type&#34;</span>: <span style="color:#e6db74">&#34;CWE&#34;</span> }
</span></span></code></pre></div><p>A parser that only reads <code>cweId</code> drops these. In 2026, 3,063 records name a CWE ID only in CNA text, and 2,559 description objects have <code>type</code> <code>CWE</code> or <code>cwe</code> with no <code>cweId</code>, 2,359 of them from WPScan. CISA&rsquo;s own <a href="https://github.com/cisagov/vulnrichment">Vulnrichment README</a> uses a WPScan record, CVE-2024-3477, as its example of CISA adding a CWE; the CNA text in that record already says &ldquo;CWE-352 Cross-Site Request Forgery (CSRF)&rdquo;.</p>
<h2 id="what-cisa-adp-adds">What CISA-ADP adds</h2>
<p>The <a href="https://github.com/cisagov/vulnrichment">Vulnrichment README</a> says that once CVEs are SSVC-scored, &ldquo;some higher-risk CVEs will also receive enrichment of CWE and/or CVSS data points, where possible&rdquo;, that CISA populates the ADP container only for CVEs &ldquo;that do not already have these fields populated by the originating CNA&rdquo;, and that &ldquo;At no point will CISA overwrite the originating CNA&rsquo;s data&rdquo;. If a CNA later adds the data, &ldquo;the CISA ADP will drop its own assessments&rdquo;. When both exist, it says the CNA&rsquo;s data &ldquo;should take precedence&rdquo;.</p>
<p>In 2026 records:</p>
<ul>
<li>CISA-ADP carries a <code>cweId</code> in 16,066 records. In 15,650 of them the CNA had none.</li>
<li>In 416 records both containers carry a <code>cweId</code>; in 101 they differ. Per the README, prefer the CNA value.</li>
<li>Of the 3,063 records where the CNA named a CWE only in text, CISA-ADP has a <code>cweId</code> in 2,591, and in 2,544 it is the same ID the text already named.</li>
<li>The only other ADP supplying <code>cweId</code> across the corpus is <code>redhat-SADP</code> (1,124 records). CISA-ADP has 44,152.</li>
</ul>
<p>The most frequent CWEs CISA-ADP adds when the CNA gave none, 2026 records:</p>
<table>
  <thead>
      <tr>
          <th>CWE</th>
          <th>Records</th>
          <th>CWE mapping note</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CWE-284 Improper Access Control</td>
          <td>3,334</td>
          <td>Pillar; usage DISCOURAGED</td>
      </tr>
      <tr>
          <td>CWE-200 Exposure of Sensitive Information</td>
          <td>975</td>
          <td></td>
      </tr>
      <tr>
          <td>CWE-79 Cross-site Scripting</td>
          <td>829</td>
          <td></td>
      </tr>
      <tr>
          <td>CWE-269 Improper Privilege Management</td>
          <td>717</td>
          <td></td>
      </tr>
      <tr>
          <td>CWE-416 Use After Free</td>
          <td>696</td>
          <td></td>
      </tr>
  </tbody>
</table>
<p>CWE-284&rsquo;s own <a href="https://cwe.mitre.org/data/definitions/284">definition page</a> (CWE 5.0) lists vulnerability mapping usage as &ldquo;DISCOURAGED&rdquo;, says the ID &ldquo;should not be used to map to real-world vulnerabilities&rdquo;, and calls it &ldquo;extremely high-level, a Pillar&rdquo; that is &ldquo;not useful for trend analysis&rdquo;. Most of those 3,334 (2,624) are Oracle records whose CNA entry is text only. CVE-2026-82997 is one, where CISA-ADP adds CWE-284 to a CNA entry that reads &ldquo;Easily exploitable vulnerability allows low privileged attacker with network access via T3, IIOP to compromise Service Delivery Platform.&rdquo; The pillar choice reflects how little the CNA text says, so treat these as &ldquo;access-control-ish, details unknown&rdquo;, not as evidence that access control is your top weakness class.</p>
<h2 id="a-parser-that-handles-all-four-cases">A parser that handles all four cases</h2>
<p>This Python 3 function (tested with Python 3.13 against the five records listed below, fetched from <code>raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/...</code>) returns the CWE set and where it came from:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> re<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>CWE_RE <span style="color:#f92672">=</span> re<span style="color:#f92672">.</span>compile(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;\bCWE-[1-9][0-9]*\b&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cwe_ids</span>(record):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Return (ids, source) for one CVE JSON 5.x record.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    source: &#39;cna&#39;, &#39;cna-text&#39;, &#39;cisa-adp&#39; or &#39;none&#39;.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    c <span style="color:#f92672">=</span> record[<span style="color:#e6db74">&#34;containers&#34;</span>]
</span></span><span style="display:flex;"><span>    texts, ids <span style="color:#f92672">=</span> [], set()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> pt <span style="color:#f92672">in</span> c[<span style="color:#e6db74">&#34;cna&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;problemTypes&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> d <span style="color:#f92672">in</span> pt<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;descriptions&#34;</span>, []):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> d<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;cweId&#34;</span>):
</span></span><span style="display:flex;"><span>                ids<span style="color:#f92672">.</span>add(d[<span style="color:#e6db74">&#34;cweId&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>                texts<span style="color:#f92672">.</span>append(d<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;description&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> ids:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> sorted(ids), <span style="color:#e6db74">&#34;cna&#34;</span>
</span></span><span style="display:flex;"><span>    found <span style="color:#f92672">=</span> {m <span style="color:#66d9ef">for</span> t <span style="color:#f92672">in</span> texts <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> CWE_RE<span style="color:#f92672">.</span>findall(t)}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> found:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> sorted(found), <span style="color:#e6db74">&#34;cna-text&#34;</span>      <span style="color:#75715e"># CWE named in prose only</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> adp[<span style="color:#e6db74">&#34;providerMetadata&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;shortName&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> pt <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;problemTypes&#34;</span>, []):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> d <span style="color:#f92672">in</span> pt<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;descriptions&#34;</span>, []):
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> d<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;cweId&#34;</span>):
</span></span><span style="display:flex;"><span>                    ids<span style="color:#f92672">.</span>add(d[<span style="color:#e6db74">&#34;cweId&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (sorted(ids), <span style="color:#e6db74">&#34;cisa-adp&#34;</span>) <span style="color:#66d9ef">if</span> ids <span style="color:#66d9ef">else</span> ([], <span style="color:#e6db74">&#34;none&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> path <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]:
</span></span><span style="display:flex;"><span>        r <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(path))
</span></span><span style="display:flex;"><span>        print(r[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>], <span style="color:#f92672">*</span>cwe_ids(r))
</span></span></code></pre></div><p>Output on the November 15, 2026 versions of those records:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2024-2905 [&#39;CWE-732&#39;] cna
</span></span><span style="display:flex;"><span>CVE-2024-3477 [&#39;CWE-352&#39;] cna-text
</span></span><span style="display:flex;"><span>CVE-2026-74687 [] none
</span></span><span style="display:flex;"><span>CVE-2026-82126 [&#39;CWE-284&#39;] cna-text
</span></span><span style="display:flex;"><span>CVE-2026-82997 [&#39;CWE-284&#39;] cisa-adp
</span></span></code></pre></div><p>Design choices, and their limits:</p>
<ul>
<li><strong>CNA text before CISA-ADP.</strong> The CNA is the authoritative source under the README&rsquo;s precedence rule, and in 2,544 of 2,591 overlapping cases the two agree anyway. Swap the order if you only trust structured fields.</li>
<li><strong>The regex can over-match.</strong> A description such as &ldquo;similar to CWE-79&rdquo; would be captured. In the corpus the prose-only CWE strings are overwhelmingly titles in the form &ldquo;CWE-NNN Name&rdquo;, but log <code>cna-text</code> results separately.</li>
<li><strong>Keep the source column.</strong> Store <code>source</code> with each row. It lets you report &ldquo;CNA-mapped&rdquo; and &ldquo;enriched&rdquo; weaknesses separately and lets you drop discouraged pillars from trend charts without losing the records.</li>
<li><strong>Re-read on updates.</strong> CISA-ADP values can disappear when a CNA adds its own. If you sync incrementally with <a href="/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/">cvelistV5 delta.json and deltaLog.json</a>, re-run the extractor on every updated record rather than caching the first answer.</li>
</ul>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Reading <code>problemTypes[0].descriptions[0].cweId</code> only.</strong> 11.2% of records with CNA CWE data list more than one distinct CWE, and some put a text entry first.</li>
<li><strong>Assuming <code>type</code> tells you what you have.</strong> Lower-case <code>cwe</code>, missing <code>type</code> and <code>Impact</code> all occur.</li>
<li><strong>Treating &ldquo;no CWE&rdquo; as &ldquo;unknown vulnerability class&rdquo; for every vendor equally.</strong> Linux kernel and Mozilla records will always be in that bucket until those CNAs change practice.</li>
<li><strong>Merging CISA-ADP CWE-284 into a Top-N chart without a label.</strong> It becomes a top-three entry by volume in 2026 for reasons unrelated to the underlying bugs.</li>
<li><strong>Mixing CWE with severity signals.</strong> CWE says what kind of bug it is, not how urgent it is. Severity and exploitation come from <a href="/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/">CVSS in the record</a>, <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">CISA&rsquo;s SSVC values</a> and KEV.</li>
</ul>
<h2 id="how-mitres-top-25-deals-with-the-same-data">How MITRE&rsquo;s Top 25 deals with the same data</h2>
<p>The <a href="https://cwe.mitre.org/top25/archive/2025/2025_methodology.html">2025 CWE Top 25 methodology</a> used 39,080 CVE records published between June 1, 2024 and June 1, 2025, including mappings &ldquo;added by CISA Vulnrichment after publication&rdquo;. About 9,468 (24%) were flagged for re-mapping review, for example where the CWE was too abstract or commonly misused. If you build an internal ranking from raw CVE data you get the unreviewed version of that dataset, so expect pillars and class-level CWEs to rank higher than they do in the published list.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>Counts come from one snapshot (the <code>cve_2026-11-15_0000Z</code> release) and change daily as records are published, updated and enriched.</li>
<li>Years are grouped by <code>datePublished</code>, not by CVE ID year.</li>
<li>We did not validate whether each CWE is the correct mapping for its vulnerability, only where it is recorded.</li>
<li>CWE usage notes are quoted from CWE 5.0 for CWE-284 only; check the definition page of any other CWE before excluding it.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>problemTypes structure, cweId pattern, CNA required fields</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CNA MUST identify vulnerability type; SHOULD use CWE (rule 5.1.7)</td>
          <td><a href="https://www.cve.org/Resources/Roles/Cnas/CNA_Rules_v4.1.0.pdf">CNA Operational Rules 4.1.0</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CNA/ADP coverage counts, CNA patterns, CISA-ADP gap-fill counts</td>
          <td>cvelistV5 release <code>cve_2026-11-15_0000Z</code>, counted with Python</td>
          <td>Verified (our count)</td>
      </tr>
      <tr>
          <td>CISA-ADP adds CWE for some CVEs, never overwrites CNA, drops on CNA update, CNA precedence</td>
          <td><a href="https://github.com/cisagov/vulnrichment">cisagov/vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CWE-284 is a Pillar with mapping usage DISCOURAGED</td>
          <td><a href="https://cwe.mitre.org/data/definitions/284">CWE-284 definition</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Top 25 2025 dataset size, window and re-mapping share</td>
          <td><a href="https://cwe.mitre.org/top25/archive/2025/2025_methodology.html">2025 CWE Top 25 methodology</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVSS v4 Supplemental Metrics in CVE Records: Who Sets Them</title><link>https://firewallsync.com/posts/cvss-v4-supplemental-metrics-in-cve-records-automatable-provider-urgency/</link><guid>https://firewallsync.com/posts/cvss-v4-supplemental-metrics-in-cve-records-automatable-provider-urgency/</guid><pubDate>Sat, 14 Nov 2026 04:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CVSS v4.0 Supplemental metrics (Safety, Automatable, Recovery, Value Density, Response Effort, Provider Urgency) never change the score. A count across every published CVE record shows only 3.4% of CNA v4.0 entries set any of them, which CNAs do, how the JSON fields and vector string drift apart, and how to use them without overtrusting them.</description><enclosure url="https://firewallsync.com/posts/cvss-v4-supplemental-metrics-in-cve-records-automatable-provider-urgency/cover_hu4154979156842941863.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cvss-v4-supplemental-metrics-in-cve-records-automatable-provider-urgency/cover_hu4154979156842941863.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cvss-v4-supplemental-metrics-in-cve-records-automatable-provider-urgency/cover_hu4154979156842941863.jpg" alt="A lit red no-entry signal with a white horizontal bar, glowing against a dark background at night" width="1200" height="630"></p><p><strong>Short answer:</strong> CVSS v4.0 added six optional <em>Supplemental</em> metrics that describe a vulnerability without changing its score. They are real fields in the CVE JSON 5 schema, but almost nobody fills them in. Across every published CVE record in the cvelistV5 snapshot of November 14, 2026, only <strong>3.4%</strong> of records with a CNA <code>cvssV4_0</code> entry set any Supplemental metric, and two vendors, Juniper and Palo Alto Networks, produce nearly a third of them (516 of 1,661 entries). If you parse them, parse both the JSON fields and the vector string, and treat them as the supplier&rsquo;s opinion at publication time.</p>
<p>This article covers the field names and values, who actually sets them, two parsing traps, and where they fit in a patch decision. For the overall flow from CVE record to ticket, see the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>. For reading the rest of a v4.0 vector, see <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">how to read a CVSS v4.0 vector string</a>.</p>
<h2 id="the-six-supplemental-metrics">The six Supplemental metrics</h2>
<p>The <a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 specification</a> introduces &ldquo;a new, optional metric group called the Supplemental metric group&rdquo; and states: &ldquo;No metric will have any impact on the final calculated CVSS score (e.g. CVSS-BTE).&rdquo; The values &ldquo;convey additional extrinsic characteristics of the vulnerability itself.&rdquo;</p>
<table>
  <thead>
      <tr>
          <th>Metric</th>
          <th>Vector</th>
          <th>Allowed values</th>
          <th>CVE JSON property</th>
          <th>JSON enum</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Safety</td>
          <td><code>S</code></td>
          <td>X, N, P</td>
          <td><code>Safety</code></td>
          <td>NEGLIGIBLE, PRESENT</td>
      </tr>
      <tr>
          <td>Automatable</td>
          <td><code>AU</code></td>
          <td>X, N, Y</td>
          <td><code>Automatable</code></td>
          <td>NO, YES</td>
      </tr>
      <tr>
          <td>Recovery</td>
          <td><code>R</code></td>
          <td>X, A, U, I</td>
          <td><code>Recovery</code></td>
          <td>AUTOMATIC, USER, IRRECOVERABLE</td>
      </tr>
      <tr>
          <td>Value Density</td>
          <td><code>V</code></td>
          <td>X, D, C</td>
          <td><code>valueDensity</code></td>
          <td>DIFFUSE, CONCENTRATED</td>
      </tr>
      <tr>
          <td>Vulnerability Response Effort</td>
          <td><code>RE</code></td>
          <td>X, L, M, H</td>
          <td><code>vulnerabilityResponseEffort</code></td>
          <td>LOW, MODERATE, HIGH</td>
      </tr>
      <tr>
          <td>Provider Urgency</td>
          <td><code>U</code></td>
          <td>X, Clear, Green, Amber, Red</td>
          <td><code>providerUrgency</code></td>
          <td>CLEAR, GREEN, AMBER, RED</td>
      </tr>
  </tbody>
</table>
<p>Every JSON enum also allows <code>NOT_DEFINED</code>, which is the schema default (<a href="https://github.com/CVEProject/cve-schema/blob/main/schema/imports/cvss/cvss-v4.0.json">cve-schema <code>cvss-v4.0.json</code></a>). In the vector, <code>X</code> means Not Defined, and the spec tells consumers to &ldquo;treat unspecified Threat, Environmental and Supplemental as Not Defined.&rdquo;</p>
<p>What each one means, in the spec&rsquo;s words:</p>
<ul>
<li><strong>Safety:</strong> whether consequences meet the IEC 61508 consequence categories. <code>P</code> (Present) versus <code>N</code> (Negligible).</li>
<li><strong>Automatable:</strong> &ldquo;Can an attacker automate exploitation events for this vulnerability across multiple targets?&rdquo; <code>Y</code> means &ldquo;Attackers can reliably automate all 4 steps of the kill chain.&rdquo;</li>
<li><strong>Recovery:</strong> whether the system recovers services automatically (<code>A</code>), needs user intervention (<code>U</code>), or is irrecoverable by the user (<code>I</code>) after an attack.</li>
<li><strong>Value Density:</strong> what one exploitation event gains the attacker. <code>D</code> (Diffuse): &ldquo;The vulnerable system has limited resources.&rdquo; <code>C</code> (Concentrated): &ldquo;The vulnerable system is rich in resources.&rdquo;</li>
<li><strong>Vulnerability Response Effort:</strong> how hard it is for the consumer to respond, from <code>L</code> (&ldquo;low/trivial&rdquo;) to <code>H</code> (&ldquo;significant and/or difficult&rdquo;).</li>
<li><strong>Provider Urgency:</strong> an optional &ldquo;pass-through&rdquo; metric for the provider&rsquo;s own assessment, from Clear (&ldquo;no urgency (Informational)&rdquo;) to Red (&ldquo;the highest urgency&rdquo;).</li>
</ul>
<p>Two schema details trip up parsers:</p>
<ol>
<li><strong>Property-name casing is inconsistent.</strong> <code>Safety</code>, <code>Automatable</code> and <code>Recovery</code> start with a capital letter; the other three are camelCase. A loop that lower-cases the first letter of every metric name will miss three of them.</li>
<li><strong>Provider Urgency uses words in the vector.</strong> The vector segment is <code>U:Amber</code>, not <code>U:A</code>, and the JSON value is <code>AMBER</code>. A regex for single-letter values silently drops it.</li>
</ol>
<h2 id="who-actually-sets-them">Who actually sets them</h2>
<p>We streamed every record in the cvelistV5 midnight release of November 14, 2026 (<code>2026-11-14_all_CVEs_at_midnight.zip.zip</code>), kept records in state <code>PUBLISHED</code>, and counted CNA <code>cvssV4_0</code> entries where either the JSON property is set to something other than <code>NOT_DEFINED</code> or the vector has a non-<code>X</code> segment. Counts below are from that run; the method is described so you can repeat it.</p>
<table>
  <thead>
      <tr>
          <th>Measure</th>
          <th>Count</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Published records with a CNA <code>cvssV4_0</code> entry</td>
          <td>46,973 (47,055 entries)</td>
      </tr>
      <tr>
          <td>Records with at least one Supplemental metric</td>
          <td>1,609 (3.4%)</td>
      </tr>
      <tr>
          <td>CNA entries with at least one Supplemental metric</td>
          <td>1,661</td>
      </tr>
      <tr>
          <td>Entries with all six set</td>
          <td>257</td>
      </tr>
      <tr>
          <td>CNAs publishing v4.0 / of those, using Supplemental at least once</td>
          <td>326 / 149</td>
      </tr>
  </tbody>
</table>
<p>By metric, across the 1,661 entries:</p>
<table>
  <thead>
      <tr>
          <th>Metric</th>
          <th>Entries set</th>
          <th>Value split</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Automatable</td>
          <td>1,210</td>
          <td>YES 732, NO 478</td>
      </tr>
      <tr>
          <td>Provider Urgency</td>
          <td>1,152</td>
          <td>AMBER 617, GREEN 212, RED 208, CLEAR 115</td>
      </tr>
      <tr>
          <td>Response Effort</td>
          <td>1,123</td>
          <td>MODERATE 726, LOW 330, HIGH 67</td>
      </tr>
      <tr>
          <td>Recovery</td>
          <td>1,067</td>
          <td>USER 679, AUTOMATIC 345, IRRECOVERABLE 43</td>
      </tr>
      <tr>
          <td>Value Density</td>
          <td>891</td>
          <td>DIFFUSE 448, CONCENTRATED 443</td>
      </tr>
      <tr>
          <td>Safety</td>
          <td>475</td>
          <td>NEGLIGIBLE 314, PRESENT 161</td>
      </tr>
  </tbody>
</table>
<p>Who sets them matters more than the totals:</p>
<table>
  <thead>
      <tr>
          <th>CNA</th>
          <th>CNA v4.0 entries</th>
          <th>With Supplemental</th>
          <th>Typical pattern</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>VulDB</td>
          <td>14,115</td>
          <td>0</td>
          <td>none</td>
      </tr>
      <tr>
          <td>VulnCheck</td>
          <td>11,610</td>
          <td>2</td>
          <td>none</td>
      </tr>
      <tr>
          <td>GitHub (GitHub_M)</td>
          <td>5,841</td>
          <td>31</td>
          <td>occasional</td>
      </tr>
      <tr>
          <td>icscert (CISA ICS)</td>
          <td>947</td>
          <td>0</td>
          <td>none</td>
      </tr>
      <tr>
          <td>Juniper</td>
          <td>426</td>
          <td>258</td>
          <td>AU, R, RE, often V</td>
      </tr>
      <tr>
          <td>Palo Alto Networks</td>
          <td>263</td>
          <td>258</td>
          <td>AU, R, V, RE, U</td>
      </tr>
  </tbody>
</table>
<p>Other frequent users in the snapshot include bcorg (mostly <code>U</code> only), Apache, OpenText, GovTech CSG and Google Cloud (mostly <code>U</code> only). By CVE ID year, 880 of the 1,661 entries are 2026 IDs, 456 are 2025 and 301 are 2024.</p>
<p>The practical reading: a missing <code>AU</code> or <code>U</code> is a statement about which CNA wrote the record, not about the vulnerability. The two largest producers of v4.0 scores leave every Supplemental metric undefined.</p>
<p>Only five ADP <code>cvssV4_0</code> entries exist in the snapshot (all CISA-ADP), and none set a Supplemental metric. CISA&rsquo;s exploitation and automatability signals arrive through SSVC instead; see <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">how to read SSVC values in a CVE record</a>.</p>
<h2 id="trap-the-vector-and-the-json-fields-drift-apart">Trap: the vector and the JSON fields drift apart</h2>
<p>The schema stores each Supplemental value twice: once as a segment in <code>vectorString</code> and once as its own property. In 61 CNA entries they disagree on which metrics are set. GitHub (22), NLnet Labs (14) and Vaadin (7) account for most of them. Two real shapes:</p>
<ul>
<li><code>CVE-2026-110803</code> (Vaadin): JSON sets all six properties (<code>Automatable: YES</code>, <code>providerUrgency: CLEAR</code> and so on) while the vector ends at <code>SA:N</code> with no Supplemental segments.</li>
<li><code>CVE-2026-41292</code> (NLnet Labs): the vector ends <code>E:U/U:Red</code>, but no <code>providerUrgency</code> property exists.</li>
</ul>
<p>We found no entry where both sides were set to different defined values; the drift is always one side missing. Tools that read only <code>vectorString</code> or only the JSON properties will each miss part of the data.</p>
<h2 id="a-parser-that-reads-both-and-flags-drift">A parser that reads both and flags drift</h2>
<p>This Python 3 function (run with Python 3.13 against the three records named in this article, downloaded from <code>raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/...</code>) returns defined Supplemental values from a <code>cvssV4_0</code> object, preferring the vector and falling back to JSON, and reports mismatches:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># vector abbreviation -&gt; (JSON property in cvss-v4.0.json, vector value -&gt; JSON enum)</span>
</span></span><span style="display:flex;"><span>SUPPLEMENTAL <span style="color:#f92672">=</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;S&#34;</span>:  (<span style="color:#e6db74">&#34;Safety&#34;</span>, {<span style="color:#e6db74">&#34;N&#34;</span>: <span style="color:#e6db74">&#34;NEGLIGIBLE&#34;</span>, <span style="color:#e6db74">&#34;P&#34;</span>: <span style="color:#e6db74">&#34;PRESENT&#34;</span>}),
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;AU&#34;</span>: (<span style="color:#e6db74">&#34;Automatable&#34;</span>, {<span style="color:#e6db74">&#34;N&#34;</span>: <span style="color:#e6db74">&#34;NO&#34;</span>, <span style="color:#e6db74">&#34;Y&#34;</span>: <span style="color:#e6db74">&#34;YES&#34;</span>}),
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;R&#34;</span>:  (<span style="color:#e6db74">&#34;Recovery&#34;</span>, {<span style="color:#e6db74">&#34;A&#34;</span>: <span style="color:#e6db74">&#34;AUTOMATIC&#34;</span>, <span style="color:#e6db74">&#34;U&#34;</span>: <span style="color:#e6db74">&#34;USER&#34;</span>, <span style="color:#e6db74">&#34;I&#34;</span>: <span style="color:#e6db74">&#34;IRRECOVERABLE&#34;</span>}),
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;V&#34;</span>:  (<span style="color:#e6db74">&#34;valueDensity&#34;</span>, {<span style="color:#e6db74">&#34;D&#34;</span>: <span style="color:#e6db74">&#34;DIFFUSE&#34;</span>, <span style="color:#e6db74">&#34;C&#34;</span>: <span style="color:#e6db74">&#34;CONCENTRATED&#34;</span>}),
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;RE&#34;</span>: (<span style="color:#e6db74">&#34;vulnerabilityResponseEffort&#34;</span>, {<span style="color:#e6db74">&#34;L&#34;</span>: <span style="color:#e6db74">&#34;LOW&#34;</span>, <span style="color:#e6db74">&#34;M&#34;</span>: <span style="color:#e6db74">&#34;MODERATE&#34;</span>, <span style="color:#e6db74">&#34;H&#34;</span>: <span style="color:#e6db74">&#34;HIGH&#34;</span>}),
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;U&#34;</span>:  (<span style="color:#e6db74">&#34;providerUrgency&#34;</span>, {<span style="color:#e6db74">&#34;Clear&#34;</span>: <span style="color:#e6db74">&#34;CLEAR&#34;</span>, <span style="color:#e6db74">&#34;Green&#34;</span>: <span style="color:#e6db74">&#34;GREEN&#34;</span>, <span style="color:#e6db74">&#34;Amber&#34;</span>: <span style="color:#e6db74">&#34;AMBER&#34;</span>, <span style="color:#e6db74">&#34;Red&#34;</span>: <span style="color:#e6db74">&#34;RED&#34;</span>}),
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">supplemental</span>(cvss4):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Return {json_field: value} for defined Supplemental metrics, plus conflicts.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    segs <span style="color:#f92672">=</span> dict(p<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;:&#34;</span>, <span style="color:#ae81ff">1</span>) <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> cvss4<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;vectorString&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;/&#34;</span>)[<span style="color:#ae81ff">1</span>:] <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;:&#34;</span> <span style="color:#f92672">in</span> p)
</span></span><span style="display:flex;"><span>    out, conflicts <span style="color:#f92672">=</span> {}, []
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> abbr, (field, mapping) <span style="color:#f92672">in</span> SUPPLEMENTAL<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>        from_vec <span style="color:#f92672">=</span> mapping<span style="color:#f92672">.</span>get(segs<span style="color:#f92672">.</span>get(abbr))            <span style="color:#75715e"># None if absent or X</span>
</span></span><span style="display:flex;"><span>        from_json <span style="color:#f92672">=</span> cvss4<span style="color:#f92672">.</span>get(field)
</span></span><span style="display:flex;"><span>        from_json <span style="color:#f92672">=</span> <span style="color:#66d9ef">None</span> <span style="color:#66d9ef">if</span> from_json <span style="color:#f92672">in</span> (<span style="color:#66d9ef">None</span>, <span style="color:#e6db74">&#34;NOT_DEFINED&#34;</span>) <span style="color:#66d9ef">else</span> from_json
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> from_vec <span style="color:#f92672">!=</span> from_json:          <span style="color:#75715e"># differs, or one side missing</span>
</span></span><span style="display:flex;"><span>            conflicts<span style="color:#f92672">.</span>append((field, from_vec, from_json))
</span></span><span style="display:flex;"><span>        value <span style="color:#f92672">=</span> from_vec <span style="color:#f92672">or</span> from_json
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> value:
</span></span><span style="display:flex;"><span>            out[field] <span style="color:#f92672">=</span> value
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out, conflicts
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    record <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> record[<span style="color:#e6db74">&#34;containers&#34;</span>][<span style="color:#e6db74">&#34;cna&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;cvssV4_0&#34;</span> <span style="color:#f92672">in</span> m:
</span></span><span style="display:flex;"><span>            vals, conflicts <span style="color:#f92672">=</span> supplemental(m[<span style="color:#e6db74">&#34;cvssV4_0&#34;</span>])
</span></span><span style="display:flex;"><span>            print(record[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>], vals)
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> conflicts:
</span></span><span style="display:flex;"><span>                print(<span style="color:#e6db74">&#34;  vector/JSON mismatch:&#34;</span>, c)
</span></span></code></pre></div><p>Output for <code>CVE-2024-0012</code> (two v4.0 entries, one per scenario) and <code>CVE-2026-41292</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2024-0012 {&#39;Automatable&#39;: &#39;NO&#39;, &#39;Recovery&#39;: &#39;USER&#39;, &#39;valueDensity&#39;: &#39;CONCENTRATED&#39;, &#39;vulnerabilityResponseEffort&#39;: &#39;HIGH&#39;, &#39;providerUrgency&#39;: &#39;RED&#39;}
</span></span><span style="display:flex;"><span>CVE-2024-0012 {&#39;Automatable&#39;: &#39;NO&#39;, &#39;Recovery&#39;: &#39;USER&#39;, &#39;valueDensity&#39;: &#39;CONCENTRATED&#39;, &#39;vulnerabilityResponseEffort&#39;: &#39;HIGH&#39;, &#39;providerUrgency&#39;: &#39;RED&#39;}
</span></span><span style="display:flex;"><span>CVE-2026-41292 {&#39;providerUrgency&#39;: &#39;RED&#39;}
</span></span><span style="display:flex;"><span>  vector/JSON mismatch: (&#39;providerUrgency&#39;, &#39;RED&#39;, None)
</span></span></code></pre></div><p>This is illustrative ingestion code: it reads the CNA container only and does not validate the rest of the vector. Store the CNA short name next to each value so a dashboard can show who said it.</p>
<h2 id="automatable-cvss-au-versus-ssvc-automatable">Automatable: CVSS AU versus SSVC Automatable</h2>
<p>CVSS <code>AU</code> and SSVC Automatable are close cousins. CERT/CC&rsquo;s SSVC decision point asks &ldquo;Can an attacker reliably automate creating exploitation events for this vulnerability?&rdquo; and defines yes as &ldquo;Attackers can reliably automate steps 1-4 of the kill chain&rdquo; (<a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/automatable.py">CERTCC/SSVC <code>automatable.py</code></a>). The CVSS definition likewise refers to &ldquo;all 4 steps of the kill chain.&rdquo;</p>
<p>They can still disagree within a single record. In <code>CVE-2024-0012</code>, the PAN-OS management web interface authentication bypass, Palo Alto Networks&rsquo; CVSS vector says <code>AU:N</code>. The CISA-ADP SSVC entry in the same record, timestamped 2024-11-18, says <code>Exploitation: active</code>, <code>Automatable: yes</code>, <code>Technical Impact: total</code>. CISA added the CVE to KEV on 2024-11-18.</p>
<p>The record does not explain the difference, so we will not guess at it. The lesson for tooling is narrower: the CNA&rsquo;s <code>AU</code> is the CNA&rsquo;s assessment, and nothing in the schema ties it to exploitation evidence elsewhere in the record. Of the 117 KEV CVEs with a CNA v4.0 entry in this snapshot, 24 carry any Supplemental metric. KEV status comes from the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV catalog</a>, not from CVSS.</p>
<h2 id="how-to-use-them-in-a-patch-decision">How to use them in a patch decision</h2>
<ul>
<li><strong>Display, do not score.</strong> Show <code>AU</code>, <code>R</code>, <code>RE</code> and <code>U</code> on the ticket as supplier context. The spec is explicit that they do not change the CVSS number, so do not fold them into one.</li>
<li><strong>Prefer SSVC and KEV for exploitation-driven urgency.</strong> Where CISA-ADP has SSVC values, they reflect exploitation status; CVSS Supplemental values do not. See <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a> for how the signals divide the work.</li>
<li><strong>Use Response Effort for scheduling, not priority.</strong> <code>RE:H</code> (as in CVE-2024-0012) is a planning hint: the fix needs real work, so book the window early.</li>
<li><strong>Treat Safety: Present as a routing flag.</strong> 161 entries set <code>S:P</code>. If you run OT, medical or vehicle systems, route those to the owners who understand IEC 61508 consequences.</li>
<li><strong>Never require a value.</strong> A rule like &ldquo;page on <code>U:Red</code>&rdquo; fires for 208 entries in the entire CVE corpus and stays silent for every VulDB, VulnCheck and CISA ICS v4.0 score.</li>
</ul>
<h2 id="common-mistakes">Common mistakes</h2>
<ol>
<li>Reading only <code>vectorString</code> or only the JSON properties. Read both; 61 CNA entries differ.</li>
<li>Assuming camelCase for all six JSON names. Three are capitalised.</li>
<li>Parsing <code>U</code> as a single letter. Its values are <code>Clear</code>, <code>Green</code>, <code>Amber</code>, <code>Red</code>.</li>
<li>Treating an absent <code>AU</code> as &ldquo;not automatable&rdquo;. It almost always means &ldquo;not evaluated&rdquo;.</li>
<li>Adding Supplemental metrics into a custom score and labelling it CVSS. The spec says they do not affect the score; a blended number is your own model, so name it as such.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>Counts come from one cvelistV5 snapshot (midnight UTC, November 14, 2026) and cover CNA containers in published records of every CVE year; they will change as CNAs publish and update records. We did not check whether CNAs revise Supplemental values over time, and we have not assessed whether any CNA&rsquo;s values are accurate. The per-CNA table names the short names used in <code>cveMetadata.assignerShortName</code>.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Six Supplemental metrics, allowed vector values, &ldquo;No metric will have any impact on the final calculated CVSS score&rdquo;</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">FIRST CVSS v4.0 Specification Document</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>JSON property names, enums, NOT_DEFINED default</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/imports/cvss/cvss-v4.0.json">cve-schema cvss-v4.0.json</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Corpus counts (3.4%, per-metric and per-CNA figures, 61 mismatches, 257 with all six)</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/releases">cvelistV5 release <code>cve_2026-11-14_0000Z</code></a>, our count</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>CVE-2024-0012 AU:N, U:Red, RE:H; CISA-ADP SSVC Automatable: yes</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2024/0xxx/CVE-2024-0012.json">CVE-2024-0012 record in cvelistV5</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2024-0012 added to KEV 2024-11-18; KEV CVE counts</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC Automatable definition</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/src/ssvc/decision_points/ssvc/automatable.py">CERTCC/SSVC automatable.py</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE JSON CVSS Metrics: cvssV4_0, cvssV3_1, baseScore Trap</title><link>https://firewallsync.com/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/</link><guid>https://firewallsync.com/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/</guid><pubDate>Fri, 13 Nov 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>How the metrics array in a CVE JSON 5 record actually looks across all 96,879 CVEs published in 2026: which CVSS versions CNAs ship, when CISA-ADP fills the gap, what scenarios mean, why cvssV4_0.baseScore is sometimes a Base+Threat score, and a tested script that picks one score per record and says where it came from.</description><enclosure url="https://firewallsync.com/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/cover_hu14428814069900906085.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/cover_hu14428814069900906085.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-json-cvss-metrics-cvssv4-0-cvssv3-1-basescore-trap/cover_hu14428814069900906085.jpg" alt="Black-and-white close-up of two round pressure gauges with needles and a regulator valve on old industrial pipework" width="1200" height="630"></p><p><strong>Short answer.</strong> A CVE JSON 5 record can hold zero, one or several CVSS scores, from the CNA and from Authorized Data Publishers such as CISA-ADP, in different CVSS versions and for different scenarios. Across the 96,879 CVEs published in 2026, under a third carry a CNA CVSS v4.0 score, a fifth carry no CNA score at all, and the field called <code>baseScore</code> in a v4.0 entry sometimes holds a Base+Threat score. If your pipeline reads &ldquo;the CVSS score&rdquo; from a record, it is making several choices whether you wrote them down or not. This article shows what the field-level data looks like and gives a tested selection script.</p>
<p><strong>Scope.</strong> We read <code>CVE_Record_Format.json</code> and <code>imports/cvss/cvss-v4.0.json</code> from the CVEProject/cve-schema main branch on November 13, 2026, and streamed every record in the cvelistV5 release <code>cve_2026-11-13_0000Z</code> (<code>2026-11-13_all_CVEs_at_midnight.zip</code>). Counts cover records in state PUBLISHED with a <code>datePublished</code> in 2026. Scores were recomputed with the Red Hat <code>cvss</code> Python library v3.6 on Python 3. For how to read a v4.0 vector itself, see <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">How to Read a CVSS v4.0 Vector String</a>; for where these scores sit in a patch decision, see the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>.</p>
<h2 id="what-the-schema-says">What the schema says</h2>
<p><code>metrics</code> lives in a container: the CNA&rsquo;s (<code>containers.cna.metrics</code>) or an ADP&rsquo;s (<code>containers.adp[].metrics</code>). It is not required in the CNA published container, whose required fields are <code>providerMetadata</code>, <code>descriptions</code>, <code>affected</code> and <code>references</code>.</p>
<p>When present, <code>metrics</code> is an array (<code>minItems: 1</code>). The schema&rsquo;s description of an entry is &ldquo;impact type information (e.g. a text description, CVSSv2, CVSSv3, CVSSV4, etc.)&rdquo;, and it uses <code>anyOf</code> to require at least one of these keys:</p>
<table>
  <thead>
      <tr>
          <th>Key</th>
          <th>Content</th>
          <th>Validated against</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>cvssV4_0</code></td>
          <td>CVSS v4.0 object</td>
          <td><code>imports/cvss/cvss-v4.0.json</code> (FIRST schema)</td>
      </tr>
      <tr>
          <td><code>cvssV3_1</code></td>
          <td>CVSS v3.1 object</td>
          <td><code>imports/cvss/cvss-v3.1.json</code></td>
      </tr>
      <tr>
          <td><code>cvssV3_0</code></td>
          <td>CVSS v3.0 object</td>
          <td><code>imports/cvss/cvss-v3.0.json</code></td>
      </tr>
      <tr>
          <td><code>cvssV2_0</code></td>
          <td>CVSS v2.0 object</td>
          <td><code>imports/cvss/cvss-v2.0.json</code></td>
      </tr>
      <tr>
          <td><code>other</code></td>
          <td><code>{type, content}</code>: &ldquo;A non-standard impact description, may be prose or JSON block&rdquo;</td>
          <td>free-form</td>
      </tr>
  </tbody>
</table>
<p>Two optional siblings matter for parsing:</p>
<ul>
<li><strong><code>format</code></strong>: &ldquo;Name of the scoring format.&rdquo; In practice, 55,806 CNA entries in 2026 set it to <code>CVSS</code>, so do not use it to detect the version. Use the key name.</li>
<li><strong><code>scenarios</code></strong>: an array of <code>{lang, value}</code>. The schema says: &ldquo;If no specific scenario is given, GENERAL is used as the default and applies when no more specific metric matches.&rdquo;</li>
</ul>
<p>The v4.0 object requires <code>version</code>, <code>vectorString</code>, <code>baseScore</code> and <code>baseSeverity</code>, and allows the individual metric fields: Base, Threat (<code>exploitMaturity</code>), Environmental (<code>modified*</code> and the requirement fields) and Supplemental (<code>Safety</code>, <code>Automatable</code>, <code>Recovery</code>, <code>valueDensity</code>, <code>vulnerabilityResponseEffort</code>, <code>providerUrgency</code>). There is <strong>no</strong> <code>threatScore</code> or <code>environmentalScore</code> field. That gap causes the trap described below.</p>
<p>ADP containers use the same <code>metrics</code> structure. CISA-ADP also uses <code>other</code> for its own data: in 2026 records we counted 90,133 <code>other</code> entries of type <code>ssvc</code> and 215 of type <code>kev</code>. Those are what <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">CISA Vulnrichment&rsquo;s SSVC values</a> and the KEV date come from, so a parser that keeps only CVSS keys throws them away.</p>
<h2 id="what-2026-records-actually-contain">What 2026 records actually contain</h2>
<table>
  <thead>
      <tr>
          <th>CNA CVSS formats present (2026 published records)</th>
          <th>Records</th>
          <th>Share</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>v3.1 only</td>
          <td>45,507</td>
          <td>47.0%</td>
      </tr>
      <tr>
          <td>none</td>
          <td>20,408</td>
          <td>21.1%</td>
      </tr>
      <tr>
          <td>v3.1 + v4.0</td>
          <td>12,458</td>
          <td>12.9%</td>
      </tr>
      <tr>
          <td>v4.0 only</td>
          <td>11,296</td>
          <td>11.7%</td>
      </tr>
      <tr>
          <td>v2.0 + v3.0 + v3.1 + v4.0</td>
          <td>6,370</td>
          <td>6.6%</td>
      </tr>
      <tr>
          <td>v3.0 only</td>
          <td>612</td>
          <td>0.6%</td>
      </tr>
      <tr>
          <td>v3.0 + v4.0</td>
          <td>216</td>
          <td>0.2%</td>
      </tr>
      <tr>
          <td>v2.0 + v3.1 + v4.0</td>
          <td>12</td>
          <td>&lt;0.1%</td>
      </tr>
      <tr>
          <td><strong>Total</strong></td>
          <td><strong>96,879</strong></td>
          <td></td>
      </tr>
  </tbody>
</table>
<p>Totals across those rows: 30,352 records (31.3%) had a CNA v4.0 entry, and 64,347 (66.4%) a CNA v3.1 entry.</p>
<p>v4.0 adoption is concentrated. VulnCheck (10,176 records), VulDB (6,370) and GitHub (4,287) published 68.6% of the CNA v4.0 records. VulDB&rsquo;s 6,370 v4.0 records match the 6,370 records in the four-version row, and the VulDB record we opened (CVE-2026-82598) lists v4.0, v3.1, v3.0 and v2.0 entries side by side. A parser that takes the first entry in the array gets whichever one the CNA happened to list first.</p>
<p>Of the 20,408 records with no CNA CVSS, <strong>15,148</strong> had a CVSS v3.1 entry from CISA-ADP, and <strong>5,260</strong> had no CVSS from either the CNA or CISA-ADP. (A few of those carry a supplier ADP score or a CNA <code>other</code> rating such as &ldquo;Red Hat severity rating&rdquo;, which appeared in 1,476 CNA entries.) Only 4 CISA-ADP entries in 2026 used v4.0.</p>
<h2 id="the-basescore-trap">The baseScore trap</h2>
<p>The CVSS v4.0 specification names scores by the metric groups used: CVSS-B (Base), CVSS-BT (Base and Threat), CVSS-BE and CVSS-BTE. Our <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">vector-string article</a> covers that nomenclature. The CVE schema, though, gives the v4.0 object only one score field, <code>baseScore</code>. When a CNA puts a Threat metric in the vector, it has to decide which number goes in that field. CNAs decide differently.</p>
<p>We took every CNA <code>cvssV4_0</code> entry from 2026 whose <code>vectorString</code> sets Exploit Maturity to something other than X, recomputed the score with and without the <code>E:</code> metric, and compared both to the stored <code>baseScore</code>:</p>
<table>
  <thead>
      <tr>
          <th>Stored <code>baseScore</code> equals&hellip;</th>
          <th>Entries</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Base-only score (CVSS-B)</td>
          <td>5,816</td>
      </tr>
      <tr>
          <td>Base+Threat score (CVSS-BT)</td>
          <td>1,067</td>
      </tr>
      <tr>
          <td>Both equal (E metric did not change the score)</td>
          <td>68</td>
      </tr>
      <tr>
          <td>Neither</td>
          <td>3</td>
      </tr>
      <tr>
          <td><strong>Total with E set</strong></td>
          <td><strong>6,954</strong></td>
      </tr>
  </tbody>
</table>
<p>The split follows the CNA. VulDB accounts for 5,790 of the Base-only entries. GitHub (432), Palo Alto Networks (120), QNAP (106) and NETGEAR (48) lead the CVSS-BT side. Two real records show the difference:</p>
<ul>
<li><strong>CVE-2026-82598 (VulDB)</strong>: vector <code>CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P</code>, stored <code>baseScore</code> 6.9. The full vector computes to 5.5; without <code>E:P</code>, 6.9. The stored number is CVSS-B.</li>
<li><strong>CVE-2026-41129 (GitHub, a Craft CMS SSRF)</strong>: vector <code>CVSS:4.0/AV:N/AC:L/AT:N/PR:H/UI:N/VC:H/VI:L/VA:N/SC:N/SI:N/SA:N/E:P</code>, stored <code>baseScore</code> 5.5, severity MEDIUM. Without <code>E:P</code> the score is 7.0 (HIGH). The stored number is CVSS-BT.</li>
</ul>
<p>Both are valid by the schema. But if your SLA says &ldquo;High and above within 14 days&rdquo;, the second record misses the bar because the CNA folded in Exploit Maturity, and the first meets the bar or misses it purely on base impact. If you also add an exploitation signal from <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV</a> or <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS</a>, a CVSS-BT number counts exploit evidence twice in one direction or the other. Decide which you want, recompute it from the vector, and label it.</p>
<p>Exploit Maturity values seen in CNA v4.0 entries in 2026: <code>NOT_DEFINED</code> 10,620, field absent 19,051, <code>UNREPORTED</code> 482, <code>PROOF_OF_CONCEPT</code> 210 and <code>ATTACKED</code> 41. Those counts come from the <code>exploitMaturity</code> JSON field, which not every producer fills even when the vector sets <code>E:</code>. Read the vector, not the field.</p>
<h2 id="scenarios-one-cve-several-scores">Scenarios: one CVE, several scores</h2>
<p>3,543 CNA metrics entries in 2026 named a scenario other than GENERAL. F5&rsquo;s CVE-2026-41225 is a clean example. It has three entries:</p>
<table>
  <thead>
      <tr>
          <th>Scenario</th>
          <th>Format</th>
          <th>Score</th>
          <th>Vector</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>GENERAL</td>
          <td>cvssV3_1</td>
          <td>7.2</td>
          <td><code>CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H</code></td>
      </tr>
      <tr>
          <td>GENERAL</td>
          <td>cvssV4_0</td>
          <td>8.6</td>
          <td><code>CVSS:4.0/AV:N/AC:L/AT:N/PR:H/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N</code></td>
      </tr>
      <tr>
          <td>Appliance Mode</td>
          <td>cvssV3_1</td>
          <td>9.1</td>
          <td><code>CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:C/C:H/I:H/A:H</code></td>
      </tr>
  </tbody>
</table>
<p>If you run the product in the named configuration, the scenario-specific score is the one that applies to you. A dashboard that shows the maximum overstates risk for everyone else, and one that shows the first entry may understate it for appliance-mode customers. Pick GENERAL by default, and let asset metadata select a named scenario when it matches.</p>
<h2 id="cna-and-cisa-adp-disagree-more-than-you-would-expect">CNA and CISA-ADP disagree more than you would expect</h2>
<p>CISA-ADP mostly adds CVSS where the CNA has none (see <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">ADP containers</a>). In 468 records from 2026 both published a v3.1 score. In 58 of those (12.4%) the base scores differed: CISA-ADP was higher in 31 and lower in 27. Some gaps are large. In CVE-2026-41080 (libexpat hash flooding, assigned by MITRE), the CNA entry is <code>AV:L/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L</code> (2.9) and CISA-ADP&rsquo;s is <code>AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N</code> (7.5). Those two analysts read the impact very differently. Averaging them would produce a number nobody asserted. Store both, with provider, and let your policy say which one drives the SLA.</p>
<h2 id="a-tested-script-pick-one-score-and-keep-its-provenance">A tested script: pick one score and keep its provenance</h2>
<p>This script applies a fixed order: CNA before CISA-ADP, v4.0 before v3.1 before v3.0, GENERAL (or unlabelled) scenarios before named ones. For v4.0 it recomputes CVSS-B and CVSS-BT from the vector and says which one <code>baseScore</code> holds. It needs <code>pip install cvss</code> (we used v3.6).</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Pick one CVSS entry per CVE JSON 5.x record, and say where it came from.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Order: CNA v4.0 &gt; CNA v3.1 &gt; CNA v3.0 &gt; CISA-ADP v3.1. GENERAL scenario first.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Requires: pip install cvss (Red Hat library, v3.6 tested).&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> cvss <span style="color:#f92672">import</span> CVSS4
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>ORDER <span style="color:#f92672">=</span> [<span style="color:#e6db74">&#34;cvssV4_0&#34;</span>, <span style="color:#e6db74">&#34;cvssV3_1&#34;</span>, <span style="color:#e6db74">&#34;cvssV3_0&#34;</span>]
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">is_general</span>(entry):
</span></span><span style="display:flex;"><span>    sc <span style="color:#f92672">=</span> entry<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;scenarios&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#f92672">not</span> sc <span style="color:#f92672">or</span> any(s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;value&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)<span style="color:#f92672">.</span>upper() <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;GENERAL&#34;</span> <span style="color:#66d9ef">for</span> s <span style="color:#f92672">in</span> sc)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">candidates</span>(record):
</span></span><span style="display:flex;"><span>    cna <span style="color:#f92672">=</span> record[<span style="color:#e6db74">&#34;containers&#34;</span>][<span style="color:#e6db74">&#34;cna&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, [])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> fmt <span style="color:#f92672">in</span> ORDER:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> sorted(cna, key<span style="color:#f92672">=</span><span style="color:#66d9ef">lambda</span> e: <span style="color:#f92672">not</span> is_general(e)):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> fmt <span style="color:#f92672">in</span> e:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">yield</span> <span style="color:#e6db74">&#34;cna&#34;</span>, fmt, e
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> record[<span style="color:#e6db74">&#34;containers&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> adp[<span style="color:#e6db74">&#34;providerMetadata&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;shortName&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []):
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;cvssV3_1&#34;</span> <span style="color:#f92672">in</span> e:
</span></span><span style="display:flex;"><span>                    <span style="color:#66d9ef">yield</span> <span style="color:#e6db74">&#34;CISA-ADP&#34;</span>, <span style="color:#e6db74">&#34;cvssV3_1&#34;</span>, e
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">pick</span>(record):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> source, fmt, e <span style="color:#f92672">in</span> candidates(record):
</span></span><span style="display:flex;"><span>        m <span style="color:#f92672">=</span> e[fmt]
</span></span><span style="display:flex;"><span>        out <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;source&#34;</span>: source, <span style="color:#e6db74">&#34;format&#34;</span>: fmt, <span style="color:#e6db74">&#34;stored&#34;</span>: m[<span style="color:#e6db74">&#34;baseScore&#34;</span>],
</span></span><span style="display:flex;"><span>               <span style="color:#e6db74">&#34;vector&#34;</span>: m[<span style="color:#e6db74">&#34;vectorString&#34;</span>],
</span></span><span style="display:flex;"><span>               <span style="color:#e6db74">&#34;scenario&#34;</span>: [s[<span style="color:#e6db74">&#34;value&#34;</span>] <span style="color:#66d9ef">for</span> s <span style="color:#f92672">in</span> e<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;scenarios&#34;</span>, [])] <span style="color:#f92672">or</span> [<span style="color:#e6db74">&#34;(none)&#34;</span>]}
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> fmt <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;cvssV4_0&#34;</span>:
</span></span><span style="display:flex;"><span>            parts <span style="color:#f92672">=</span> m[<span style="color:#e6db74">&#34;vectorString&#34;</span>]<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;/&#34;</span>)
</span></span><span style="display:flex;"><span>            base_only <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;/&#34;</span><span style="color:#f92672">.</span>join(p <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> parts <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> p<span style="color:#f92672">.</span>startswith(<span style="color:#e6db74">&#34;E:&#34;</span>))
</span></span><span style="display:flex;"><span>            out[<span style="color:#e6db74">&#34;cvss_b&#34;</span>] <span style="color:#f92672">=</span> float(CVSS4(base_only)<span style="color:#f92672">.</span>base_score)
</span></span><span style="display:flex;"><span>            out[<span style="color:#e6db74">&#34;cvss_bt&#34;</span>] <span style="color:#f92672">=</span> float(CVSS4(m[<span style="color:#e6db74">&#34;vectorString&#34;</span>])<span style="color:#f92672">.</span>base_score)
</span></span><span style="display:flex;"><span>            out[<span style="color:#e6db74">&#34;stored_is&#34;</span>] <span style="color:#f92672">=</span> (<span style="color:#e6db74">&#34;B and BT equal&#34;</span> <span style="color:#66d9ef">if</span> out[<span style="color:#e6db74">&#34;cvss_b&#34;</span>] <span style="color:#f92672">==</span> out[<span style="color:#e6db74">&#34;cvss_bt&#34;</span>] <span style="color:#66d9ef">else</span>
</span></span><span style="display:flex;"><span>                                <span style="color:#e6db74">&#34;CVSS-B&#34;</span> <span style="color:#66d9ef">if</span> out[<span style="color:#e6db74">&#34;stored&#34;</span>] <span style="color:#f92672">==</span> out[<span style="color:#e6db74">&#34;cvss_b&#34;</span>] <span style="color:#66d9ef">else</span>
</span></span><span style="display:flex;"><span>                                <span style="color:#e6db74">&#34;CVSS-BT&#34;</span> <span style="color:#66d9ef">if</span> out[<span style="color:#e6db74">&#34;stored&#34;</span>] <span style="color:#f92672">==</span> out[<span style="color:#e6db74">&#34;cvss_bt&#34;</span>] <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;neither&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> {<span style="color:#e6db74">&#34;source&#34;</span>: <span style="color:#66d9ef">None</span>, <span style="color:#e6db74">&#34;note&#34;</span>: <span style="color:#e6db74">&#34;no CVSS from CNA or CISA-ADP&#34;</span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> path <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]:
</span></span><span style="display:flex;"><span>        rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(path))
</span></span><span style="display:flex;"><span>        print(rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>], json<span style="color:#f92672">.</span>dumps(pick(rec)))
</span></span></code></pre></div><p>We ran it on record files fetched from <code>raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/...</code> on November 13, 2026 (output trimmed to the key fields):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2026-1995   source=CISA-ADP format=cvssV3_1 stored=7.8
</span></span><span style="display:flex;"><span>CVE-2026-41080  source=cna format=cvssV3_1 stored=2.9 scenario=GENERAL
</span></span><span style="display:flex;"><span>CVE-2026-41129  source=cna format=cvssV4_0 stored=5.5 cvss_b=7.0 cvss_bt=5.5 stored_is=CVSS-BT
</span></span><span style="display:flex;"><span>CVE-2026-41225  source=cna format=cvssV4_0 stored=8.6 cvss_b=8.6 cvss_bt=8.6 stored_is=&#34;B and BT equal&#34; scenario=GENERAL
</span></span><span style="display:flex;"><span>CVE-2026-82598  source=cna format=cvssV4_0 stored=6.9 cvss_b=6.9 cvss_bt=5.5 stored_is=CVSS-B
</span></span><span style="display:flex;"><span>CVE-2026-86471  source=cna format=cvssV4_0 stored=9.3 cvss_b=9.3 cvss_bt=9.3 stored_is=&#34;B and BT equal&#34; scenario=GENERAL
</span></span></code></pre></div><p>CVE-2026-1995 (an IDrive Cloud Backup Client privilege escalation, CNA CERT/CC) has no CNA CVSS, so the CISA-ADP v3.1 score is used and labelled. CVE-2026-86471 (Apache CouchDB) sets <code>E:A</code>; Attacked is the worst-case Exploit Maturity, so Base and Base+Threat are equal.</p>
<p>This is a starting point, not a policy. You might prefer v3.1 over v4.0 while your SLAs are still written in v3.1 bands (the two versions&rsquo; scores are not comparable), keep the CISA-ADP score alongside the CNA one, or pick a named scenario from asset data.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Reading <code>metrics[0]</code>.</strong> The array order is the CNA&rsquo;s choice; some records list four CVSS versions.</li>
<li><strong>Detecting the version from <code>format</code>.</strong> Most entries just say <code>CVSS</code>. Use the key (<code>cvssV4_0</code>, <code>cvssV3_1</code>, &hellip;) or the <code>version</code> field.</li>
<li><strong>Treating v4.0 <code>baseScore</code> as CVSS-B.</strong> Recompute from <code>vectorString</code> if your policy cares.</li>
<li><strong>Dropping <code>other</code>.</strong> CISA-ADP&rsquo;s SSVC and KEV entries are <code>other</code>; so are vendor ratings such as Red Hat&rsquo;s.</li>
<li><strong>Overwriting CNA with ADP (or the reverse).</strong> Keep both with <code>providerMetadata.shortName</code>.</li>
<li><strong>Ignoring <code>scenarios</code>.</strong> The highest score may belong to a configuration you do not run.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>Counts are a snapshot of cvelistV5 at 00:00 UTC on November 13, 2026, restricted to records published in 2026. CNAs update records, so numbers will drift. Score recomputation used one library; where we compared stored and recomputed values, three entries matched neither, and we did not investigate them. We do not say which CNA practice is correct. The schema permits both, and the CVSS v4.0 specification&rsquo;s naming guidance applies to how scores are displayed, not to this JSON field.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>metrics optional in CNA container; array structure; anyOf of five keys; <code>format</code>, <code>scenarios</code> and GENERAL default; <code>other</code> = {type, content}</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE_Record_Format.json, CVEProject/cve-schema</a> (read Nov 13, 2026)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>v4.0 object required fields; no Threat/Environmental score fields; Supplemental field names</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/imports/cvss/cvss-v4.0.json">cvss-v4.0.json import, CVEProject/cve-schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS-B / BT / BE / BTE nomenclature</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document, FIRST</a></td>
          <td>Verified in our Oct 17, 2026 article; not re-read today</td>
      </tr>
      <tr>
          <td>All 2026 counts (formats, CISA-ADP fill, other types, scenarios, B vs BT split, CNA vs ADP disagreement, CNA concentration)</td>
          <td>Our script over <a href="https://github.com/CVEProject/cvelistV5/releases">cvelistV5 release cve_2026-11-13_0000Z</a></td>
          <td>Qualified: our computation, snapshot date stated</td>
      </tr>
      <tr>
          <td>Example records CVE-2026-1995, -41080, -41129, -41225, -82598, -86471 (vectors, stored scores, scenarios)</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/tree/main/cves/2026">cvelistV5 record files</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Recomputed CVSS-B / CVSS-BT values</td>
          <td>Red Hat <a href="https://pypi.org/project/cvss/"><code>cvss</code></a> library v3.6, run Nov 13, 2026</td>
          <td>Qualified: one implementation</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>vexctl: Create, Add, Merge and Attest OpenVEX Documents</title><link>https://firewallsync.com/posts/vexctl-create-add-merge-attest-openvex-documents/</link><guid>https://firewallsync.com/posts/vexctl-create-add-merge-attest-openvex-documents/</guid><pubDate>Thu, 12 Nov 2026 04:40:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>A source-checked guide to vexctl v0.4.4: the create, add, merge and attest commands, the validation rules they enforce, why merge keeps every statement instead of resolving them, and the defaults that quietly produce weak VEX documents.</description><enclosure url="https://firewallsync.com/posts/vexctl-create-add-merge-attest-openvex-documents/cover_hu12264946083305802858.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/vexctl-create-add-merge-attest-openvex-documents/cover_hu12264946083305802858.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/vexctl-create-add-merge-attest-openvex-documents/cover_hu12264946083305802858.jpg" alt="Tall stacks of paper documents and green and brown file folders with yellow-edged pages on an office desk under ceiling lights" width="1200" height="630"></p><p><code>vexctl</code> is the OpenVEX project&rsquo;s own tool for writing VEX (Vulnerability Exploitability eXchange) documents: <code>create</code> starts a document, <code>add</code> appends a statement and bumps the version, <code>merge</code> combines documents into one timeline, and <code>attest</code> wraps a document in an in-toto attestation you can sign and attach to an image. The commands are short. The defaults behind them are where weak VEX comes from: an anonymous author, a placeholder action statement, and a merge that resolves nothing.</p>
<p>This guide reads the <a href="https://github.com/openvex/vexctl/tree/v0.4.4">vexctl v0.4.4 source</a> (the newest tag on November 12, 2026, committed June 16, 2026) and the <a href="https://github.com/openvex/go-vex/tree/v0.2.8">go-vex v0.2.8</a> library it builds on. It fills the gap left by our <a href="/posts/trivy-vex-oci-openvex-attestations-cosign-attest/">Trivy <code>--vex oci</code> guide</a>: Trivy consumes VEX but does not write it. If you need the basics of what a statement asserts, start with <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">what a not_affected statement means</a>.</p>
<h2 id="install-and-version-check">Install and version check</h2>
<p>The <a href="https://github.com/openvex/vexctl/blob/v0.4.4/README.md">README</a> gives two install routes:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>go install github.com/openvex/vexctl@latest   <span style="color:#75715e"># or pin: @v0.4.4</span>
</span></span><span style="display:flex;"><span>brew install vexctl
</span></span></code></pre></div><p>v0.4.4&rsquo;s <code>go.mod</code> requires Go 1.26.4, <code>github.com/openvex/go-vex v0.2.8</code> and <code>github.com/sigstore/cosign/v2 v2.6.3</code>. Pin the version in CI: the behaviour below is read from that tag, and a few help texts in it are out of step with the code (covered under merge).</p>
<h2 id="vexctl-create-one-statement-new-document">vexctl create: one statement, new document</h2>
<p><code>create</code> takes positional arguments or flags (<code>create [flags] [product_id [vuln_id [status]]]</code>):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>vexctl create <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --author<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;Example Corp PSIRT &lt;psirt@example.com&gt;&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --product<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;pkg:oci/api-gateway@sha256:&lt;digest&gt;?repository_url=registry.example.com/team&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --vuln<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;CVE-2026-12345&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --status<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;not_affected&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --justification<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --impact-statement<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;The parser that contains the flaw is never loaded by this image.&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --file<span style="color:#f92672">=</span>api-gateway.vex.json
</span></span></code></pre></div><p><em>Illustrative: the command and flag names are taken from the v0.4.4 source; we did not run the binary for this article.</em></p>
<p>The flags that shape the statement, from <code>internal/cmd/options.go</code>:</p>
<table>
  <thead>
      <tr>
          <th>Flag</th>
          <th>Short</th>
          <th>Default</th>
          <th>Notes</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>--product</code></td>
          <td><code>-p</code></td>
          <td>none</td>
          <td>Repeatable. At least one is required. A purl or other IRI.</td>
      </tr>
      <tr>
          <td><code>--vuln</code></td>
          <td><code>-v</code></td>
          <td>none</td>
          <td>Required.</td>
      </tr>
      <tr>
          <td><code>--status</code></td>
          <td><code>-s</code></td>
          <td>none</td>
          <td><code>not_affected</code>, <code>affected</code>, <code>fixed</code> or <code>under_investigation</code>.</td>
      </tr>
      <tr>
          <td><code>--justification</code></td>
          <td><code>-j</code></td>
          <td>none</td>
          <td>Only allowed with <code>not_affected</code>. <code>vexctl list</code> prints the valid labels.</td>
      </tr>
      <tr>
          <td><code>--impact-statement</code></td>
          <td></td>
          <td>none</td>
          <td>Only allowed with <code>not_affected</code>.</td>
      </tr>
      <tr>
          <td><code>--action-statement</code></td>
          <td><code>-a</code></td>
          <td><code>No action statement provided</code></td>
          <td>Only kept when status is <code>affected</code>.</td>
      </tr>
      <tr>
          <td><code>--status-note</code></td>
          <td></td>
          <td>none</td>
          <td>How the status was determined.</td>
      </tr>
      <tr>
          <td><code>--aliases</code></td>
          <td></td>
          <td>none</td>
          <td>Other IDs (GHSA, GO-&hellip;). Duplicates, or an alias equal to <code>--vuln</code>, are rejected.</td>
      </tr>
      <tr>
          <td><code>--subcomponents</code></td>
          <td></td>
          <td>none</td>
          <td>Rejected when more than one <code>--product</code> is given.</td>
      </tr>
      <tr>
          <td><code>--author</code></td>
          <td></td>
          <td><code>Unknown Author</code></td>
          <td>Document author.</td>
      </tr>
      <tr>
          <td><code>--author-role</code></td>
          <td></td>
          <td>empty</td>
          <td>Optional.</td>
      </tr>
      <tr>
          <td><code>--id</code></td>
          <td></td>
          <td>generated</td>
          <td>Generated from the document&rsquo;s canonical hash when omitted.</td>
      </tr>
      <tr>
          <td><code>--file</code></td>
          <td></td>
          <td>stdout</td>
          <td>Output path.</td>
      </tr>
  </tbody>
</table>
<p>Two validation layers apply. vexctl&rsquo;s own checks (above) run first; then go-vex&rsquo;s statement validator enforces the <a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX spec</a> rules: a <code>not_affected</code> statement needs a <code>justification</code> or an <code>impact_statement</code>, and an <code>affected</code> statement needs an <code>action_statement</code>. The spec words them as MUSTs: &ldquo;For statements conveying a <code>not_affected</code> status, a VEX statement MUST include either a status justification or an impact_statement&rdquo;, and &ldquo;For a statement with &lsquo;affected&rsquo; status, a VEX statement MUST include a statement that SHOULD describe actions to remediate or mitigate the vulnerability.&rdquo;</p>
<h3 id="two-defaults-to-override-every-time">Two defaults to override every time</h3>
<p><strong>The author.</strong> go-vex sets <code>DefaultAuthor = &quot;Unknown Author&quot;</code>, and vexctl&rsquo;s validator only rejects a <em>blank</em> author, so the placeholder passes. The spec says <code>author</code> &ldquo;MUST be an individual or organization&rdquo;. A consumer that weighs VEX by issuer cannot trust an anonymous statement, so pass <code>--author</code> on <code>create</code>, <code>add</code> and <code>merge</code>.</p>
<p><strong>The action statement.</strong> The <code>-a</code> flag defaults to <code>No action statement provided</code>. For an <code>affected</code> statement that default fills the required field, so the validator is satisfied and the document carries no remediation advice. Write the real action (upgrade target, configuration change, compensating control) every time.</p>
<h3 id="reproducible-timestamps">Reproducible timestamps</h3>
<p><code>create</code> and <code>add</code> honour <code>SOURCE_DATE_EPOCH</code>; the help text says it may be &ldquo;formatted in UNIX time or RFC3339&rdquo;. Set it in release pipelines so a rebuilt document gets the same timestamp and, with the default ID, the same <code>@id</code>. go-vex serializes statement timestamps in UTC.</p>
<h2 id="vexctl-add-append-and-bump-the-version">vexctl add: append and bump the version</h2>
<p>The spec says a document&rsquo;s <code>version</code> &ldquo;must be incremented when any content within the VEX document changes&rdquo;. <code>add</code> does that for you:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>vexctl add --in-place api-gateway.vex.json <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  <span style="color:#e6db74">&#34;pkg:oci/api-gateway@sha256:&lt;digest&gt;?repository_url=registry.example.com/team&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  CVE-2026-23456 under_investigation
</span></span></code></pre></div><p><em>Illustrative.</em> Per <code>internal/cmd/add.go</code>, <code>add</code>:</p>
<ul>
<li>reads the document from the first positional argument or <code>-d/--document</code>, and writes to stdout unless you pass <code>-i/--in-place</code> or <code>--file</code>;</li>
<li>copies the document&rsquo;s <code>last_updated</code> (or <code>timestamp</code>) into any statement that has no timestamp of its own, so those statements keep their inherited date once the document changes;</li>
<li>refuses the new statement if any existing statement&rsquo;s <code>timestamp</code> or <code>last_updated</code> is later than the new time (&ldquo;date cannot be older than other statements&rdquo;);</li>
<li>appends the statement, sets the document&rsquo;s <code>last_updated</code> and increments <code>version</code>.</li>
</ul>
<p>The time check matters with <code>SOURCE_DATE_EPOCH</code>: an epoch older than the newest statement makes <code>add</code> fail rather than write an out-of-order history.</p>
<h2 id="vexctl-merge-a-timeline-not-a-verdict">vexctl merge: a timeline, not a verdict</h2>
<p><code>merge</code> exists for the case the README describes, &ldquo;when more than one stakeholder is issuing VEX metadata about a piece of software&rdquo;:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>vexctl merge --author<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;Example Corp PSIRT&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  --product<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;pkg:apk/wolfi/bash@1.0.0&#34;</span> <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  examples/openvex/document1.vex.json examples/openvex/document2.vex.json &gt; merged.vex.json
</span></span></code></pre></div><p>What it does, from <code>Merge()</code> in <code>pkg/ctl/implementation.go</code>:</p>
<ol>
<li><strong>Filters, optionally.</strong> With <code>--product</code> and/or <code>--vuln</code>, only matching statements are kept. Product matching is go-vex&rsquo;s, which compares purls rather than plain strings.</li>
<li><strong>Fills timestamps.</strong> A statement without a timestamp takes its source document&rsquo;s timestamp. If both are missing, merge fails: &ldquo;unable to cascade timestamp from doc to timeless statement&rdquo;.</li>
<li><strong>Keeps everything.</strong> Every matching statement is copied. There is no deduplication and no conflict resolution.</li>
<li><strong>Sorts.</strong> Statements are sorted by vulnerability name (string order), then by timestamp, oldest first.</li>
<li><strong>Names the result deterministically.</strong> Without <code>--id</code>, the new <code>@id</code> is <code>merged-vex-</code> plus the SHA-256 of the input documents&rsquo; <code>@id</code> values, sorted and joined with <code>:</code>. A document with no <code>@id</code> contributes <code>VEX-DOC-&lt;position&gt;</code>.</li>
<li><strong>Starts a fresh document.</strong> Version 1, a new timestamp, and <code>Unknown Author</code> unless you pass <code>--author</code>.</li>
</ol>
<p>Who wins? Not merge. go-vex&rsquo;s <code>EffectiveStatement()</code> returns the <em>last</em> matching statement after the same sort, so the newest statement for a product and vulnerability is the current status. That is how the README explains a merged history of <code>under_investigation</code> followed by <code>fixed</code>: tools &ldquo;replay&rdquo; the statements in order.</p>
<p>We ported steps 1 to 5 to Python (product matching simplified to exact strings) and ran the port on the two example documents in the vexctl repository:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 -I merge_port.py document2.vex.json document1.vex.json
</span></span><span style="display:flex;"><span>merged-vex-6b33f63783819578804350ba1cde277ddd8f90aa2bbcb706d19a61d211a8443f
</span></span><span style="display:flex;"><span>2023-12-05T05:04:34.77929922Z CVE-1234-5678 under_investigation
</span></span><span style="display:flex;"><span>2023-12-05T05:06:38.099731287Z CVE-1234-5678 fixed
</span></span><span style="display:flex;"><span>effective: fixed
</span></span></code></pre></div><p>Passing the files in the other order produced the same <code>@id</code>, because the IDs are sorted before hashing. That is a property of the port matching the source; treat it as a reading aid, not a test of the vexctl binary.</p>
<h3 id="merge-pitfalls">Merge pitfalls</h3>
<ul>
<li><strong>Clock skew decides conflicts.</strong> Two issuers stating different statuses for the same product rely on their timestamps alone. A supplier&rsquo;s <code>not_affected</code> with a later clock beats your own <code>affected</code>. Merge per issuer, or review conflicts before publishing.</li>
<li><strong>Output is stdout only.</strong> <code>merge</code> registers no <code>--file</code> flag in v0.4.4; redirect the output.</li>
<li><strong>The flag is <code>--vuln</code>.</strong> The command&rsquo;s own help example shows <code>--vulnerability=CVE-2022-3294</code>, but the flag registered in v0.4.4 is <code>--vuln</code> (constant <code>vulnLongFlag = &quot;vuln&quot;</code>). Use <code>--vuln</code>.</li>
<li><strong>Merged documents grow.</strong> History is kept, so a long-lived merged file accumulates every status change. That is by design; prune by publishing a new document rather than editing statements in place.</li>
</ul>
<h2 id="vexctl-attest-wrap-sign-attach">vexctl attest: wrap, sign, attach</h2>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>vexctl attest api-gateway.vex.json                      <span style="color:#75715e"># unsigned in-toto statement to stdout</span>
</span></span><span style="display:flex;"><span>vexctl attest --sign api-gateway.vex.json               <span style="color:#75715e"># sign with the sigstore flow</span>
</span></span><span style="display:flex;"><span>vexctl attest --attach api-gateway.vex.json             <span style="color:#75715e"># sign and attach to each image subject</span>
</span></span><span style="display:flex;"><span>vexctl attest --attach api-gateway.vex.json registry.example.com/team/api-gateway@sha256:&lt;digest&gt;
</span></span></code></pre></div><p><em>Illustrative.</em> From <code>internal/cmd/attest.go</code> and <code>pkg/ctl/ctl.go</code>:</p>
<ul>
<li>The predicate is the OpenVEX document; go-vex sets the predicate type to <code>https://openvex.dev/ns/v0.2.0</code>.</li>
<li>Without positional image arguments, subjects come from the document&rsquo;s products. <code>pkg:oci</code> purls become image references (the help text&rsquo;s example: <code>pkg:oci/kube-apiserver?repository_url=registry.k8s.io&amp;tag=v1.26.0</code> becomes <code>registry.k8s.io/kube-apiserver:v1.26.0</code>), and references without a digest &ldquo;will trigger a network lookup to read the image digest from the registry&rdquo;.</li>
<li>Products that are neither oci purls nor image references are kept only if they carry <code>sha256</code> or <code>sha512</code> hashes. Otherwise they are skipped with a warning, or rejected with an error if you named them explicitly.</li>
<li><code>--attach</code> (<code>-a</code>) &ldquo;always implies &ndash;sign as sigstore does not support attaching unsigned attestations&rdquo;. <code>--refs</code> (<code>-r</code>) names the images to attach to when automatic attachment fails.</li>
</ul>
<p>If your scanner reads the attestation from the registry, check the predicate type it accepts. Our <a href="/posts/trivy-vex-oci-openvex-attestations-cosign-attest/">Trivy <code>--vex oci</code> guide</a> covers how Trivy matches OpenVEX predicate types and why you should verify the signature yourself before trusting what it finds.</p>
<h2 id="where-vexctl-fits-in-a-patch-decision-flow">Where vexctl fits in a patch-decision flow</h2>
<p>VEX is the last filter before a finding becomes work: a <code>not_affected</code> statement with a sound justification closes a scanner finding that KEV, EPSS or CVSS would otherwise push up the queue. Our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a> shows where that step sits between the CVE record and the ticket. Product IDs are where most VEX fails in practice, so check the matching rules in our <a href="/posts/openvex-products-subcomponents-purl-matching-trivy/">OpenVEX products and subcomponents guide</a> before you publish, and see the <a href="/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/">CycloneDX VEX mapping</a> if your consumers expect CycloneDX.</p>
<p>Two more commands exist and are out of scope here: <code>vexctl filter</code> applies VEX to SARIF results (the README says &ldquo;We support results files in SARIF for now&rdquo;), and <code>vexctl generate</code> builds documents from &ldquo;golden&rdquo; template files (default directory <code>.openvex/templates</code>).</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>We read the v0.4.4 source and README; we did not run the vexctl binary. Commands are marked illustrative. The merge output comes from a Python port of the merge and sort logic, not from vexctl.</li>
<li>Behaviour may change after v0.4.4. Re-check flags with <code>vexctl &lt;command&gt; --help</code> against the version you install.</li>
<li>vexctl writes OpenVEX only. CSAF and CycloneDX VEX producers follow their own formats.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>v0.4.4 is the newest tag; dependency versions</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/go.mod">vexctl v0.4.4 go.mod</a>, git tags</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Flags, defaults and validation rules</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/internal/cmd/options.go">internal/cmd/options.go</a>, <a href="https://github.com/openvex/go-vex/blob/v0.2.8/pkg/vex/vex.go">go-vex vex.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>add version bump and time checks</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/internal/cmd/add.go">internal/cmd/add.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Merge filters, keeps, sorts and hashes IDs</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/pkg/ctl/implementation.go">pkg/ctl/implementation.go</a>, <a href="https://github.com/openvex/go-vex/blob/v0.2.8/pkg/vex/statement.go">go-vex statement.go</a></td>
          <td>Verified (source); example run via Python port</td>
      </tr>
      <tr>
          <td>Latest statement is the effective one</td>
          <td><a href="https://github.com/openvex/go-vex/blob/v0.2.8/pkg/vex/vex.go">go-vex vex.go EffectiveStatement</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Attest predicate type and subject rules</td>
          <td><a href="https://github.com/openvex/vexctl/blob/v0.4.4/internal/cmd/attest.go">internal/cmd/attest.go</a>, <a href="https://github.com/openvex/vexctl/blob/v0.4.4/pkg/ctl/ctl.go">pkg/ctl/ctl.go</a>, <a href="https://github.com/openvex/go-vex/tree/v0.2.8/pkg/attestation">go-vex attestation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Spec MUSTs for author, version, justification, action statement</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>KEV knownRansomwareCampaignUse: 114 Unknown-to-Known Flips</title><link>https://firewallsync.com/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/</link><guid>https://firewallsync.com/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/</guid><pubDate>Tue, 10 Nov 2026 04:50:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>CISA's KEV catalog marks each entry's ransomware use as Known or Unknown, and it changes that field without an announcement. We replayed every kev-data commit since January 2025: 114 entries flipped from Unknown to Known, none flipped back, and 25 of the 27 Known entries added in 2026 started as Unknown. Here is how often it happens, how long it takes, and a tested script to catch the next flip.</description><enclosure url="https://firewallsync.com/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/cover_hu14181278352977741825.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/cover_hu14181278352977741825.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/kev-knownransomwarecampaignuse-unknown-to-known-flips/cover_hu14181278352977741825.jpg" alt="Beside a large yellow circular panel, a grey industrial control panel with a round ALARM indicator, a RESET button, a key switch and a column of small indicator lamps, one of them lit red" width="1200" height="630"></p><p>CISA&rsquo;s Known Exploited Vulnerabilities (KEV) catalog tags every entry with <code>knownRansomwareCampaignUse</code>: <code>Known</code> or <code>Unknown</code>. That value is not fixed at listing. Replaying the public <a href="https://github.com/cisagov/kev-data">kev-data</a> Git history from January 27, 2025 to November 10, 2026, we found <strong>114 entries that changed from <code>Unknown</code> to <code>Known</code></strong>, and none that went the other way. Of the 27 entries added in 2026 that are now marked <code>Known</code>, <strong>25 were listed as <code>Unknown</code> first</strong>.</p>
<p>A flip leaves <code>dateAdded</code> and <code>dueDate</code> untouched, so a pipeline that only watches for new CVE IDs will never notice. If your process gives ransomware-linked KEV entries a shorter SLA, you need to diff this field between catalog versions. This article shows how often it changes, how long it takes, and a tested script that catches it. For where KEV sits in the wider triage flow, see our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow from CVE record to ticket</a>.</p>
<h2 id="what-the-field-means">What the field means</h2>
<p>The KEV JSON schema in kev-data defines the field like this:</p>
<blockquote>
<p>&ldquo;&lsquo;Known&rsquo; if this vulnerability is known to have been leveraged as part of a ransomware campaign; &lsquo;Unknown&rsquo; if CISA lacks confirmation that the vulnerability has been utilized for ransomware&rdquo;</p>
</blockquote>
<p>Two consequences follow:</p>
<ul>
<li><strong><code>Unknown</code> is not &ldquo;no&rdquo;.</strong> It means CISA has not confirmed ransomware use. Don&rsquo;t treat it as a negative signal or a reason to slow down.</li>
<li><strong><code>Known</code> is a one-way upgrade in practice.</strong> In the window we measured, no entry went from <code>Known</code> back to <code>Unknown</code>.</li>
</ul>
<p>The field sits alongside <code>dateAdded</code>, <code>dueDate</code>, <code>requiredAction</code> and the rest. Our <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">field guide to the KEV JSON feed</a> covers each of them.</p>
<h2 id="how-we-measured">How we measured</h2>
<ul>
<li><strong>Data:</strong> every commit to <code>known_exploited_vulnerabilities.json</code> in <a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a>, from the first commit (2025-01-27) to the commit for catalog version 2026.11.09 (pushed 2026-11-10 03:13 UTC). That&rsquo;s 341 commits. One commit, from 2025-02-25, did not parse as JSON and was skipped; any change in it shows up in the next commit.</li>
<li><strong>Flip:</strong> a <code>cveID</code> whose <code>knownRansomwareCampaignUse</code> value differs from its value in the previous parsed commit.</li>
<li><strong>Flip date:</strong> the Git commit timestamp. It records when the change reached the repository, not when CISA&rsquo;s analysts decided it, so treat it as an upper bound with day-level precision.</li>
<li><strong>Lag:</strong> flip date minus the entry&rsquo;s <code>dateAdded</code>, in whole days.</li>
</ul>
<p>The repository only starts in January 2025, so flips before that date are invisible here. The 1,252 entries in the first commit form the baseline: 245 were already <code>Known</code>.</p>
<h2 id="how-often-the-flag-flips">How often the flag flips</h2>
<table>
  <thead>
      <tr>
          <th>Period</th>
          <th>Unknown → Known</th>
          <th>Known → Unknown</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2025 (from Jan 27)</td>
          <td>59</td>
          <td>0</td>
      </tr>
      <tr>
          <td>2026 (to Nov 10)</td>
          <td>55</td>
          <td>0</td>
      </tr>
      <tr>
          <td><strong>Total</strong></td>
          <td><strong>114</strong></td>
          <td><strong>0</strong></td>
      </tr>
  </tbody>
</table>
<p>The 2025 total matches the 59 flips GreyNoise&rsquo;s Glenn Thorpe counted for 2025, as <a href="https://www.theregister.com/2026/02/03/greynoise_cisa_ransomware_gripe/">reported by The Register</a> in February 2026. Thorpe described the change as &ldquo;no alert, no announcement&rdquo; and &ldquo;Just a field change in a JSON file&rdquo;. Our count comes from an independent replay of the same public history.</p>
<p>Flips come in batches rather than a steady trickle. The largest single days were 2025-05-12 (14 flips) and 2026-08-14 (10). The most recent flip in the data was CVE-2026-63077 (JetBrains TeamCity) on 2026-09-23, 48 days before this article. A quiet spell is normal and does not mean the field has stopped changing.</p>
<p>By vendor, Microsoft accounts for 33 of the 114 flips, followed by Ivanti (6), Fortinet (5), SonicWall (5) and Oracle (5).</p>
<h2 id="new-entries-almost-always-start-as-unknown">New entries almost always start as Unknown</h2>
<p>Of the 524 entries that first appeared in the catalog during the window, <strong>521 were listed as <code>Unknown</code></strong> and only 3 as <code>Known</code>. Ransomware attribution usually arrives after exploitation evidence, so the first version of an entry rarely carries it.</p>
<p>For the 50 entries that were both added and flipped inside the window:</p>
<table>
  <thead>
      <tr>
          <th>Lag from dateAdded to flip</th>
          <th>Entries</th>
          <th>Cumulative</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>7 days or less</td>
          <td>17</td>
          <td>17</td>
      </tr>
      <tr>
          <td>8–30 days</td>
          <td>12</td>
          <td>29</td>
      </tr>
      <tr>
          <td>31–90 days</td>
          <td>10</td>
          <td>39</td>
      </tr>
      <tr>
          <td>More than 90 days</td>
          <td>11</td>
          <td>50</td>
      </tr>
  </tbody>
</table>
<p>The median lag is 20 days and the longest is 554 days. One entry, CVE-2025-61882 (Oracle E-Business Suite), shows a lag of −1: it appeared in the repository on 2025-10-06 with a <code>dateAdded</code> of 2025-10-07 and flipped the same day.</p>
<p>Seen from the other side: of the 289 entries in the current catalog with a 2026 <code>dateAdded</code>, 27 are <code>Known</code>, and 25 of those got there through a flip. If you only read the field once, at listing, you would have missed almost all of this year&rsquo;s ransomware signal.</p>
<h2 id="old-entries-flip-too">Old entries flip too</h2>
<p>64 of the 114 flips were on entries that were already in the catalog when the repository started. Across all 114, the median lag is <strong>352 days</strong>, and the longest are CVE-2016-7255 and CVE-2019-0859 (both Microsoft Win32k, added 2021-11-03), which flipped on 2026-09-09, 1,771 days later.</p>
<p>For triage, this means an entry you closed or risk-accepted years ago can pick up a ransomware link today. That&rsquo;s a reason to reopen it, especially on systems that never got the fix. The same logic applies to KEV backfills of old CVEs, covered in our <a href="/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/">KEV time-to-listing analysis</a>.</p>
<h2 id="detect-flips-between-two-catalog-versions">Detect flips between two catalog versions</h2>
<p>The script below compares two KEV JSON snapshots and prints every entry whose <code>knownRansomwareCampaignUse</code> changed. It uses only the Python standard library. We ran it with Python 3 on the kev-data snapshots for catalog versions 2026.08.11 and 2026.08.14. The output (abbreviated) matches the 2026-08-14 batch in the full replay:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Report KEV entries whose knownRansomwareCampaignUse changed between two snapshots.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 kev_ransomware_flips.py OLD.json NEW.json
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load</span>(path):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> open(path) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>        doc <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(f)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> doc[<span style="color:#e6db74">&#34;catalogVersion&#34;</span>], {v[<span style="color:#e6db74">&#34;cveID&#34;</span>]: v <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> doc[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>old_ver, old <span style="color:#f92672">=</span> load(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>])
</span></span><span style="display:flex;"><span>new_ver, new <span style="color:#f92672">=</span> load(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;# </span><span style="color:#e6db74">{</span>old_ver<span style="color:#e6db74">}</span><span style="color:#e6db74"> -&gt; </span><span style="color:#e6db74">{</span>new_ver<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> cve, v <span style="color:#f92672">in</span> sorted(new<span style="color:#f92672">.</span>items()):
</span></span><span style="display:flex;"><span>    before <span style="color:#f92672">=</span> old<span style="color:#f92672">.</span>get(cve, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;knownRansomwareCampaignUse&#34;</span>)
</span></span><span style="display:flex;"><span>    after <span style="color:#f92672">=</span> v[<span style="color:#e6db74">&#34;knownRansomwareCampaignUse&#34;</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> before <span style="color:#f92672">is</span> <span style="color:#f92672">not</span> <span style="color:#66d9ef">None</span> <span style="color:#f92672">and</span> before <span style="color:#f92672">!=</span> after:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>before<span style="color:#e6db74">}</span><span style="color:#e6db74"> -&gt; </span><span style="color:#e6db74">{</span>after<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">added </span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;dateAdded&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>              <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;vendorProject&#39;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;product&#39;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span># 2026.08.11 -&gt; 2026.08.14
</span></span><span style="display:flex;"><span>CVE-2016-0189	Unknown -&gt; Known	added 2022-03-28	Microsoft Internet Explorer
</span></span><span style="display:flex;"><span>CVE-2018-0802	Unknown -&gt; Known	added 2021-11-03	Microsoft Office
</span></span><span style="display:flex;"><span>CVE-2019-5591	Unknown -&gt; Known	added 2021-11-03	Fortinet FortiOS
</span></span><span style="display:flex;"><span>CVE-2021-4034	Unknown -&gt; Known	added 2022-06-27	Red Hat Polkit
</span></span><span style="display:flex;"><span>CVE-2025-60710	Unknown -&gt; Known	added 2026-04-13	Microsoft Windows
</span></span><span style="display:flex;"><span>...
</span></span></code></pre></div><p>To run it in production, keep the last catalog you processed, and when a new <code>catalogVersion</code> arrives, run the diff before you overwrite the old file. To backfill history, clone kev-data and feed consecutive commits to the same function (<code>git log --reverse --format=%H -- known_exploited_vulnerabilities.json</code> lists them; <code>git show &lt;commit&gt;:known_exploited_vulnerabilities.json</code> extracts each one).</p>
<p>The script skips entries that are new in the second snapshot, because their listing value is not a change. Handle new entries in your normal KEV ingest.</p>
<h2 id="what-to-do-with-a-flip">What to do with a flip</h2>
<ol>
<li><strong>Treat Unknown → Known as a new event.</strong> Open or re-prioritize tickets for every asset that still carries the CVE, even if the KEV due date has already passed or the finding was risk-accepted.</li>
<li><strong>Don&rsquo;t read Unknown as low risk.</strong> Of this year&rsquo;s <code>Known</code> entries, 25 of 27 started as <code>Unknown</code>. A ransomware-only fast lane that ignores everything else will lag the data by weeks.</li>
<li><strong>Record the flip date yourself.</strong> The catalog keeps no history of the field, so your ticket or data store is the only place that will remember when the link appeared.</li>
<li><strong>Combine it with other signals, don&rsquo;t substitute it for them.</strong> EPSS scores for newly listed KEV CVEs often start low (see our <a href="/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/">analysis of EPSS for KEV entries</a>), so a flip is a useful second trigger rather than the only one.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>The data covers January 27, 2025 to November 10, 2026. Earlier flips are not visible in the kev-data repository.</li>
<li>Flip dates are Git commit times, not CISA decision times.</li>
<li>CISA does not publish the evidence behind a <code>Known</code> value, so this article measures when the label changed, not why.</li>
<li>One entry that flipped, CVE-2025-6264 (Rapid7 Velociraptor), has since been removed from the catalog. It is counted in the 114 but not in the 361 current <code>Known</code> entries.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Definition of Known and Unknown</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/develop/known_exploited_vulnerabilities_schema.json">KEV JSON schema, cisagov/kev-data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>114 Unknown → Known flips, 0 reversals; lag and vendor figures</td>
          <td>Replay of <a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a> commit history, 2025-01-27 to 2026-11-10</td>
          <td>Verified (our computation)</td>
      </tr>
      <tr>
          <td>361 of 1,773 entries Known (20.4%)</td>
          <td><a href="https://github.com/cisagov/kev-data">KEV JSON</a>, catalogVersion 2026.11.09</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>GreyNoise counted 59 flips in 2025; &ldquo;no alert, no announcement&rdquo;</td>
          <td><a href="https://www.theregister.com/2026/02/03/greynoise_cisa_ransomware_gripe/">The Register, Feb 3, 2026</a></td>
          <td>Qualified (press report of third-party research; matches our count)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>KEV Time to Listing: CVE datePublished to dateAdded</title><link>https://firewallsync.com/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/</link><guid>https://firewallsync.com/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/</guid><pubDate>Mon, 09 Nov 2026 04:55:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>How long after a CVE record is published does CISA add it to KEV? We joined all 1,769 KEV entries to their CVE records' datePublished. For 2026 additions the median gap is 14 days, but the catalog mixes same-week listings with backfills of decade-old CVEs. Numbers by year and vendor, a tested script, and what it means for SLA clocks.</description><enclosure url="https://firewallsync.com/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/cover_hu7963031346552311960.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/cover_hu7963031346552311960.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/kev-time-to-listing-cve-published-to-dateadded-days-by-vendor/cover_hu7963031346552311960.jpg" alt="A pale blue-grey multiple-exposure image of several pocket-watch and stopwatch dials overlapping, their hands pointing in different directions" width="1200" height="630"></p><p>For CVEs that CISA added to its Known Exploited Vulnerabilities (KEV) catalog in 2026, the median gap between the CVE record&rsquo;s publication and the KEV <code>dateAdded</code> is <strong>14 days</strong>. For 2026 CVE IDs it is <strong>4 days</strong>, and 28% of those were listed on or before the day their record went public. But the same catalog also backfills old bugs: 18% of this year&rsquo;s additions were published more than a year before they were listed.</p>
<p>That split is the useful part for anyone who sets patch SLAs. A KEV due date starts its clock at listing; your exposure clock started at publication. This article measures the gap between the two across the whole catalog, by year and by vendor, and gives you a script to rerun the numbers yourself. For where KEV sits in the full triage flow, see our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow from CVE record to ticket</a>.</p>
<h2 id="what-we-measured">What we measured</h2>
<ul>
<li><strong>KEV side:</strong> <code>dateAdded</code> for every entry in the <a href="https://github.com/cisagov/kev-data">KEV JSON</a> (catalog version 2026.11.06, released 2026-11-06T19:44Z, 1,769 entries). Our <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">field guide to the KEV JSON feed</a> covers the other fields.</li>
<li><strong>CVE side:</strong> <code>cveMetadata.datePublished</code> from each entry&rsquo;s record in the <a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> bulk release tagged <code>cve_2026-11-09_0000Z</code>. All 1,769 KEV CVEs had a record, all in the PUBLISHED state.</li>
<li><strong>Gap:</strong> <code>dateAdded − date(datePublished)</code> in whole days. Negative means KEV listed the CVE before the record was published.</li>
</ul>
<p>The <a href="https://github.com/CVEProject/cve-schema">CVE Record Format schema</a> defines <code>datePublished</code> as &ldquo;The date/time the CVE Record was first published in the CVE List.&rdquo; That is not the vendor&rsquo;s advisory date or the first public report, which can be earlier, so treat the gap as &ldquo;time since the CVE record existed&rdquo;, not &ldquo;time since disclosure&rdquo;. KEV&rsquo;s <code>dateAdded</code> is a date with no time zone, so a gap of 0 or ±1 day is within rounding.</p>
<p>CISA lists three <a href="https://www.cisa.gov/known-exploited-vulnerabilities">criteria for adding an entry</a>: the vulnerability &ldquo;has an assigned Common Vulnerabilities and Exposures (CVE) ID&rdquo;, there is &ldquo;reliable evidence that the vulnerability has been actively exploited in the wild&rdquo;, and there is &ldquo;a clear remediation action for the vulnerability, such as a vendor-provided update.&rdquo; The gap therefore reflects when exploitation evidence reached CISA, not processing speed.</p>
<h2 id="the-gap-by-year">The gap by year</h2>
<p>The catalog launched on 2021-11-03 with 287 entries at once, which is why 2021 and 2022 look so different from later years: the early catalog was mostly a backlog of known-exploited older CVEs.</p>
<table>
  <thead>
      <tr>
          <th>Year added</th>
          <th>Entries</th>
          <th>Median gap (days)</th>
          <th>Listed on/before publication day</th>
          <th>Within 7 days</th>
          <th>Within 30 days</th>
          <th>More than 365 days</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2021</td>
          <td>311</td>
          <td>446</td>
          <td>2%</td>
          <td>3%</td>
          <td>6%</td>
          <td>56%</td>
      </tr>
      <tr>
          <td>2022</td>
          <td>555</td>
          <td>1,436</td>
          <td>7%</td>
          <td>9%</td>
          <td>12%</td>
          <td>75%</td>
      </tr>
      <tr>
          <td>2023</td>
          <td>187</td>
          <td>12</td>
          <td>24%</td>
          <td>45%</td>
          <td>55%</td>
          <td>25%</td>
      </tr>
      <tr>
          <td>2024</td>
          <td>186</td>
          <td>21.5</td>
          <td>21%</td>
          <td>42%</td>
          <td>51%</td>
          <td>24%</td>
      </tr>
      <tr>
          <td>2025</td>
          <td>245</td>
          <td>26</td>
          <td>20%</td>
          <td>40%</td>
          <td>51%</td>
          <td>23%</td>
      </tr>
      <tr>
          <td>2026 (to Nov 6)</td>
          <td>285</td>
          <td>14</td>
          <td>19%</td>
          <td>43%</td>
          <td>59%</td>
          <td>18%</td>
      </tr>
  </tbody>
</table>
<p>Since 2023 the shape has been stable: roughly a fifth of entries are listed by the day the record is published, a little over two fifths within a week, and about a quarter or less are CVEs more than a year old.</p>
<h3 id="fresh-cves-versus-backfill">Fresh CVEs versus backfill</h3>
<p>Splitting 2026 additions by CVE ID year separates the two populations. The 197 entries with 2026 IDs had a median gap of <strong>4 days</strong>: 28% were listed on or before publication day, 60% within 7 days and 82% within 30 days. The rest are mostly backfill. Of all 285 entries added in 2026, 51 were published more than 365 days before listing; the oldest is CVE-2008-4128 (Cisco), listed about 6,500 days after its record was published.</p>
<p>The same pattern holds in earlier years. For same-year CVE IDs, the median gap was 2 days in 2023, 3 in 2024 and 3 in 2025.</p>
<h2 id="the-gap-by-vendor">The gap by vendor</h2>
<p>For entries added in 2025 and 2026, vendors with at least eight additions:</p>
<table>
  <thead>
      <tr>
          <th>Vendor (vendorProject)</th>
          <th>Entries</th>
          <th>Median gap (days)</th>
          <th>On/before publication day</th>
          <th>Within 7 days</th>
          <th>More than 365 days</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Microsoft</td>
          <td>79</td>
          <td>1</td>
          <td>49%</td>
          <td>54%</td>
          <td>24%</td>
      </tr>
      <tr>
          <td>Cisco</td>
          <td>28</td>
          <td>3</td>
          <td>46%</td>
          <td>57%</td>
          <td>11%</td>
      </tr>
      <tr>
          <td>Apple</td>
          <td>19</td>
          <td>2</td>
          <td>16%</td>
          <td>53%</td>
          <td>21%</td>
      </tr>
      <tr>
          <td>Fortinet</td>
          <td>17</td>
          <td>7</td>
          <td>29%</td>
          <td>53%</td>
          <td>6%</td>
      </tr>
      <tr>
          <td>Google</td>
          <td>16</td>
          <td>1</td>
          <td>25%</td>
          <td>100%</td>
          <td>0%</td>
      </tr>
      <tr>
          <td>Linux</td>
          <td>16</td>
          <td>132</td>
          <td>0%</td>
          <td>0%</td>
          <td>38%</td>
      </tr>
      <tr>
          <td>Ivanti</td>
          <td>15</td>
          <td>6</td>
          <td>33%</td>
          <td>60%</td>
          <td>0%</td>
      </tr>
      <tr>
          <td>Citrix</td>
          <td>11</td>
          <td>7</td>
          <td>36%</td>
          <td>55%</td>
          <td>0%</td>
      </tr>
      <tr>
          <td>Adobe</td>
          <td>10</td>
          <td>58</td>
          <td>0%</td>
          <td>30%</td>
          <td>30%</td>
      </tr>
      <tr>
          <td>SonicWall</td>
          <td>10</td>
          <td>1</td>
          <td>40%</td>
          <td>70%</td>
          <td>20%</td>
      </tr>
      <tr>
          <td>Oracle</td>
          <td>9</td>
          <td>48</td>
          <td>0%</td>
          <td>22%</td>
          <td>33%</td>
      </tr>
      <tr>
          <td>Synacor</td>
          <td>9</td>
          <td>280</td>
          <td>0%</td>
          <td>0%</td>
          <td>33%</td>
      </tr>
  </tbody>
</table>
<p>Two readings, both cautious because the samples are small:</p>
<ul>
<li><strong>Vendors that publish exploited-in-the-wild status with the fix</strong> (Microsoft, Cisco, Google, SonicWall) show median gaps of one to three days. For Microsoft, almost half the entries were listed on or before the day the CVE record appeared. If your SLA for these products waits for KEV, it waits very little, but it still waits.</li>
<li><strong>Linux kernel entries</strong> behave differently: a median of 132 days and none within a week. Kernel CVE records are published in volume by the kernel CNA, and exploitation evidence for a specific one usually arrives much later. Our guide to <a href="/posts/linux-kernel-cve-records-affected-versions-stable-branches/">Linux kernel CVE records and stable-branch ranges</a> covers how to tell whether you are affected once it does.</li>
</ul>
<p>Microsoft&rsquo;s 24% over a year reflects backfill of old Windows and Office CVEs (several from 2008-2009 were added in 2026), not slow listing of new ones.</p>
<h2 id="kev-can-come-before-the-cve-record">KEV can come before the CVE record</h2>
<p>62 entries have a negative gap: KEV listed them before their CVE record was published. Four were added in 2026, including CVE-2026-118527 (Ivanti Connect Secure), added on 2026-10-30 with a record published on 2026-11-03, and CVE-2026-87886, added on 2026-09-16 one day before its record.</p>
<p>Some older negative gaps are artefacts: 210 KEV CVE records carry a <code>datePublished</code> with a midnight timestamp, typical of records migrated from the legacy CVE format, and a few of those dates are clearly later than the CVE&rsquo;s real disclosure. Treat large negative gaps on pre-2023 CVEs as data quality, not as evidence of early listing.</p>
<p>The engineering consequence is the same either way: <strong>join KEV to your inventory by <code>cveID</code> alone.</strong> A pipeline that only checks KEV membership for CVEs it has already ingested from the CVE List or NVD will skip the entries CISA lists first, which are exactly the most urgent ones.</p>
<h2 id="what-this-means-for-sla-clocks">What this means for SLA clocks</h2>
<p><strong>Two clocks, both worth storing.</strong> The KEV <code>dueDate</code> is set from <code>dateAdded</code>. For 2026 additions it was 3 days after listing for 151 entries, 14 days for 87 and 21 days for 44; our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04 explainer</a> covers where those tiers come from. Measured from <code>datePublished</code> instead, the median 2026 entry was due 22 days after its record appeared. If you only report &ldquo;days since KEV listing&rdquo;, your metrics will look better than your actual exposure window.</p>
<p><strong>Do not use KEV as the start signal for internet-facing products.</strong> For fresh CVEs in edge and enterprise products the median wait for a listing is a few days, and attackers were by definition already active when CISA got its evidence. KEV confirms urgency; it should not be the first thing that creates it. EPSS has the same lag problem for new KEV entries, as our <a href="/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/">EPSS-for-KEV audit</a> showed.</p>
<p><strong>Treat backfill as a reopen trigger.</strong> Nearly one in five 2026 additions was a CVE more than a year old. If your team accepted risk on an old CVE (&ldquo;internal only&rdquo;, &ldquo;no known exploit&rdquo;), a KEV listing invalidates the second half of that reasoning. Alert on any new KEV <code>cveID</code> that matches an open risk acceptance or a suppressed finding, not only on new scanner findings.</p>
<p><strong>Watch for the CISA-ADP update too.</strong> In our snapshot, 1,767 of the 1,769 KEV CVE records also carried a <code>kev</code> metric in a CISA-ADP container, so the listing shows up in the CVE record itself; our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC guide</a> explains the <code>Exploitation</code> value, and the <a href="/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/">cvelistV5 delta sync guide</a> shows how to pick up those updates hourly.</p>
<h2 id="a-script-to-rerun-the-numbers">A script to rerun the numbers</h2>
<p>This Python 3 script (standard library only) pulls the KEV JSON, fetches each recent entry&rsquo;s CVE record from cvelistV5 on GitHub, and prints the gap. We ran it on 2026-11-09 with the default 90-day window:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Days from CVE record publication to CISA KEV listing, for recent KEV adds.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 kev_lag.py [days_back]   (default 90)&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> date, timedelta
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> statistics <span style="color:#f92672">import</span> median
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>KEV <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/cisagov/kev-data/main/known_exploited_vulnerabilities.json&#34;</span>
</span></span><span style="display:flex;"><span>CVE <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/</span><span style="color:#e6db74">{y}</span><span style="color:#e6db74">/</span><span style="color:#e6db74">{b}</span><span style="color:#e6db74">xxx/</span><span style="color:#e6db74">{id}</span><span style="color:#e6db74">.json&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">get</span>(url):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(url, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">30</span>) <span style="color:#66d9ef">as</span> r:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> json<span style="color:#f92672">.</span>load(r)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">record_url</span>(cve_id):
</span></span><span style="display:flex;"><span>    _, year, num <span style="color:#f92672">=</span> cve_id<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;-&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> CVE<span style="color:#f92672">.</span>format(y<span style="color:#f92672">=</span>year, b<span style="color:#f92672">=</span>num[:<span style="color:#f92672">-</span><span style="color:#ae81ff">3</span>] <span style="color:#66d9ef">if</span> len(num) <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">3</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;0&#34;</span>, id<span style="color:#f92672">=</span>cve_id)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>days_back <span style="color:#f92672">=</span> int(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]) <span style="color:#66d9ef">if</span> len(sys<span style="color:#f92672">.</span>argv) <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">1</span> <span style="color:#66d9ef">else</span> <span style="color:#ae81ff">90</span>
</span></span><span style="display:flex;"><span>cutoff <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>today() <span style="color:#f92672">-</span> timedelta(days<span style="color:#f92672">=</span>days_back)
</span></span><span style="display:flex;"><span>kev <span style="color:#f92672">=</span> get(KEV)
</span></span><span style="display:flex;"><span>rows <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> kev[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    added <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>fromisoformat(v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> added <span style="color:#f92672">&lt;</span> cutoff:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>        meta <span style="color:#f92672">=</span> get(record_url(v[<span style="color:#e6db74">&#34;cveID&#34;</span>]))[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>]
</span></span><span style="display:flex;"><span>        published <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>fromisoformat(meta[<span style="color:#e6db74">&#34;datePublished&#34;</span>][:<span style="color:#ae81ff">10</span>])
</span></span><span style="display:flex;"><span>        lag <span style="color:#f92672">=</span> (added <span style="color:#f92672">-</span> published)<span style="color:#f92672">.</span>days
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">except</span> <span style="color:#a6e22e">Exception</span>:           <span style="color:#75715e"># no CVE record in cvelistV5 yet (404)</span>
</span></span><span style="display:flex;"><span>        published, lag <span style="color:#f92672">=</span> <span style="color:#66d9ef">None</span>, <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>    rows<span style="color:#f92672">.</span>append((v[<span style="color:#e6db74">&#34;cveID&#34;</span>], v[<span style="color:#e6db74">&#34;vendorProject&#34;</span>], str(published), v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>], lag))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> sorted(rows, key<span style="color:#f92672">=</span><span style="color:#66d9ef">lambda</span> r: r[<span style="color:#ae81ff">3</span>]):
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">&#34;</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">&#34;</span><span style="color:#f92672">.</span>join(map(str, r)))
</span></span><span style="display:flex;"><span>lags <span style="color:#f92672">=</span> [r[<span style="color:#ae81ff">4</span>] <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> rows <span style="color:#66d9ef">if</span> r[<span style="color:#ae81ff">4</span>] <span style="color:#f92672">is</span> <span style="color:#f92672">not</span> <span style="color:#66d9ef">None</span>]
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#ae81ff">\n</span><span style="color:#e6db74">KEV adds since </span><span style="color:#e6db74">{</span>cutoff<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>len(rows)<span style="color:#e6db74">}</span><span style="color:#e6db74">; with a CVE record: </span><span style="color:#e6db74">{</span>len(lags)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> lags:
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;median lag </span><span style="color:#e6db74">{</span>median(lags)<span style="color:#e6db74">}</span><span style="color:#e6db74"> days; listed on or before publication day: &#34;</span>
</span></span><span style="display:flex;"><span>          <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>sum(l <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">0</span> <span style="color:#66d9ef">for</span> l <span style="color:#f92672">in</span> lags)<span style="color:#e6db74">}</span><span style="color:#e6db74">; within 7 days: </span><span style="color:#e6db74">{</span>sum(l <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">7</span> <span style="color:#66d9ef">for</span> l <span style="color:#f92672">in</span> lags)<span style="color:#e6db74">}</span><span style="color:#e6db74">; &#34;</span>
</span></span><span style="display:flex;"><span>          <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;older than 365 days: </span><span style="color:#e6db74">{</span>sum(l <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">365</span> <span style="color:#66d9ef">for</span> l <span style="color:#f92672">in</span> lags)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>The last lines of our run:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2026-118527	Ivanti	2026-11-03	2026-10-30	-4
</span></span><span style="display:flex;"><span>CVE-2026-111357	Ivanti	2026-10-15	2026-10-30	15
</span></span><span style="display:flex;"><span>CVE-2026-119977	WordPress	2026-10-31	2026-11-03	3
</span></span><span style="display:flex;"><span>CVE-2024-27834	Apple	2024-05-13	2026-11-04	905
</span></span><span style="display:flex;"><span>CVE-2026-19868	SonicWall	2026-11-06	2026-11-06	0
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>KEV adds since 2026-08-11: 107; with a CVE record: 107
</span></span><span style="display:flex;"><span>median lag 8 days; listed on or before publication day: 24; within 7 days: 53; older than 365 days: 13
</span></span></code></pre></div><p>The figures matched our corpus-wide calculation for the same window. The script makes one HTTP request per KEV entry in the window; for the whole catalog, download a cvelistV5 bulk release zip instead, as we did for the tables above. To turn the output into an alert, swap the print loop for a check of each new <code>cveID</code> against your asset inventory and open risk acceptances.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li><code>datePublished</code> is the CVE record&rsquo;s publication date, not the earliest public disclosure. For CVEs where a vendor advisory or exploit write-up came first, the real exposure window is longer than our gap.</li>
<li>Pre-2023 records include migrated legacy dates; we did not correct them. The 2021-2022 rows describe the catalog&rsquo;s backlog phase and should not be compared directly with later years.</li>
<li>Vendor rows with 8 to 20 entries are small samples. They show a pattern, not a vendor benchmark.</li>
<li>We measured when CISA listed entries, not when exploitation started. KEV does not publish a first-seen-exploited date.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV catalog 2026.11.06, 1,769 entries, dateAdded and dueDate values</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a></td>
          <td>Verified (data)</td>
      </tr>
      <tr>
          <td>datePublished for all 1,769 KEV CVEs, all PUBLISHED</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a>, release cve_2026-11-09_0000Z</td>
          <td>Verified (data)</td>
      </tr>
      <tr>
          <td>Definition of datePublished</td>
          <td><a href="https://github.com/CVEProject/cve-schema">CVE Record Format schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>1,767 of 1,769 KEV CVE records carry a CISA-ADP <code>kev</code> metric</td>
          <td>cvelistV5 release cve_2026-11-09_0000Z</td>
          <td>Verified (data)</td>
      </tr>
      <tr>
          <td>KEV inclusion criteria</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities">CISA KEV page</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>All medians, shares and counts in the tables</td>
          <td>FirewallSync calculation from the two datasets above</td>
          <td>Verified (recomputed with code)</td>
      </tr>
      <tr>
          <td>Reasons for vendor differences (advisory practice, kernel CNA volume)</td>
          <td>Editorial interpretation</td>
          <td>Qualified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>cvelistV5 delta.json and deltaLog.json: Sync CVE Changes</title><link>https://firewallsync.com/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/</link><guid>https://firewallsync.com/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/</guid><pubDate>Sun, 08 Nov 2026 04:55:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>The CVE Program's cvelistV5 repository publishes two small index files, cves/delta.json and cves/deltaLog.json, that tell you which CVE records changed and when. We read 30 days of deltaLog (2,870 entries, 122,761 changes) to show what each field means, what 'new' and 'updated' really cover, how often the feed moves, and a tested Python script that keeps a local mirror current from a checkpoint.</description><enclosure url="https://firewallsync.com/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/cover_hu13465230699450244882.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/cover_hu13465230699450244882.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cvelistv5-delta-json-deltalog-incremental-cve-sync/cover_hu13465230699450244882.jpg" alt="Rows of colourful printed cards sorted into clear acrylic racks on wall-mounted wooden shelves" width="1200" height="630"></p><p>If you keep your own copy of the CVE List, you do not need to re-download all 421,074 record files (the count on <code>main</code> on November 8, 2026) to stay current. The CVE Program&rsquo;s <a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> repository publishes two index files, <code>cves/delta.json</code> and <code>cves/deltaLog.json</code>, that list which records were added or modified and when. Store the last <code>fetchTime</code> you processed, read every batch after it, fetch the newest version of each changed record, and fall back to a full baseline when your checkpoint ages out of the log&rsquo;s 30-day window.</p>
<p>This article explains the fields, what the two change queues really mean, and how the feed behaves in practice, based on the copy of <code>deltaLog.json</code> we downloaded on November 8, 2026 (2,870 batches, October 9 to November 8). It ends with a sync script we ran against that feed. Where a record&rsquo;s fields fit into a patch decision, see our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow from CVE record to ticket</a>.</p>
<h2 id="the-two-files-and-how-they-relate">The two files and how they relate</h2>
<table>
  <thead>
      <tr>
          <th>File</th>
          <th>What it holds</th>
          <th>Size on Nov 8, 2026</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>cves/delta.json</code></td>
          <td>The single most recent change batch</td>
          <td>684 bytes (2 changes)</td>
      </tr>
      <tr>
          <td><code>cves/deltaLog.json</code></td>
          <td>An array of change batches, newest first, for a rolling 30 days</td>
          <td>36.8 MB (2,870 batches)</td>
      </tr>
  </tbody>
</table>
<p>Both files sit on the <code>main</code> branch and are readable over <code>raw.githubusercontent.com</code> without authentication. On the morning we checked, <code>delta.json</code> was byte-for-byte the same object as the first element of <code>deltaLog.json</code>.</p>
<p>The 30-day window comes from two places. The repository README says the log &ldquo;normally retains a rolling 30 days worth of CVE record modification history&rdquo;, and notes that it was temporarily cut to 15 days during a date-normalization exercise in February 2026. The repository&rsquo;s <code>update.yml</code> workflow sets <code>CVES_DEFAULT_DELTA_LOG_HISTORY_IN_DAYS: 30</code>. The window is a setting, so code that assumes exactly 30 days will eventually be wrong. Check the timestamp of the oldest entry instead.</p>
<h2 id="fields-in-a-delta-batch">Fields in a delta batch</h2>
<p>Each batch (the whole of <code>delta.json</code>, or one element of <code>deltaLog.json</code>) looks like this. The example is the real <code>delta.json</code> from 03:15 UTC on November 8, 2026, trimmed to one item:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;fetchTime&#34;</span>: <span style="color:#e6db74">&#34;2026-11-08T03:15:41.690Z&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;numberOfChanges&#34;</span>: <span style="color:#ae81ff">2</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;new&#34;</span>: [],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;updated&#34;</span>: [
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;cveId&#34;</span>: <span style="color:#e6db74">&#34;CVE-2024-28834&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;cveOrgLink&#34;</span>: <span style="color:#e6db74">&#34;https://www.cve.org/CVERecord?id=CVE-2024-28834&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;githubLink&#34;</span>: <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2024/28xxx/CVE-2024-28834.json&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;dateUpdated&#34;</span>: <span style="color:#e6db74">&#34;2026-11-08T03:06:52.175Z&#34;</span>
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>  ],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;error&#34;</span>: []
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Meaning</th>
          <th>What to do with it</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>fetchTime</code></td>
          <td>When the update job pulled changes from CVE Services (UTC)</td>
          <td>Your checkpoint. Store the newest one you processed.</td>
      </tr>
      <tr>
          <td><code>numberOfChanges</code></td>
          <td>Count of items in <code>new</code> + <code>updated</code> + <code>error</code></td>
          <td>A sanity check; it matched the list lengths in all 2,870 batches we read.</td>
      </tr>
      <tr>
          <td><code>new</code></td>
          <td>Records whose JSON file was added to the repository in this batch</td>
          <td>Fetch and insert. Do not assume they are newly disclosed vulnerabilities (see below).</td>
      </tr>
      <tr>
          <td><code>updated</code></td>
          <td>Records whose existing file was modified</td>
          <td>Fetch and replace your copy.</td>
      </tr>
      <tr>
          <td><code>error</code></td>
          <td>Items the tooling could not classify</td>
          <td>Log and alert. Empty in every batch we read.</td>
      </tr>
      <tr>
          <td><code>cveId</code></td>
          <td>The CVE ID</td>
          <td>Your deduplication key.</td>
      </tr>
      <tr>
          <td><code>githubLink</code></td>
          <td>Raw URL of the record JSON on <code>main</code></td>
          <td>Fetch this to get the current record.</td>
      </tr>
      <tr>
          <td><code>cveOrgLink</code></td>
          <td>The record&rsquo;s page on cve.org</td>
          <td>For humans and tickets.</td>
      </tr>
      <tr>
          <td><code>dateUpdated</code></td>
          <td>The record&rsquo;s <code>cveMetadata.dateUpdated</code> when the batch was built</td>
          <td>Use it to keep only the newest change per CVE.</td>
      </tr>
  </tbody>
</table>
<p>Every one of the 122,761 change items we read had exactly those four keys. None of them tells you <em>what</em> changed in the record. If you need a diff, keep the previous version and compare.</p>
<h2 id="what-new-and-updated-actually-mean">What &ldquo;new&rdquo; and &ldquo;updated&rdquo; actually mean</h2>
<p>The labels come from git, not from the CVE lifecycle. The repository&rsquo;s bundled tooling (<code>.github/workflows/dist/index.js</code>) builds each delta from git status: files that were not in the repository go to the new queue, modified files go to updated. Two consequences matter for triage.</p>
<p><strong>A &ldquo;new&rdquo; record can be REJECTED.</strong> We fetched 300 randomly sampled IDs from the new queue in the most recent 600 batches. 272 are PUBLISHED records today and 28 are REJECTED. A reserved ID that is rejected before publication still lands as a new file. If you feed &ldquo;new&rdquo; straight into an alerting channel, filter on <code>cveMetadata.state</code> first. Our <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">guide to CVE record states</a> covers what each state means for a ticket.</p>
<p><strong>Most updates are enrichment.</strong> We fetched 300 sampled IDs from the updated queue (299 parsed) and compared each record&rsquo;s <code>cveMetadata.dateUpdated</code> with the timestamps on its containers. In 244 records the closest match (within two minutes) was the CISA-ADP container, in 51 it was the CNA container and in 4 it was the CVE Program container. In other words, the typical update is CISA adding or revising SSVC, KEV or CVSS data rather than the vendor changing the description or affected versions. That is useful: an update can be the moment a CVE gains an SSVC <code>Exploitation: active</code> value or a KEV entry. See <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">how ADP containers work</a> and <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">how to read the SSVC values CISA adds</a>.</p>
<p>Both samples reflect the records as they stood when we fetched them on November 8, 2026, which can be later than the batch that listed them. Treat the proportions as indicative of one month, not a constant.</p>
<h2 id="how-the-feed-behaves">How the feed behaves</h2>
<p>We counted the whole log with Python (the counts below are recomputed from the file, not estimated).</p>
<table>
  <thead>
      <tr>
          <th>Measure (Oct 9 to Nov 8, 2026)</th>
          <th>Value</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Batches</td>
          <td>2,870</td>
      </tr>
      <tr>
          <td>Items in <code>new</code></td>
          <td>17,699</td>
      </tr>
      <tr>
          <td>Items in <code>updated</code></td>
          <td>105,062</td>
      </tr>
      <tr>
          <td>Items in <code>error</code></td>
          <td>0</td>
      </tr>
      <tr>
          <td>Distinct CVE IDs touched</td>
          <td>31,848</td>
      </tr>
      <tr>
          <td>IDs that appear in more than one batch</td>
          <td>21,774</td>
      </tr>
      <tr>
          <td>Most appearances for one ID</td>
          <td>118 (two IDs, each updated roughly every 6 hours)</td>
      </tr>
      <tr>
          <td>Median gap between batches</td>
          <td>10.2 minutes</td>
      </tr>
      <tr>
          <td>90th percentile gap</td>
          <td>28.9 minutes</td>
      </tr>
      <tr>
          <td>Longest gap</td>
          <td>228.8 minutes</td>
      </tr>
      <tr>
          <td>Largest single batch</td>
          <td>1,417 changes (Oct 13, 21:42 UTC)</td>
      </tr>
  </tbody>
</table>
<p>Three points follow for anyone building on the feed:</p>
<ol>
<li><strong>The same CVE appears many times, and each appearance is a real change.</strong> No <code>(cveId, dateUpdated)</code> pair repeated across the 30 days. If you process batches one by one you will fetch the same record repeatedly; collapse to the newest <code>dateUpdated</code> per ID first.</li>
<li><strong>Schedule is not cadence.</strong> <code>update.yml</code> is set to run every five minutes, but the gaps above show batches arriving anywhere from 2 minutes to nearly 4 hours apart. Alert on &ldquo;no new batch for several hours&rdquo;, not &ldquo;no batch in 5 minutes&rdquo;.</li>
<li><strong><code>dateUpdated</code> trails <code>fetchTime</code> slightly.</strong> Across all items the median lag from a record&rsquo;s <code>dateUpdated</code> to the batch&rsquo;s <code>fetchTime</code> was 4.8 minutes and the maximum was about 102 minutes. That is fine for vulnerability management, but it is not a real-time feed.</li>
</ol>
<h2 id="the-other-ways-to-sync">The other ways to sync</h2>
<p>The README offers two download routes, and the delta files sit alongside them.</p>
<ul>
<li><strong>git.</strong> Clone once, then <code>git pull</code>. Git works out what changed for you. A shallow, blob-less clone keeps the first download small, but a full working tree still means checking out more than 400,000 JSON files.</li>
<li><strong>Release zips.</strong> The README says a baseline named <code>YYYY-MM-DD_all_CVEs_at_midnight.zip</code> is &ldquo;issued at the end of each day at midnight&rdquo; and that hourly files named like <code>2024-04-04_delta_CVEs_at_0100Z.zip</code> &ldquo;only [contain] the deltas since the baseline zip file&rdquo;. We downloaded <code>2026-11-08_delta_CVEs_at_0300Z.zip</code> from the release tagged <code>cve_2026-11-08_0300Z</code>: it held 194 records under <code>deltaCves/</code>, and all 194 IDs appear in <code>deltaLog.json</code> batches between midnight and 03:30 UTC. Note the hourly zip is cumulative since midnight, not just the last hour. In the release assets we fetched, the baseline file actually carried a doubled extension (<code>2026-11-08_all_CVEs_at_midnight.zip.zip</code>).</li>
<li><strong>delta files over HTTPS.</strong> Best when you only want the records that changed and do not want a git clone at all, for example in a serverless function or a scanner plugin.</li>
</ul>
<h2 id="a-tested-sync-script">A tested sync script</h2>
<p>The script below keeps a local folder of records current from a checkpoint. It uses only the Python standard library.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Incremental CVE record sync from cvelistV5&#39;s cves/deltaLog.json (Python 3.9+, stdlib only).&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> pathlib<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> datetime
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>RAW <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/deltaLog.json&#34;</span>
</span></span><span style="display:flex;"><span>STATE <span style="color:#f92672">=</span> pathlib<span style="color:#f92672">.</span>Path(<span style="color:#e6db74">&#34;state.json&#34;</span>)      <span style="color:#75715e"># remembers the last fetchTime we processed</span>
</span></span><span style="display:flex;"><span>MIRROR <span style="color:#f92672">=</span> pathlib<span style="color:#f92672">.</span>Path(<span style="color:#e6db74">&#34;records&#34;</span>)        <span style="color:#75715e"># one JSON file per CVE ID</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">ts</span>(s):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> datetime<span style="color:#f92672">.</span>fromisoformat(s<span style="color:#f92672">.</span>replace(<span style="color:#e6db74">&#34;Z&#34;</span>, <span style="color:#e6db74">&#34;+00:00&#34;</span>))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">get_json</span>(url):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(url, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">60</span>) <span style="color:#66d9ef">as</span> r:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> json<span style="color:#f92672">.</span>load(r)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>():
</span></span><span style="display:flex;"><span>    state <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>loads(STATE<span style="color:#f92672">.</span>read_text()) <span style="color:#66d9ef">if</span> STATE<span style="color:#f92672">.</span>exists() <span style="color:#66d9ef">else</span> {}
</span></span><span style="display:flex;"><span>    log <span style="color:#f92672">=</span> get_json(RAW)                              <span style="color:#75715e"># newest entry first</span>
</span></span><span style="display:flex;"><span>    oldest <span style="color:#f92672">=</span> ts(log[<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>][<span style="color:#e6db74">&#34;fetchTime&#34;</span>])
</span></span><span style="display:flex;"><span>    since <span style="color:#f92672">=</span> ts(state[<span style="color:#e6db74">&#34;lastFetchTime&#34;</span>]) <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lastFetchTime&#34;</span> <span style="color:#f92672">in</span> state <span style="color:#66d9ef">else</span> <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> since <span style="color:#f92672">is</span> <span style="color:#66d9ef">None</span> <span style="color:#f92672">or</span> since <span style="color:#f92672">&lt;</span> oldest:
</span></span><span style="display:flex;"><span>        sys<span style="color:#f92672">.</span>exit(<span style="color:#e6db74">&#34;No checkpoint, or checkpoint older than the log window: &#34;</span>
</span></span><span style="display:flex;"><span>                 <span style="color:#e6db74">&#34;load a midnight baseline zip first, then set lastFetchTime.&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># Collect every change after the checkpoint; keep the newest dateUpdated per CVE.</span>
</span></span><span style="display:flex;"><span>    todo <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> entry <span style="color:#f92672">in</span> log:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> ts(entry[<span style="color:#e6db74">&#34;fetchTime&#34;</span>]) <span style="color:#f92672">&lt;=</span> since:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">break</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> queue <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;new&#34;</span>, <span style="color:#e6db74">&#34;updated&#34;</span>):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> item <span style="color:#f92672">in</span> entry[queue]:
</span></span><span style="display:flex;"><span>                cur <span style="color:#f92672">=</span> todo<span style="color:#f92672">.</span>get(item[<span style="color:#e6db74">&#34;cveId&#34;</span>])
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> cur <span style="color:#f92672">is</span> <span style="color:#66d9ef">None</span> <span style="color:#f92672">or</span> ts(item[<span style="color:#e6db74">&#34;dateUpdated&#34;</span>]) <span style="color:#f92672">&gt;</span> ts(cur[<span style="color:#e6db74">&#34;dateUpdated&#34;</span>]):
</span></span><span style="display:flex;"><span>                    todo[item[<span style="color:#e6db74">&#34;cveId&#34;</span>]] <span style="color:#f92672">=</span> item
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> entry[<span style="color:#e6db74">&#34;error&#34;</span>]:
</span></span><span style="display:flex;"><span>            print(<span style="color:#e6db74">&#34;delta reported errors:&#34;</span>, entry[<span style="color:#e6db74">&#34;error&#34;</span>], file<span style="color:#f92672">=</span>sys<span style="color:#f92672">.</span>stderr)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>    MIRROR<span style="color:#f92672">.</span>mkdir(exist_ok<span style="color:#f92672">=</span><span style="color:#66d9ef">True</span>)
</span></span><span style="display:flex;"><span>    counts <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> cve_id, item <span style="color:#f92672">in</span> todo<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>        record <span style="color:#f92672">=</span> get_json(item[<span style="color:#e6db74">&#34;githubLink&#34;</span>])
</span></span><span style="display:flex;"><span>        state_name <span style="color:#f92672">=</span> record[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;state&#34;</span>]          <span style="color:#75715e"># PUBLISHED or REJECTED</span>
</span></span><span style="display:flex;"><span>        counts[state_name] <span style="color:#f92672">=</span> counts<span style="color:#f92672">.</span>get(state_name, <span style="color:#ae81ff">0</span>) <span style="color:#f92672">+</span> <span style="color:#ae81ff">1</span>
</span></span><span style="display:flex;"><span>        (MIRROR <span style="color:#f92672">/</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve_id<span style="color:#e6db74">}</span><span style="color:#e6db74">.json&#34;</span>)<span style="color:#f92672">.</span>write_text(json<span style="color:#f92672">.</span>dumps(record))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>    STATE<span style="color:#f92672">.</span>write_text(json<span style="color:#f92672">.</span>dumps({<span style="color:#e6db74">&#34;lastFetchTime&#34;</span>: log[<span style="color:#ae81ff">0</span>][<span style="color:#e6db74">&#34;fetchTime&#34;</span>]}))
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>len(todo)<span style="color:#e6db74">}</span><span style="color:#e6db74"> records refreshed since </span><span style="color:#e6db74">{</span>since<span style="color:#f92672">.</span>isoformat()<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>counts<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main()
</span></span></code></pre></div><p><strong>How we tested it.</strong> On November 8, 2026 we ran it with Python 3 on Linux, with a checkpoint set to the sixth-newest batch (<code>2026-11-08T01:47:15.999Z</code>). It downloaded the log, refreshed 29 records (all PUBLISHED) in about five seconds, and moved the checkpoint to <code>2026-11-08T03:15:41.690Z</code>. With the state file deleted it exited with the re-baseline message, as intended.</p>
<p><strong>Limits of the script.</strong> It is a starting point, not a production sync:</p>
<ul>
<li>It downloads the whole 36.8 MB log on every run. For runs every few minutes, poll <code>delta.json</code> and only pull the log when you have missed batches.</li>
<li>It fetches records one at a time and does not retry. Add parallelism, retries and backoff before you point it at a large backlog.</li>
<li><code>githubLink</code> points at <code>main</code>, so you get the record as it is when you fetch it, which may be newer than the batch. That is usually what you want, but it means <code>dateUpdated</code> in your copy can be later than the one in the delta.</li>
<li>It does not handle the rare case of a record file being removed from the repository; the delta format has no &ldquo;deleted&rdquo; queue.</li>
</ul>
<h2 id="where-this-fits-in-a-vulnerability-workflow">Where this fits in a vulnerability workflow</h2>
<p>A local, current mirror is what makes the rest of the prioritization stack cheap: join it to the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">CISA KEV JSON feed</a> and the <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS daily file</a>, and re-run the decision only for the CVE IDs that the delta says changed. Because most updates are CISA-ADP enrichment, a changed record is worth re-scoring even when the vendor text did not move.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>deltaLog retains a rolling 30 days; baseline and hourly zip naming; hourly zip holds deltas since the baseline</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/README.md">cvelistV5 README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Update job scheduled every 5 minutes; 30-day history setting</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/.github/workflows/update.yml">cvelistV5 update.yml</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>new = added file, updated = modified file</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/tree/main/.github/workflows/dist">cvelistV5 workflow tooling, dist/index.js</a></td>
          <td>Verified (bundled source read on Nov 8, 2026)</td>
      </tr>
      <tr>
          <td>Field names, counts, gaps, largest batch</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/deltaLog.json">cves/deltaLog.json</a>, downloaded Nov 8, 2026</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>28 of 300 sampled new IDs REJECTED; 244 of 299 sampled updates match CISA-ADP</td>
          <td>Sampled record files from cvelistV5 <code>main</code>, fetched Nov 8, 2026</td>
          <td>Qualified (random sample, current record state)</td>
      </tr>
      <tr>
          <td>Hourly zip contents (194 records, all in deltaLog)</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/releases">cvelistV5 releases</a>, tag cve_2026-11-08_0300Z</td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Linux Kernel CVE Records: Read Stable-Branch Affected Ranges</title><link>https://firewallsync.com/posts/linux-kernel-cve-records-affected-versions-stable-branches/</link><guid>https://firewallsync.com/posts/linux-kernel-cve-records-affected-versions-stable-branches/</guid><pubDate>Sat, 07 Nov 2026 04:50:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>Every Linux kernel CVE record carries two product entries: git commit ranges and kernel version ranges. We read all 7,987 kernel CNA records published in 2026 to show how the two fit together, where reading only one goes wrong (1,056 records), why the CPE block over-matches fixed kernels, and a tested Python checker for a kernel.org version.</description><enclosure url="https://firewallsync.com/posts/linux-kernel-cve-records-affected-versions-stable-branches/cover_hu12897958757966638349.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/linux-kernel-cve-records-affected-versions-stable-branches/cover_hu12897958757966638349.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/linux-kernel-cve-records-affected-versions-stable-branches/cover_hu12897958757966638349.jpg" alt="A grove of leafless trees with dark, twisting trunks that split into many crossing branches above a pale yellow lawn" width="1200" height="630"></p><p>A Linux kernel CVE record answers &ldquo;is my kernel affected?&rdquo; in two places at once: a list of <strong>git commit ranges</strong> and a list of <strong>kernel version ranges</strong>, each in its own product entry with a different <code>defaultStatus</code>. Read the version entry from top to bottom, stop at the first match, and fall back to <code>affected</code> for anything unlisted. Then check the git entry too, because in about one record in eight it carries version ranges the other entry does not.</p>
<p>We read every record the Linux kernel CNA published in 2026 (7,987 records, January 21 to November 6) from the CVE Program&rsquo;s <a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> bulk export of November 7, 2026, and counted how they are built. This article covers the kernel-specific layout. For the general rules of <code>version</code>, <code>lessThan</code>, <code>lessThanOrEqual</code> and <code>defaultStatus</code> in any CVE record, see our <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">guide to CVE affected versions</a>.</p>
<h2 id="what-a-kernel-cve-record-contains">What a kernel CVE record contains</h2>
<p>The kernel&rsquo;s <a href="https://docs.kernel.org/process/cve.html">CVE process document</a> sets the ground rules. CVEs are assigned &ldquo;after a fix is available and applied to a stable kernel tree&rdquo;, and are tracked &ldquo;by the git commit id of the original fix&rdquo;. No CVEs are assigned for kernel versions that are not &ldquo;currently being actively supported by the Stable/LTS kernel team&rdquo;. And applicability is yours to decide: &ldquo;the applicability of any specific CVE is up to the user of Linux to determine, it is not up to the CVE assignment team.&rdquo;</p>
<p>That produces a very regular structure. Across the 7,987 records:</p>
<table>
  <thead>
      <tr>
          <th>Feature</th>
          <th>Records</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Exactly two entries in <code>affected</code>, both product &ldquo;Linux&rdquo;, vendor &ldquo;Linux&rdquo;</td>
          <td>7,987</td>
      </tr>
      <tr>
          <td><code>repo</code> set to the stable tree (<code>git.kernel.org/.../stable/linux.git</code>)</td>
          <td>7,987</td>
      </tr>
      <tr>
          <td><code>programFiles</code> listing the changed source files</td>
          <td>7,987</td>
      </tr>
      <tr>
          <td>A <code>cpeApplicability</code> block</td>
          <td>7,987</td>
      </tr>
      <tr>
          <td>Version entry starts with the introducing release, then <code>0</code> to it unaffected</td>
          <td>7,911</td>
      </tr>
      <tr>
          <td>Version entry ends with an <code>original_commit_for_fix</code> range to <code>*</code></td>
          <td>7,917</td>
      </tr>
      <tr>
          <td>CNA-supplied CVSS v3.1 metrics</td>
          <td>3,154</td>
      </tr>
  </tbody>
</table>
<p>Every record has the two entries. In 7,923 of them the order is the same: the git entry (<code>defaultStatus</code> unaffected) first, the version entry (<code>defaultStatus</code> affected) second. In the remaining 64, both entries have <code>defaultStatus</code> unaffected, so only explicitly listed versions are affected. The checker below handles both shapes because it evaluates every entry.</p>
<h2 id="entry-1-git-commit-ranges">Entry 1: git commit ranges</h2>
<p>The first product entry has <code>&quot;defaultStatus&quot;: &quot;unaffected&quot;</code> and a list of commit ranges. Here is CVE-2026-99190 (gpiolib, published October 28, 2026), trimmed to the version list:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;d73d2a569e6212f58d89efc6e66ffbb2386a47b1&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;68a2cd5381b8d89dc80bd16dd9cc9ca2d5c25e7f&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;git&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;9b407312755fd5db012413ca005f0f3a661db8dd&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;78189c0e35abbcdab3b8ab0b764d578c02900734&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;git&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;9b407312755fd5db012413ca005f0f3a661db8dd&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;890c3bf08e0fc09eff788a6cfeded7d21b4d3e9d&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;git&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;9b407312755fd5db012413ca005f0f3a661db8dd&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;c28846ca1ac6134881114d272314623c8433e372&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;git&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;6.12.114&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;6.12.115&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}
</span></span></code></pre></div><p>Each git range reads &ldquo;from the commit that introduced the bug, up to but not including a commit that fixed it&rdquo;. The fix commits are the same ones listed in the record&rsquo;s <code>references</code> as <code>git.kernel.org/stable/c/&lt;hash&gt;</code> links, one per branch that received the fix. The CVE schema says <code>repo</code> exists &ldquo;to resolve git hash version ranges&rdquo;, which is the point: comparing commits is not a string comparison. You need a kernel git checkout to ask whether your build contains the introducing commit and not the fix.</p>
<p>If you build kernels from source, this is the most precise data in the record. Most teams run a released version instead, which is what the second entry is for. But note the last line above: a kernel version range inside the git entry. We come back to it below.</p>
<h2 id="entry-2-kernel-version-ranges-first-match-wins">Entry 2: kernel version ranges, first match wins</h2>
<p>The second entry has <code>&quot;defaultStatus&quot;: &quot;affected&quot;</code>. For CVE-2026-99190 it is:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;6.15&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;0&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;6.15&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;6.12.115&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;6.12.*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;6.18.59&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;6.18.*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;7.2.13&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;7.2.*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;7.3&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;original_commit_for_fix&#34;</span>}
</span></span></code></pre></div><p>The <a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE schema</a> gives the rule: walk the list, return the status of the first entry that matches, and return <code>defaultStatus</code> if none do. Applied here:</p>
<ol>
<li><strong>6.15 is affected.</strong> A bare <code>version</code> with no range is an exact match. This is the release that introduced the bug.</li>
<li><strong>Anything before 6.15 is unaffected.</strong> <code>0</code> is the schema&rsquo;s convention for &ldquo;the earliest possible version&rdquo;.</li>
<li><strong>One unaffected range per fixed stable branch.</strong> <code>6.12.115</code> to <code>6.12.*</code> means 6.12.115 and every later 6.12 release. The schema says a trailing <code>*</code> means &ldquo;an arbitrarily large number in the version ordering&rdquo;, so <code>6.12.*</code> closes the range at the end of the 6.12 branch without touching 6.13.</li>
<li><strong>The mainline fix onward is unaffected.</strong> The <code>7.3 ... *</code> entry covers 7.3 and every later release.</li>
<li><strong>Everything else is affected.</strong> 6.18.58, any 6.19 kernel, or 7.2.12 matches no entry and takes <code>defaultStatus</code>.</li>
</ol>
<p>Point 5 is the reason the order and the default both matter. A branch that does not appear at all, such as 6.19 here, has no fix listed and falls through to affected. The kernel&rsquo;s <a href="https://kernel.org/releases.html">releases page</a> currently lists 6.18, 6.12, 6.6, 6.1, 5.15 and 5.10 as longterm branches. If your kernel sits on a branch that a record does not list, the record gives you no fixed version on that branch; the way out is moving to a branch that has one.</p>
<h3 id="the-original_commit_for_fix-type-is-not-defined">The <code>original_commit_for_fix</code> type is not defined</h3>
<p>The last entry&rsquo;s <code>versionType</code> is <code>original_commit_for_fix</code>. It is not among the schema&rsquo;s <code>versionType</code> examples (&ldquo;custom&rdquo;, &ldquo;git&rdquo;, &ldquo;maven&rdquo;, &ldquo;python&rdquo;, &ldquo;rpm&rdquo;, &ldquo;semver&rdquo;), and the kernel&rsquo;s CVE process document does not mention it. From the records, the value looks like the first mainline release containing the original fix, and every one we saw was a release number such as <code>7.3</code> or <code>7.4-rc1</code>. That is our reading of the data, not a published definition.</p>
<p>The practical effect: a parser that refuses unknown version types stops at this entry for every version not matched earlier, which in practice means &ldquo;unknown&rdquo; for any kernel above the newest fixed stable branch. Our <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">generic affected-versions guide</a> shows a strict evaluator doing exactly that on a kernel record. If you choose to compare it as a normal kernel version, which is what the checker below does, make that a documented decision.</p>
<h2 id="why-you-must-read-both-entries">Why you must read both entries</h2>
<p>The <code>6.12.114 lessThan 6.12.115</code> line in the git entry is the trap. In CVE-2026-99190, the bug arrived in mainline at 6.15, so the version entry says every 6.12 kernel is &ldquo;before 6.15&rdquo; and unaffected. But the git entry marks 6.12.114 as affected and the version entry lists a 6.12.115 fix. The commit ranges hint at why: the first range starts from a different commit (<code>d73d2a5...</code>) than the other three (<code>9b40731...</code>), consistent with the bug reaching 6.12 as a stable backport. Read only the version entry and 6.12.114 comes out clean.</p>
<p>This is not rare. In 1,084 of the 7,987 records, the first (git) entry also contains <code>semver</code> ranges. In <strong>1,056 records</strong> (3,136 ranges), at least one of those ranges marks a stable release affected that the version entry, evaluated alone, calls unaffected.</p>
<p>Running the checker below against the record:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 kernel_affected.py CVE-2026-99190.json 6.12.113 6.12.114 6.18.58 6.18.59 6.19.10 7.3 6.12.114-1-amd64
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.12.113: unaffected
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.12.114: affected
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.18.58: affected
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.18.59: unaffected
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.19.10: affected
</span></span><span style="display:flex;"><span>CVE-2026-99190 7.3: unaffected
</span></span><span style="display:flex;"><span>CVE-2026-99190 6.12.114-1-amd64: distribution kernel suffix; use your distro&#39;s tracker
</span></span></code></pre></div><p>(Per-entry lines trimmed; for 6.12.114 the checker reports entry[0] affected via the <code>6.12.114</code>-<code>6.12.115</code> range and entry[1] unaffected via <code>0</code>-<code>6.15</code>.)</p>
<h2 id="why-cpe-matching-over-flags-fixed-kernels">Why CPE matching over-flags fixed kernels</h2>
<p>Every 2026 kernel record also has a <code>cpeApplicability</code> block. For CVE-2026-100075, it is one OR node of eight ranges on <code>cpe:2.3:o:linux:linux_kernel</code>, each starting at 4.7 and ending before one fixed version: 5.10.270, 5.15.221, 6.1.188, 6.6.157, 6.12.110, 6.18.52, 7.2.6, and 7.3. OR&rsquo;d together, those ranges collapse to &ldquo;4.7 up to 7.3&rdquo;, so 5.10.300, a fixed 5.10 kernel, still matches the 7.3 range.</p>
<p>We checked this across the corpus. The version entries name 36,296 fixed stable releases (the start of each <code>X.Y.N</code> to <code>X.Y.*</code> unaffected range). The OR&rsquo;d CPE ranges in the same records match <strong>35,340 of them (97.4%) as vulnerable</strong>, in 7,917 records. CPE ranges have no way to say &ldquo;fixed on the 6.12 branch at 6.12.110&rdquo;; the affected block does. We covered this limit with one example in our <a href="/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/">cpeApplicability guide</a>; the corpus count shows it is the norm for kernel records, not an edge case.</p>
<p>If your scanner or SBOM tool matches kernel CVEs by CPE and reports patched LTS kernels as vulnerable, this is the first thing to check.</p>
<h2 id="distribution-kernels-stop-here">Distribution kernels: stop here</h2>
<p>kernel.org is direct about this: &ldquo;If you see anything at all after the dash, you are running a distribution kernel&rdquo;, and &ldquo;kernel developers can provide no support for them&rdquo;. A RHEL, Ubuntu or Debian kernel version reflects the distribution&rsquo;s base plus its own backports, so <code>5.14.0-...</code> or <code>6.8.0-...</code> cannot be compared to the kernel.org ranges in a CVE record. The kernel&rsquo;s CVE document also says issues only present in a distribution&rsquo;s kernel get CVEs from that distribution, not from the kernel team.</p>
<p>For those kernels, use the distribution&rsquo;s security tracker or its VEX/CSAF data. Our <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explainer</a> covers how to read a &ldquo;not affected&rdquo; statement from a vendor.</p>
<h2 id="a-tested-checker-for-kernelorg-versions">A tested checker for kernel.org versions</h2>
<p>The script below implements the schema&rsquo;s first-match algorithm over <strong>both</strong> product entries, skips git commit ranges (they need a kernel checkout), and reports affected if either entry says so. It refuses versions with a distribution suffix. We ran it with Python 3.12 against CVE-2026-99190 (output above) and across all 7,987 2026 records to produce the counts in this article.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Evaluate a kernel.org version against a Linux kernel CNA record (CVE JSON 5).
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">usage: kernel_affected.py CVE-record.json VERSION [VERSION ...]&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> re<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">key</span>(v):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Kernel version -&gt; comparable tuple. &#39;*&#39; sorts above any number; -rcN sorts before final.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> v <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;*&#34;</span>:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> ((float(<span style="color:#e6db74">&#34;inf&#34;</span>),),)
</span></span><span style="display:flex;"><span>    m <span style="color:#f92672">=</span> re<span style="color:#f92672">.</span>fullmatch(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;(\d+(?:\.\d+)*)(?:\.(\*))?(?:-rc(\d+))?&#34;</span>, v)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> m:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">raise</span> <span style="color:#a6e22e">ValueError</span>(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;not a kernel.org version: </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    nums <span style="color:#f92672">=</span> tuple(int(p) <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> m<span style="color:#f92672">.</span>group(<span style="color:#ae81ff">1</span>)<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;.&#34;</span>))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> m<span style="color:#f92672">.</span>group(<span style="color:#ae81ff">2</span>):                                   <span style="color:#75715e"># &#34;6.12.*&#34; -&gt; 6.12.&lt;infinity&gt;</span>
</span></span><span style="display:flex;"><span>        nums <span style="color:#f92672">+=</span> (float(<span style="color:#e6db74">&#34;inf&#34;</span>),)
</span></span><span style="display:flex;"><span>    rc <span style="color:#f92672">=</span> int(m<span style="color:#f92672">.</span>group(<span style="color:#ae81ff">3</span>)) <span style="color:#66d9ef">if</span> m<span style="color:#f92672">.</span>group(<span style="color:#ae81ff">3</span>) <span style="color:#66d9ef">else</span> float(<span style="color:#e6db74">&#34;inf&#34;</span>)  <span style="color:#75715e"># 7.4-rc1 &lt; 7.4</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (nums <span style="color:#f92672">+</span> (<span style="color:#ae81ff">0</span>,) <span style="color:#f92672">*</span> (<span style="color:#ae81ff">4</span> <span style="color:#f92672">-</span> len(nums)), rc)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">status</span>(product, v):
</span></span><span style="display:flex;"><span>    k <span style="color:#f92672">=</span> key(v)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> product<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versions&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> e<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versionType&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;git&#34;</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>                                 <span style="color:#75715e"># commit ranges need a kernel git checkout</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lessThan&#34;</span> <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> e <span style="color:#f92672">and</span> <span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span> <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> e:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> key(e[<span style="color:#e6db74">&#34;version&#34;</span>]) <span style="color:#f92672">==</span> k:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">return</span> e[<span style="color:#e6db74">&#34;status&#34;</span>], e
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        lo <span style="color:#f92672">=</span> key(e[<span style="color:#e6db74">&#34;version&#34;</span>])
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lessThan&#34;</span> <span style="color:#f92672">in</span> e <span style="color:#f92672">and</span> lo <span style="color:#f92672">&lt;=</span> k <span style="color:#f92672">&lt;</span> key(e[<span style="color:#e6db74">&#34;lessThan&#34;</span>]):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> e[<span style="color:#e6db74">&#34;status&#34;</span>], e
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span> <span style="color:#f92672">in</span> e <span style="color:#f92672">and</span> lo <span style="color:#f92672">&lt;=</span> k <span style="color:#f92672">&lt;=</span> key(e[<span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span>]):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> e[<span style="color:#e6db74">&#34;status&#34;</span>], e
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> product<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;defaultStatus&#34;</span>, <span style="color:#e6db74">&#34;unknown&#34;</span>), <span style="color:#66d9ef">None</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>cve <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>]
</span></span><span style="display:flex;"><span>prods <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;containers&#34;</span>][<span style="color:#e6db74">&#34;cna&#34;</span>][<span style="color:#e6db74">&#34;affected&#34;</span>]       <span style="color:#75715e"># kernel records carry two product entries</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>:]:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;-&#34;</span> <span style="color:#f92672">in</span> v <span style="color:#f92672">and</span> <span style="color:#f92672">not</span> re<span style="color:#f92672">.</span>search(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;-rc\d+$&#34;</span>, v):
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: distribution kernel suffix; use your distro&#39;s tracker&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    verdicts <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> i, p <span style="color:#f92672">in</span> enumerate(prods):
</span></span><span style="display:flex;"><span>        s, e <span style="color:#f92672">=</span> status(p, v)
</span></span><span style="display:flex;"><span>        why <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;defaultStatus&#34;</span> <span style="color:#66d9ef">if</span> e <span style="color:#f92672">is</span> <span style="color:#66d9ef">None</span> <span style="color:#66d9ef">else</span> json<span style="color:#f92672">.</span>dumps({k: e[k] <span style="color:#66d9ef">for</span> k <span style="color:#f92672">in</span> e <span style="color:#66d9ef">if</span> k <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;status&#34;</span>})
</span></span><span style="display:flex;"><span>        verdicts<span style="color:#f92672">.</span>append(s)
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;  entry[</span><span style="color:#e6db74">{</span>i<span style="color:#e6db74">}</span><span style="color:#e6db74">] defaultStatus=</span><span style="color:#e6db74">{</span>p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;defaultStatus&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>s<span style="color:#e6db74">}</span><span style="color:#e6db74"> (matched </span><span style="color:#e6db74">{</span>why<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span><span style="display:flex;"><span>    final <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;affected&#34;</span> <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;affected&#34;</span> <span style="color:#f92672">in</span> verdicts <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;unaffected&#34;</span> <span style="color:#66d9ef">if</span> all(x <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;unaffected&#34;</span> <span style="color:#66d9ef">for</span> x <span style="color:#f92672">in</span> verdicts) <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;unknown&#34;</span>
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>final<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>Limits: it treats <code>original_commit_for_fix</code> as a kernel version comparison (see above), ignores git ranges, and only understands kernel.org version strings. It is a reference for reading the data, not a vulnerability scanner.</p>
<h2 id="turning-the-answer-into-a-decision">Turning the answer into a decision</h2>
<p>&ldquo;Affected&rdquo; in a kernel record means the code is in your build, not that it is reachable. The kernel team says plainly that &ldquo;large numbers of assigned CVEs are not relevant for their systems&rdquo;, and the <code>programFiles</code> list (here <code>drivers/gpio/gpiolib.c</code>) is a quick first filter against your kernel config. After that, the usual signals apply: is it in KEV, what does EPSS say, what is exposed. Our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a> walks through that decision from the CVE record to a ticket.</p>
<p>The kernel team&rsquo;s own advice is to take whole stable releases rather than cherry-pick: &ldquo;it is best to take all released kernel changes, as they are tested together in a unified whole&rdquo;. For most teams, staying on a current LTS point release answers more of these records than evaluating them one by one.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The counts cover Linux kernel CNA records with IDs CVE-2026-* in the PUBLISHED state, from the cvelistV5 export of November 7, 2026. The CNA updates records, so counts will drift. We did not check git ranges against a kernel tree, so the backport explanation for CVE-2026-99190 is an inference from the record. The meaning of <code>original_commit_for_fix</code> is our reading of the data, not a published definition.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Record counts, entry structure, 1,056 records / 3,136 ranges, 36,296 fixed releases and 35,340 CPE matches</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">CVEProject/cvelistV5</a> bulk export 2026-11-07, computed with the script above</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>CVE-2026-99190 and CVE-2026-100075 affected and cpeApplicability data</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">cvelistV5 records</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>First-match algorithm, <code>defaultStatus</code> fallback, <code>*</code> wildcard, <code>repo</code> for git ranges, <code>versionType</code> examples</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE JSON schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVEs assigned after a fix reaches stable, tracked by original fix commit; only supported branches; applicability is the user&rsquo;s; take all stable changes</td>
          <td><a href="https://docs.kernel.org/process/cve.html">Linux kernel CVE process</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Longterm branches; text after the dash means a distribution kernel</td>
          <td><a href="https://kernel.org/releases.html">kernel.org releases</a></td>
          <td>Verified, Nov 7, 2026</td>
      </tr>
      <tr>
          <td>Meaning of <code>original_commit_for_fix</code></td>
          <td>Record data only; no definition found</td>
          <td>Qualified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>EPSS Scores for KEV CVEs: Why New Entries Score Low</title><link>https://firewallsync.com/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/</link><guid>https://firewallsync.com/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/</guid><pubDate>Sat, 07 Nov 2026 04:50:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>An EPSS cutoff of 0.1 misses a quarter of CISA's KEV catalog today, and half of the entries added in the last 90 days. We tracked 113 KEV additions through 145 daily EPSS v5 files: the median score on the day of listing was 0.0076, and entries that crossed 0.1 after listing took a median of 14 days. Data, a tested audit script and the rule it implies.</description><enclosure url="https://firewallsync.com/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/cover_hu7102096413798716976.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/cover_hu7102096413798716976.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/epss-scores-for-kev-cves-low-after-listing-threshold-audit/cover_hu7102096413798716976.jpg" alt="A wooden-framed hourglass with grey sand running from the upper bulb into the lower one, photographed against a black background" width="1200" height="630"></p><p>If your patch pipeline uses an EPSS cutoff to decide what is urgent, it will miss about a quarter of CISA&rsquo;s Known Exploited Vulnerabilities (KEV) catalog, and roughly <strong>half of the entries CISA added in the last 90 days</strong>. Joining KEV catalog 2026.11.06 to the EPSS file for November 6, 2026, 436 of 1,768 scored KEV CVEs sit below 0.1. Tracking recent additions day by day shows why: EPSS scores for newly listed CVEs typically start low and take about two weeks to cross 0.1, which is the window in which a KEV listing matters most.</p>
<p>We argued in our <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV comparison</a> that KEV should be checked before EPSS. This article measures what happens when it is not, and gives you a script to test your own cutoff.</p>
<h2 id="the-snapshot-kev-cves-below-common-epss-cutoffs">The snapshot: KEV CVEs below common EPSS cutoffs</h2>
<p>Data: the <a href="https://github.com/cisagov/kev-data">KEV JSON</a> (catalog version 2026.11.06, released 2026-11-06T19:44Z, 1,769 entries) and the <a href="https://github.com/empiricalsec/epss_scores">EPSS daily file</a> for November 6, 2026 (<code>model_version:v2026.06.15</code>, 400,985 scored CVEs). One KEV entry, CVE-2026-19868, added that day, had no EPSS score yet.</p>
<table>
  <thead>
      <tr>
          <th>EPSS cutoff</th>
          <th>KEV CVEs below it (of 1,768)</th>
          <th>Share</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>0.5</td>
          <td>861</td>
          <td>48.7%</td>
      </tr>
      <tr>
          <td>0.1</td>
          <td>436</td>
          <td>24.7%</td>
      </tr>
      <tr>
          <td>0.05</td>
          <td>307</td>
          <td>17.4%</td>
      </tr>
      <tr>
          <td>0.01</td>
          <td>54</td>
          <td>3.1%</td>
      </tr>
      <tr>
          <td>90th percentile</td>
          <td>248</td>
          <td>14.0%</td>
      </tr>
      <tr>
          <td>95th percentile</td>
          <td>405</td>
          <td>22.9%</td>
      </tr>
  </tbody>
</table>
<p>The median KEV CVE scores 0.536 (99.0th percentile), so the typical entry is well above any cutoff. The problem is the tail, and the tail is not random. If you are unsure what the <code>epss</code> and <code>percentile</code> columns mean, our <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed guide</a> covers them.</p>
<h2 id="newest-kev-entries-score-lowest">Newest KEV entries score lowest</h2>
<p>Split by <code>dateAdded</code>, the share below 0.1 falls as entries age:</p>
<table>
  <thead>
      <tr>
          <th>Listed in KEV</th>
          <th>KEV CVEs</th>
          <th>Below 0.1</th>
          <th>Share</th>
          <th>Median EPSS</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Within 90 days</td>
          <td>106</td>
          <td>52</td>
          <td>49.1%</td>
          <td>0.104</td>
      </tr>
      <tr>
          <td>91 to 365 days ago</td>
          <td>207</td>
          <td>57</td>
          <td>27.5%</td>
          <td>0.444</td>
      </tr>
      <tr>
          <td>More than 1 year ago</td>
          <td>1,455</td>
          <td>327</td>
          <td>22.5%</td>
          <td>0.583</td>
      </tr>
  </tbody>
</table>
<p>Several recent listings we covered are in that first row. On November 6, WordPress CVE-2026-119977 (added November 3) scored 0.0116, Ivanti CVE-2026-111357 and CVE-2026-118527 (added October 30) scored 0.0053 and 0.0082, and VMware ESX CVE-2026-41709 (added October 28) scored 0.0069. Each would sit below even a 0.05 cutoff.</p>
<h2 id="tracking-the-lag-113-kev-additions-day-by-day">Tracking the lag: 113 KEV additions, day by day</h2>
<p>A snapshot cannot show whether low scores are permanent or just early. So we took the 113 KEV entries added between June 16 and October 7, 2026, and looked up each one in every daily EPSS file from June 15 to November 6 (145 files). That window starts the day after EPSS v5 replaced v4, so every score comes from the same model (<code>v2026.06.15</code>); our <a href="/posts/epss-v5-score-shift-model-version-v2026-06-15/">EPSS v5 analysis</a> explains why mixing models would distort this. The October 7 end date gives every entry at least 30 days of follow-up.</p>
<table>
  <thead>
      <tr>
          <th>Day relative to KEV listing</th>
          <th>Entries with a score</th>
          <th>At or above 0.1</th>
          <th>Share</th>
          <th>Median EPSS</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Day before</td>
          <td>77</td>
          <td>15</td>
          <td>19.5%</td>
          <td>0.0072</td>
      </tr>
      <tr>
          <td>Day of listing</td>
          <td>91</td>
          <td>15</td>
          <td>16.5%</td>
          <td>0.0076</td>
      </tr>
      <tr>
          <td>7 days after</td>
          <td>113</td>
          <td>38</td>
          <td>33.6%</td>
          <td>0.0220</td>
      </tr>
      <tr>
          <td>30 days after</td>
          <td>113</td>
          <td>68</td>
          <td>60.2%</td>
          <td>0.1689</td>
      </tr>
  </tbody>
</table>
<p>Notes on reading it:</p>
<ul>
<li><strong>Missing scores are real.</strong> 36 of the 113 had no EPSS score the day before listing and 22 had none on the day itself, usually because the CVE had only just been published. By day 7 all 113 were scored.</li>
<li><strong>Crossing time.</strong> 15 entries were already at or above 0.1 the day before listing. Of the rest, 64 crossed 0.1 later, after a median of <strong>14 days</strong> (23 within a week). 34 had not reached 0.1 by November 6, 36 to 115 days after listing.</li>
<li><strong>Timing, not cause.</strong> EPSS files publish only the score. We cannot see which inputs moved a score, so we do not claim the KEV listing itself caused any rise.</li>
</ul>
<p>Why the gap exists is consistent with how each list is built. CISA adds a CVE when there is &ldquo;reliable evidence that the vulnerability has been actively exploited in the wild&rdquo; (<a href="https://cisa.gov/known-exploited-vulnerabilities">CISA</a>). EPSS is &ldquo;the probability [0-1] of exploitation in the wild in the next 30 days (following score publication)&rdquo;, estimated from data (<a href="https://github.com/empiricalsec/epss_scores">EPSS scores README</a>). The community <a href="https://riskbasedprioritization.github.io/epss/EPSS_and_CISA_KEV">Risk Based Prioritization guide</a> puts it this way: EPSS reflects &ldquo;the automated data we receive and collect&rdquo;, while KEV is &ldquo;a list of human curated CVEs with historical exploitation activity&rdquo;, so a CVE can be in one and score low in the other. It also quotes FIRST&rsquo;s EPSS FAQ: where exploitation is evidenced, &ldquo;that information should supersede anything EPSS has to say&rdquo;.</p>
<h2 id="what-to-change-in-a-pipeline">What to change in a pipeline</h2>
<p>These are our recommendations from the data above; we have not measured them on anyone&rsquo;s backlog.</p>
<ol>
<li><strong>Route on KEV membership before any EPSS rule.</strong> A listed CVE goes to the urgent queue whatever its score. This is the first step in our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">prioritization workflow</a>.</li>
<li><strong>Do not &ldquo;promote&rdquo; KEV entries by adding a bonus to the EPSS score.</strong> A bonus sized for the median KEV entry (0.536) does nothing for an entry at 0.005.</li>
<li><strong>Treat a missing EPSS score as &ldquo;unknown&rdquo;, not zero.</strong> 22 of the 113 entries had no score on the day they were listed. A join that fills nulls with 0 drops exactly those.</li>
<li><strong>Refresh KEV more often than you refresh thresholds.</strong> The <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed</a> updates on listing days; the lag above means the KEV check is doing nearly all of the work for the first two weeks.</li>
<li><strong>Re-run the audit after every EPSS model change</strong>, since cutoffs shift with the model.</li>
</ol>
<h2 id="audit-your-own-cutoff">Audit your own cutoff</h2>
<p>This script joins the two public files and reports how many KEV entries your cutoff misses, by listing age. We ran it with Python 3.12 on the files named above; the output below is for 0.1.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;How many CISA KEV entries would an EPSS cutoff miss?
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">usage: kev_epss_audit.py known_exploited_vulnerabilities.json epss_scores-YYYY-MM-DD.csv.gz CUTOFF&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> gzip<span style="color:#f92672">,</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> date
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>kev_path, epss_path, cutoff <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>], float(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">3</span>])
</span></span><span style="display:flex;"><span>catalog <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(kev_path))
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> gzip<span style="color:#f92672">.</span>open(epss_path, <span style="color:#e6db74">&#34;rt&#34;</span>) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>    header <span style="color:#f92672">=</span> f<span style="color:#f92672">.</span>readline()<span style="color:#f92672">.</span>strip()                 <span style="color:#75715e"># &#34;#model_version:...,score_date:...&#34;</span>
</span></span><span style="display:flex;"><span>    scores <span style="color:#f92672">=</span> {r[<span style="color:#e6db74">&#34;cve&#34;</span>]: float(r[<span style="color:#e6db74">&#34;epss&#34;</span>]) <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(f)}
</span></span><span style="display:flex;"><span>score_day <span style="color:#f92672">=</span> date<span style="color:#f92672">.</span>fromisoformat(header<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;score_date:&#34;</span>)[<span style="color:#ae81ff">1</span>][:<span style="color:#ae81ff">10</span>])
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;KEV catalog </span><span style="color:#e6db74">{</span>catalog[<span style="color:#e6db74">&#39;catalogVersion&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">, EPSS </span><span style="color:#e6db74">{</span>header[<span style="color:#ae81ff">1</span>:]<span style="color:#e6db74">}</span><span style="color:#e6db74">, cutoff </span><span style="color:#e6db74">{</span>cutoff<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>buckets <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;added &lt;= 90 days ago&#34;</span>: [], <span style="color:#e6db74">&#34;91-365 days&#34;</span>: [], <span style="color:#e6db74">&#34;over 1 year&#34;</span>: []}
</span></span><span style="display:flex;"><span>unscored <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> catalog[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> v[<span style="color:#e6db74">&#34;cveID&#34;</span>] <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> scores:
</span></span><span style="display:flex;"><span>        unscored<span style="color:#f92672">.</span>append(v[<span style="color:#e6db74">&#34;cveID&#34;</span>])
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    age <span style="color:#f92672">=</span> (score_day <span style="color:#f92672">-</span> date<span style="color:#f92672">.</span>fromisoformat(v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>]))<span style="color:#f92672">.</span>days
</span></span><span style="display:flex;"><span>    name <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;added &lt;= 90 days ago&#34;</span> <span style="color:#66d9ef">if</span> age <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">90</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;91-365 days&#34;</span> <span style="color:#66d9ef">if</span> age <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">365</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;over 1 year&#34;</span>
</span></span><span style="display:flex;"><span>    buckets[name]<span style="color:#f92672">.</span>append((v[<span style="color:#e6db74">&#34;cveID&#34;</span>], scores[v[<span style="color:#e6db74">&#34;cveID&#34;</span>]]))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>total <span style="color:#f92672">=</span> below <span style="color:#f92672">=</span> <span style="color:#ae81ff">0</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> name, rows <span style="color:#f92672">in</span> buckets<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>    n_below <span style="color:#f92672">=</span> sum(<span style="color:#ae81ff">1</span> <span style="color:#66d9ef">for</span> _, s <span style="color:#f92672">in</span> rows <span style="color:#66d9ef">if</span> s <span style="color:#f92672">&lt;</span> cutoff)
</span></span><span style="display:flex;"><span>    total <span style="color:#f92672">+=</span> len(rows); below <span style="color:#f92672">+=</span> n_below
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>name<span style="color:#e6db74">:</span><span style="color:#e6db74">22</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>len(rows)<span style="color:#e6db74">:</span><span style="color:#e6db74">5</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> KEV CVEs, </span><span style="color:#e6db74">{</span>n_below<span style="color:#e6db74">:</span><span style="color:#e6db74">4</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> below cutoff (</span><span style="color:#e6db74">{</span>n_below<span style="color:#f92672">/</span>len(rows)<span style="color:#e6db74">:</span><span style="color:#e6db74">.1%</span><span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span><span style="color:#e6db74">&#39;all scored&#39;</span><span style="color:#e6db74">:</span><span style="color:#e6db74">22</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>total<span style="color:#e6db74">:</span><span style="color:#e6db74">5</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> KEV CVEs, </span><span style="color:#e6db74">{</span>below<span style="color:#e6db74">:</span><span style="color:#e6db74">4</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> below cutoff (</span><span style="color:#e6db74">{</span>below<span style="color:#f92672">/</span>total<span style="color:#e6db74">:</span><span style="color:#e6db74">.1%</span><span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;not in EPSS file: </span><span style="color:#e6db74">{</span>len(unscored)<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>unscored[:<span style="color:#ae81ff">5</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 kev_epss_audit.py known_exploited_vulnerabilities.json epss_scores-2026-11-06.csv.gz 0.1
</span></span><span style="display:flex;"><span>KEV catalog 2026.11.06, EPSS model_version:v2026.06.15,score_date:2026-11-06T12:00:23Z, cutoff 0.1
</span></span><span style="display:flex;"><span>added &lt;= 90 days ago     106 KEV CVEs,   52 below cutoff (49.1%)
</span></span><span style="display:flex;"><span>91-365 days              207 KEV CVEs,   57 below cutoff (27.5%)
</span></span><span style="display:flex;"><span>over 1 year             1455 KEV CVEs,  327 below cutoff (22.5%)
</span></span><span style="display:flex;"><span>all scored              1768 KEV CVEs,  436 below cutoff (24.7%)
</span></span><span style="display:flex;"><span>not in EPSS file: 1 [&#39;CVE-2026-19868&#39;]
</span></span></code></pre></div><p>Get the KEV file from <code>raw.githubusercontent.com/cisagov/kev-data/main/known_exploited_vulnerabilities.json</code> and the EPSS file from the <code>empiricalsec/epss_scores</code> repository (one gzip file per day under the year folder). The script reads the whole EPSS file into memory, which is fine for about 400,000 rows.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The snapshot is one day (November 6, 2026). The day-by-day sample is 113 entries over one model version, which is enough to show a pattern but not to estimate a precise lag for any vendor or CVE type. &ldquo;Day of listing&rdquo; compares KEV <code>dateAdded</code> to the EPSS file of the same date; their publication times within the day differ. We did not examine why the 34 entries that stayed below 0.1 scored as they did.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV counts, below-cutoff shares, medians, age buckets, per-CVE scores</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a> catalog 2026.11.06 joined to <a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a> 2026-11-06, computed with the script above</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>113-entry day-by-day results (medians, shares, 14-day median crossing, 34 never)</td>
          <td>Same KEV file and 145 daily EPSS files 2026-06-15 to 2026-11-06, all <code>v2026.06.15</code>, computed in Python</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>EPSS column definitions and 30-day window</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">EPSS scores README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV inclusion criterion</td>
          <td><a href="https://cisa.gov/known-exploited-vulnerabilities">CISA KEV catalog page</a></td>
          <td>Verified, Nov 7, 2026</td>
      </tr>
      <tr>
          <td>Data-generating difference between EPSS and KEV; FIRST FAQ &ldquo;supersede&rdquo; quote</td>
          <td><a href="https://riskbasedprioritization.github.io/epss/EPSS_and_CISA_KEV">Risk Based Prioritization guide</a> (community guide quoting FIRST)</td>
          <td>Qualified: secondary for the FIRST quote</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>EPSS v5: How Scores Shifted and What to Re-baseline</title><link>https://firewallsync.com/posts/epss-v5-score-shift-model-version-v2026-06-15/</link><guid>https://firewallsync.com/posts/epss-v5-score-shift-model-version-v2026-06-15/</guid><pubDate>Fri, 06 Nov 2026 04:45:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>EPSS v5 (model v2026.06.15) replaced v4 in the daily feed on June 15, 2026. We diffed the last v4 file against the first v5 file: 26% fewer CVEs at or above 0.1, 41% fewer at or above 0.5, a median score almost three times higher, and only 588 of the top 1,000 CVEs unchanged. Here is what that means for threshold rules, and a tested guard that stops your pipeline on the next model change.</description><enclosure url="https://firewallsync.com/posts/epss-v5-score-shift-model-version-v2026-06-15/cover_hu8687825451073364504.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/epss-v5-score-shift-model-version-v2026-06-15/cover_hu8687825451073364504.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/epss-v5-score-shift-model-version-v2026-06-15/cover_hu8687825451073364504.jpg" alt="Close-up of an oval dial thermometer whose glass is covered in ice crystals, with a red needle visible beneath the frost" width="1200" height="630"></p><p>The EPSS daily file changed models on <strong>June 15, 2026</strong>. The first line of the June 14 file reads <code>#model_version:v2025.03.14</code> (v4). The June 15 file reads <code>#model_version:v2026.06.15</code> (v5). We compared those two files, which together hold 340,204 CVEs scored on both days. If your pipeline gates on a fixed EPSS probability, it lost about a quarter of its flagged CVEs overnight: <strong>23,069 CVEs scored 0.1 or higher under v4 and 17,041 under v5</strong>. Nothing about those vulnerabilities changed that day; the model did.</p>
<p>This article covers what changed in v5, what the shift looks like in the data, and what to re-baseline. For the file format itself (columns, header line, how to join it to a scanner export), see our <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed guide</a>. That guide already lists the five model versions and their start dates; this one measures the most recent change.</p>
<h2 id="what-changed-in-v5">What changed in v5</h2>
<p>Empirical Security, which runs the EPSS model and hosts the data, <a href="https://lifestyle.kbew98country.com/story/220743/empirical-security-releases-epss-v5/">announced EPSS V5 on May 13, 2026</a>. The release says v5 &ldquo;incorporates improved modeling techniques and optimization algorithms&rdquo; and that &ldquo;the calibration step has been refined.&rdquo; It also cites a better classifier for published exploit code and &ldquo;the addition of new data feeds.&rdquo;</p>
<p>The headline claim is a 23% improvement. Empirical&rsquo;s <a href="https://research.empiricalsecurity.com/research/epss-v5-is-here">research post</a> defines the metric only loosely, as one that rewards &ldquo;ranking genuinely exploited vulnerabilities above ones that pose no real threat.&rdquo; On data from May 4, 2026, it reports that v4 scored 0.514 and v5 scored 0.633, where &ldquo;a random guess would score around 0.025, and a perfect model would score 1.0.&rdquo; We did not reproduce that evaluation.</p>
<p>The rollout is visible in the public <a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a> repository:</p>
<table>
  <thead>
      <tr>
          <th>Date</th>
          <th>What the repository shows</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2026-05-22 to 2026-06-08</td>
          <td><code>beta_scores/epssv5_beta-*.csv.gz</code>, header <code>#model_version:v2026.05.12</code></td>
      </tr>
      <tr>
          <td>2026-06-14</td>
          <td>Last daily file on <code>v2025.03.14</code> (v4), <code>score_date</code> 12:55:00Z</td>
      </tr>
      <tr>
          <td>2026-06-15</td>
          <td>First daily file on <code>v2026.06.15</code> (v5), <code>score_date</code> 12:03:41Z</td>
      </tr>
      <tr>
          <td>2026-11-05</td>
          <td>Still <code>v2026.06.15</code></td>
      </tr>
  </tbody>
</table>
<p>The beta files carry a different model label (<code>v2026.05.12</code>) from the production model, so treat beta scores as a preview of v5, not as v5 itself. The repository README lists the model changes for v2, v3 and v4 but, as of November 6, 2026, has no v5 entry. The file header was the reliable signal.</p>
<p>The CSV format did not change. Both files have the same comment line followed by <code>cve,epss,percentile</code>.</p>
<h2 id="how-the-scores-moved">How the scores moved</h2>
<p>We loaded <code>2026/epss_scores-2026-06-14.csv.gz</code> and <code>2026/epss_scores-2026-06-15.csv.gz</code> from the repository and compared CVEs present in both. For a baseline, we ran the same comparison on June 13 and June 14, two consecutive v4 days.</p>
<table>
  <thead>
      <tr>
          <th>Measure</th>
          <th>v4 (June 14)</th>
          <th>v5 (June 15)</th>
          <th>Normal day (June 13 to 14)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVEs scored</td>
          <td>341,350</td>
          <td>340,247</td>
          <td>341,309 to 341,350</td>
      </tr>
      <tr>
          <td>Median EPSS</td>
          <td>0.00262</td>
          <td>0.00751</td>
          <td>unchanged</td>
      </tr>
      <tr>
          <td>Mean EPSS</td>
          <td>0.03328</td>
          <td>0.02934</td>
          <td>unchanged</td>
      </tr>
      <tr>
          <td>99th percentile of EPSS</td>
          <td>0.77230</td>
          <td>0.59470</td>
          <td>n/a</td>
      </tr>
      <tr>
          <td>CVEs at or above 0.1</td>
          <td>23,069</td>
          <td>17,041</td>
          <td>23,071 to 23,069</td>
      </tr>
      <tr>
          <td>CVEs at or above 0.5</td>
          <td>7,193</td>
          <td>4,249</td>
          <td>7,194 to 7,193</td>
      </tr>
      <tr>
          <td>Crossed 0.1 upward / downward</td>
          <td>3,820 / 9,847</td>
          <td></td>
          <td>9 / 11</td>
      </tr>
      <tr>
          <td>Crossed the 90th percentile in / out</td>
          <td>9,837 / 9,944</td>
          <td></td>
          <td>25 / 29</td>
      </tr>
      <tr>
          <td>Moved by 2x or more</td>
          <td>243,259 of 340,204 (71.5%)</td>
          <td></td>
          <td>313 of 341,309</td>
      </tr>
      <tr>
          <td>Top 1,000 by EPSS that stayed top 1,000</td>
          <td>588</td>
          <td></td>
          <td>999</td>
      </tr>
  </tbody>
</table>
<p>Three patterns stand out.</p>
<p><strong>The distribution got narrower.</strong> The lowest v4 score was 0.00001 and 93,484 CVEs sat below 0.001. Under v5 the minimum is 0.00046 and only 3,552 CVEs are below 0.001. At the top end, the 99th percentile fell from 0.772 to 0.595. The median ratio of v5 to v4 score was 2.89: most low-scored CVEs went up, while many of the high scorers came down. That fits the release&rsquo;s description of recalibration, but the release does not say this is the intended effect, so treat it as our reading of the data.</p>
<p><strong>Fixed cutoffs shrank.</strong> A rule like <code>epss &gt;= 0.1</code> matched 26.1% fewer CVEs the day v5 went live, and <code>epss &gt;= 0.5</code> matched 40.9% fewer. Of the CVEs that crossed 0.1, 9,847 dropped below it and only 3,820 rose above it.</p>
<p><strong>Percentile rules kept their size, not their members.</strong> A percentile is a rank, so the 90th-percentile bucket stays about 10% of scored CVEs. But 9,837 CVEs entered it and 9,944 left it in one day, compared with 25 and 29 on a normal v4 day. If you track &ldquo;new CVEs in the top decile&rdquo; as a work queue, June 15 would have looked like a flood.</p>
<p>1,146 CVEs scored on June 14 were missing from the June 15 file and 43 appeared for the first time. By November 5, all but 119 of the missing ones were back in the feed. We did not establish why they were absent that day.</p>
<h2 id="what-happened-to-kev-listed-cves">What happened to KEV-listed CVEs</h2>
<p>To see whether v5 demoted known-exploited vulnerabilities, we took the <a href="https://github.com/cisagov/kev-data">CISA KEV catalog</a> (catalog version 2026.11.04) and kept only the 1,619 entries whose <code>dateAdded</code> is on or before June 14, 2026, so the set matches what was known on the switch date. All 1,619 were scored on both days.</p>
<table>
  <thead>
      <tr>
          <th>KEV CVEs (added by 2026-06-14)</th>
          <th>v4</th>
          <th>v5</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Median EPSS</td>
          <td>0.704</td>
          <td>0.535</td>
      </tr>
      <tr>
          <td>At or above 0.1</td>
          <td>1,224</td>
          <td>1,219</td>
      </tr>
      <tr>
          <td>At or above the 90th percentile</td>
          <td>1,323</td>
          <td>1,366</td>
      </tr>
  </tbody>
</table>
<p>The probability for KEV entries came down along with the rest of the high end. But almost the same number cleared 0.1, and more of them landed in the top decile. KEV-listed CVEs that a 0.1 rule already caught were still caught. A rule set to 0.5 or 0.7 on the assumption that KEV CVEs sit there would have lost many of them, since the median is now about 0.54.</p>
<p>This is one reason KEV should stay a separate input instead of being folded into an EPSS cutoff. Our <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV comparison</a> covers why; the <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">prioritization workflow</a> shows where each signal sits in the patch decision.</p>
<h2 id="what-to-re-baseline">What to re-baseline</h2>
<p>These are our recommendations, based on the data above. We have not measured their effect on any team&rsquo;s backlog.</p>
<ol>
<li><strong>Find every fixed probability cutoff.</strong> SLA rules, ticket triggers, dashboard colours and scanner policies. If one was tuned against v4 data, check how many CVEs it matches today versus before June 15.</li>
<li><strong>Decide what the cutoff is for.</strong> If it sizes a queue (&ldquo;the top few thousand CVEs&rdquo;), switch it to a percentile. If it expresses risk (&ldquo;at least a 10% chance of exploitation in 30 days&rdquo;), keep the probability, because v5&rsquo;s calibration is meant to make that reading more accurate, but accept that the queue size will move.</li>
<li><strong>Do not compare scores across the switch.</strong> A CVE that went from 0.4 to 0.2 on June 15 did not become less likely to be exploited that day. If you chart EPSS history, mark June 15, 2026 (and the earlier switches on 2022-02-04, 2023-03-07 and 2025-03-17).</li>
<li><strong>Keep KEV as an override.</strong> Do not rely on EPSS to bring KEV entries above a high cutoff.</li>
<li><strong>Store <code>model_version</code> next to every score.</strong> Then you can always tell which model produced a score your system acted on.</li>
</ol>
<h2 id="a-guard-for-the-next-model-change">A guard for the next model change</h2>
<p>There is no API callback when the model changes, but the file tells you. This script compares the header of yesterday&rsquo;s and today&rsquo;s file and exits non-zero on a model change, so a CI job or cron wrapper can stop before ingesting.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Stop an EPSS ingest when the model version changes.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 epss_model_guard.py yesterday.csv.gz today.csv.gz&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> gzip<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">header</span>(path):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> gzip<span style="color:#f92672">.</span>open(path, <span style="color:#e6db74">&#34;rt&#34;</span>) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>        first <span style="color:#f92672">=</span> f<span style="color:#f92672">.</span>readline()<span style="color:#f92672">.</span>strip()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> first<span style="color:#f92672">.</span>startswith(<span style="color:#e6db74">&#34;#&#34;</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> {<span style="color:#e6db74">&#34;model_version&#34;</span>: <span style="color:#e6db74">&#34;pre-v2 (no header)&#34;</span>, <span style="color:#e6db74">&#34;score_date&#34;</span>: <span style="color:#e6db74">&#34;&#34;</span>}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> dict(kv<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;:&#34;</span>, <span style="color:#ae81ff">1</span>) <span style="color:#66d9ef">for</span> kv <span style="color:#f92672">in</span> first<span style="color:#f92672">.</span>lstrip(<span style="color:#e6db74">&#34;#&#34;</span>)<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;,&#34;</span>))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>old, new <span style="color:#f92672">=</span> header(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]), header(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;previous: </span><span style="color:#e6db74">{</span>old[<span style="color:#e6db74">&#39;model_version&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> scored </span><span style="color:#e6db74">{</span>old[<span style="color:#e6db74">&#39;score_date&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;current:  </span><span style="color:#e6db74">{</span>new[<span style="color:#e6db74">&#39;model_version&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> scored </span><span style="color:#e6db74">{</span>new[<span style="color:#e6db74">&#39;score_date&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> old[<span style="color:#e6db74">&#34;model_version&#34;</span>] <span style="color:#f92672">!=</span> new[<span style="color:#e6db74">&#34;model_version&#34;</span>]:
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">&#34;MODEL CHANGE: hold threshold-based alerts and re-baseline before ingesting&#34;</span>)
</span></span><span style="display:flex;"><span>    sys<span style="color:#f92672">.</span>exit(<span style="color:#ae81ff">2</span>)
</span></span></code></pre></div><p>We ran it with Python 3 against files from the repository:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 epss_model_guard.py epss_scores-2026-06-14.csv.gz epss_scores-2026-06-15.csv.gz
</span></span><span style="display:flex;"><span>previous: v2025.03.14 scored 2026-06-14T12:55:00Z
</span></span><span style="display:flex;"><span>current:  v2026.06.15 scored 2026-06-15T12:03:41Z
</span></span><span style="display:flex;"><span>MODEL CHANGE: hold threshold-based alerts and re-baseline before ingesting
</span></span><span style="display:flex;"><span>exit=2
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>$ python3 epss_model_guard.py epss_scores-2026-11-04.csv.gz epss_scores-2026-11-05.csv.gz
</span></span><span style="display:flex;"><span>previous: v2026.06.15 scored 2026-11-04T12:03:12Z
</span></span><span style="display:flex;"><span>current:  v2026.06.15 scored 2026-11-05T12:03:04Z
</span></span><span style="display:flex;"><span>exit=0
</span></span></code></pre></div><p>When the guard trips, run the same comparison we did above (counts at your cutoffs, percentile churn, KEV coverage) before you turn alerts back on. Note the <code>score_date</code> format also changed with v4 (from <code>+0000</code> offsets to <code>Z</code>), so parse it as ISO 8601 instead of matching a fixed string.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>We compared one pair of days. Scores on later v5 days have kept moving as new signals arrive, so today&rsquo;s counts differ from June 15.</li>
<li>The 23% improvement figure and the description of model changes come from Empirical Security. We did not test model accuracy.</li>
<li>KEV is a list of vulnerabilities known to be exploited, not a ground truth for EPSS&rsquo;s 30-day prediction window, so the KEV table shows coverage, not accuracy.</li>
<li>We did not find a FIRST page describing v5 in the sources we could reach. If FIRST publishes a different start date or model description, the file headers above are still what your pipeline will see.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Model version and score date of each daily and beta file; all score counts, medians, percentiles and crossings</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a> files for 2026-06-13, 06-14, 06-15, 11-04, 11-05 and <code>beta_scores/</code>, computed with Python</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>README lists v2, v3 and v4 start dates and no v5 entry</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">epss_scores README</a></td>
          <td>Verified, Nov 6, 2026</td>
      </tr>
      <tr>
          <td>v5 announced May 13, 2026; modeling, calibration, exploit-code classifier and data-feed changes; 23% improvement</td>
          <td><a href="https://lifestyle.kbew98country.com/story/220743/empirical-security-releases-epss-v5/">Empirical Security press release</a></td>
          <td>Verified (vendor claim)</td>
      </tr>
      <tr>
          <td>v4 scored 0.514 and v5 0.633 on May 4, 2026 data; random about 0.025</td>
          <td><a href="https://research.empiricalsecurity.com/research/epss-v5-is-here">EPSS V5 Is Here</a></td>
          <td>Qualified: metric not fully specified</td>
      </tr>
      <tr>
          <td>KEV entries added by 2026-06-14 (1,619) and their scores</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a>, catalog 2026.11.04, joined to the EPSS files</td>
          <td>Verified (computed)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>OpenVEX Products and Subcomponents: purl Matching Rules</title><link>https://firewallsync.com/posts/openvex-products-subcomponents-purl-matching-trivy/</link><guid>https://firewallsync.com/posts/openvex-products-subcomponents-purl-matching-trivy/</guid><pubDate>Fri, 06 Nov 2026 04:40:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>An OpenVEX statement suppresses a finding only when its products, subcomponents and purls line up with what the scanner found. Here is how the go-vex library behind Trivy v0.76.2 matches them: versionless purls, qualifier subsets, subcomponent scoping, timestamp inheritance and which status wins, with a schema-valid example document and a tested Python port of the rules.</description><enclosure url="https://firewallsync.com/posts/openvex-products-subcomponents-purl-matching-trivy/cover_hu10637076205791324779.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/openvex-products-subcomponents-purl-matching-trivy/cover_hu10637076205791324779.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/openvex-products-subcomponents-purl-matching-trivy/cover_hu10637076205791324779.jpg" alt="Three weathered wooden crates side by side on a white counter in front of a white brick wall" width="1200" height="630"></p><p>An OpenVEX statement only changes a scanner&rsquo;s output if its <strong>product</strong>, its <strong>subcomponents</strong> and the <strong>purls</strong> inside them match what the scanner found. In the go-vex library that Trivy v0.76.2 uses, the rules are short: the VEX purl is a pattern (no version means every version, and every qualifier it lists must be present), subcomponents narrow a statement to specific packages, and among matching statements the one with the newest timestamp wins. Most &ldquo;VEX didn&rsquo;t work&rdquo; tickets come down to one of those three.</p>
<p>This article is about the identifiers and the matching. For what the statuses and justifications mean, see <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explained</a>. For converting CycloneDX VEX to OpenVEX, see the <a href="/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/">CycloneDX VEX mapping</a>. For CSAF&rsquo;s product tree, see <a href="/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/">CSAF VEX product_tree</a>.</p>
<h2 id="products-and-subcomponents-in-the-spec">Products and subcomponents in the spec</h2>
<p>The <a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification v0.2.0</a> defines a product as &ldquo;a logical unit representing a piece of software&rdquo;, broad enough that &ldquo;anything that can be described in a Software Bill of Materials (SBOM) can be thought of as a product.&rdquo; A subcomponent is &ldquo;any components possibly included in the product where the vulnerability originates.&rdquo; The spec adds that subcomponents &ldquo;SHOULD also list software identifiers and they SHOULD also be listed in the product SBOM.&rdquo;</p>
<p>Products and subcomponents share one structure, which the spec calls <code>Component</code>:</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Required</th>
          <th>What it holds</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>@id</code></td>
          <td>No</td>
          <td>An IRI for the component. A purl is a valid IRI, so <code>@id</code> can be a purl.</td>
      </tr>
      <tr>
          <td><code>identifiers</code></td>
          <td>No</td>
          <td>A map of identifier type to value. The spec lists <code>purl</code>, <code>cpe22</code> and <code>cpe23</code>, and recommends purls.</td>
      </tr>
      <tr>
          <td><code>hashes</code></td>
          <td>No</td>
          <td>A map of hash algorithm to digest, using IANA hash names.</td>
      </tr>
      <tr>
          <td><code>subcomponents</code></td>
          <td>No (products only)</td>
          <td>A list of components the statement is scoped to.</td>
      </tr>
  </tbody>
</table>
<p>All three identifying fields are optional, but the spec says a valid statement &ldquo;MUST identify a product.&rdquo; It also says the product should list &ldquo;as many software identifiers as possible to help VEX processors when matching the product.&rdquo;</p>
<p><code>products</code> itself is optional in a statement because it can be inherited from an encapsulating document, such as the subject of an in-toto attestation. Timestamps can be inherited too. The spec&rsquo;s inheritance rules are: a statement timestamp overrides the document timestamp, which overrides the encapsulating document&rsquo;s; a product ID in a statement overrides product data in the encapsulating document.</p>
<h2 id="how-go-vex-matches-a-statement">How go-vex matches a statement</h2>
<p>Trivy loads OpenVEX documents with <a href="https://github.com/openvex/go-vex">openvex/go-vex</a>, pinned to v0.2.9 in Trivy v0.76.2&rsquo;s <code>go.mod</code>. Its matching code is small enough to read in full. For a finding (CVE, product, optional subcomponent), a statement matches when all of these hold:</p>
<ol>
<li><strong>Vulnerability.</strong> The CVE ID equals the statement&rsquo;s <code>vulnerability.name</code>, its <code>@id</code>, or one of its <code>aliases</code>.</li>
<li><strong>Product.</strong> One entry in <code>products</code> matches the product identifier. A component matches if its <code>@id</code> equals the identifier exactly, or its <code>@id</code> is a purl that matches by the purl rules below, or any value in <code>identifiers</code> equals it (with purl matching for the <code>purl</code> key), or any value in <code>hashes</code> equals it.</li>
<li><strong>Subcomponent.</strong> If that product entry has no <code>subcomponents</code>, or the caller passed no subcomponent, the product match is enough. Otherwise one listed subcomponent must match the subcomponent identifier by the same component rules.</li>
</ol>
<h3 id="the-purl-rules">The purl rules</h3>
<p>go-vex&rsquo;s <code>PurlMatches(purl1, purl2)</code> treats the VEX purl as the general one and the scanned purl as the specific one:</p>
<table>
  <thead>
      <tr>
          <th>VEX purl</th>
          <th>Scanned purl</th>
          <th>Match?</th>
          <th>Rule</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>pkg:apk/alpine/git</code></td>
          <td><code>pkg:apk/alpine/git@2.43.0-r0?arch=x86_64</code></td>
          <td>Yes</td>
          <td>No version in the VEX purl matches any version</td>
      </tr>
      <tr>
          <td><code>pkg:apk/alpine/git@2.43.0-r0</code></td>
          <td><code>pkg:apk/alpine/git@2.43.0-r1</code></td>
          <td>No</td>
          <td>Versions must be equal when both are set</td>
      </tr>
      <tr>
          <td><code>pkg:apk/alpine/git@2.43.0-r0</code></td>
          <td><code>pkg:apk/alpine/git</code></td>
          <td>No</td>
          <td>A versioned VEX purl never matches a versionless scan result</td>
      </tr>
      <tr>
          <td><code>pkg:oci/app?repository_url=ghcr.io%2Facme%2Fapp</code></td>
          <td><code>pkg:oci/app?repository_url=ghcr.io%2Facme%2Fapp&amp;tag=v2</code></td>
          <td>Yes</td>
          <td>Extra qualifiers on the scanned side are ignored</td>
      </tr>
      <tr>
          <td><code>pkg:oci/app?repository_url=ghcr.io%2Facme%2Fapp</code></td>
          <td><code>pkg:oci/app</code></td>
          <td>No</td>
          <td>Every VEX qualifier must be present with the same value</td>
      </tr>
      <tr>
          <td><code>pkg:apk/alpine/git</code></td>
          <td><code>pkg:apk/alpine/git-daemon@2.43.0-r0</code></td>
          <td>No</td>
          <td>Type, namespace and name must be equal</td>
      </tr>
  </tbody>
</table>
<p>The function&rsquo;s own comment says &ldquo;Purl version ranges are not supported yet.&rdquo; A statement covering versions 1.0 to 1.4 needs either one purl per version or no version at all. Leaving the version off is the broadest possible statement, so use it only when you mean every version. Trivy&rsquo;s documentation notes that these matching rules come from a community discussion, not from the purl specification.</p>
<h3 id="which-statement-wins">Which statement wins</h3>
<p>go-vex collects every matching statement and sorts them by timestamp, using the document <code>timestamp</code> for any statement without its own. Trivy takes the last one. It removes the finding only if that statement&rsquo;s status is <code>not_affected</code> or <code>fixed</code>; <code>affected</code> and <code>under_investigation</code> leave the finding in the report.</p>
<p>This is why the spec&rsquo;s update example keeps the original document timestamp on the first statement when a second one is added. If you edit a document, bump <code>timestamp</code> and <code>version</code> at the document level, and the old statements lose their original time, the ordering of statements for the same CVE can change.</p>
<h2 id="a-tested-example">A tested example</h2>
<p>The document below has two statements for CVE-2024-32002 on a container image (the first <code>under_investigation</code> with its own timestamp, the second <code>not_affected</code> inheriting the newer document timestamp) and one statement for a Go service. It validates against the OpenVEX JSON schema (<code>openvex_json_schema.json</code> from the spec repository, checked with Python <code>jsonschema</code>).</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@context&#34;</span>: <span style="color:#e6db74">&#34;https://openvex.dev/ns/v0.2.0&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;https://example.com/vex/app-2026-11-06&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;author&#34;</span>: <span style="color:#e6db74">&#34;Example Security Team &lt;security@example.com&gt;&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-11-06T09:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#ae81ff">2</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;statements&#34;</span>: [
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-11-01T10:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [
</span></span><span style="display:flex;"><span>        {
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/app?repository_url=ghcr.io%2Facme%2Fapp&#34;</span>,
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;subcomponents&#34;</span>: [{<span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:apk/alpine/git&#34;</span>}]
</span></span><span style="display:flex;"><span>        }
</span></span><span style="display:flex;"><span>      ],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;under_investigation&#34;</span>
</span></span><span style="display:flex;"><span>    },
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2024-32002&#34;</span>},
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [
</span></span><span style="display:flex;"><span>        {
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/app?repository_url=ghcr.io%2Facme%2Fapp&#34;</span>,
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;subcomponents&#34;</span>: [{<span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:apk/alpine/git&#34;</span>}]
</span></span><span style="display:flex;"><span>        }
</span></span><span style="display:flex;"><span>      ],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;justification&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status_notes&#34;</span>: <span style="color:#e6db74">&#34;git is installed for build-time use only; the runtime entrypoint never invokes git clone.&#34;</span>
</span></span><span style="display:flex;"><span>    },
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2023-0001&#34;</span>},
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [
</span></span><span style="display:flex;"><span>        {
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:golang/example.com/svc@v1.4.0&#34;</span>,
</span></span><span style="display:flex;"><span>          <span style="color:#f92672">&#34;subcomponents&#34;</span>: [{<span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:golang/golang.org/x/net@v0.17.0&#34;</span>}]
</span></span><span style="display:flex;"><span>        }
</span></span><span style="display:flex;"><span>      ],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;justification&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_present&#34;</span>
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>  ]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>The image, service names and CVE-2023-0001 statement are illustrative. To check the matching without a registry or a real image, we wrote a Python port of go-vex v0.2.9&rsquo;s <code>PurlMatches</code>, <code>Component.Matches</code>, <code>Product.Matches</code>, <code>Statement.Matches</code> and timestamp sort, using the <code>packageurl-python</code> parser, and ran it with Python 3 against the document. It is a port, not the Go library itself, so treat it as a reading aid. Output, with long purls shortened:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>product                                    subcomponent                          result
</span></span><span style="display:flex;"><span>pkg:oci/app?repository_url=…%2Fapp         pkg:apk/alpine/git@2.43.0-r0?arch=…   2 matches, effective=not_affected
</span></span><span style="display:flex;"><span>pkg:oci/app?repository_url=…%2Fapp         pkg:apk/alpine/git-daemon@2.43.0-r0   no match
</span></span><span style="display:flex;"><span>pkg:oci/app?repository_url=…%2Fapp         (none)                                2 matches, effective=not_affected
</span></span><span style="display:flex;"><span>pkg:oci/app                                pkg:apk/alpine/git@2.43.0-r0          no match
</span></span><span style="display:flex;"><span>pkg:oci/app?repository_url=…%2Fapp&amp;tag=v2  pkg:apk/alpine/git@2.43.0-r0          2 matches, effective=not_affected
</span></span><span style="display:flex;"><span>pkg:golang/example.com/svc@v1.4.0          pkg:golang/golang.org/x/net@v0.17.0   1 match,  effective=not_affected
</span></span><span style="display:flex;"><span>pkg:golang/example.com/svc@v1.5.0          pkg:golang/golang.org/x/net@v0.17.0   no match
</span></span><span style="display:flex;"><span>pkg:golang/example.com/svc@v1.4.0          pkg:golang/golang.org/x/net@v0.18.0   no match
</span></span></code></pre></div><p>What the rows show:</p>
<ul>
<li><strong>Row 1:</strong> the versionless <code>pkg:apk/alpine/git</code> subcomponent covers the installed <code>git@2.43.0-r0</code>. Both statements match, and the <code>not_affected</code> one wins because it inherits the later document timestamp.</li>
<li><strong>Row 2:</strong> <code>git-daemon</code> is a different package name, so the statement does not cover it, even though it ships from the same source.</li>
<li><strong>Row 3:</strong> with no subcomponent passed, the product match alone is enough. Trivy does not hit this case for an image root (see below), but other consumers of go-vex may.</li>
<li><strong>Rows 4 and 5:</strong> the <code>repository_url</code> qualifier on the VEX side must be present on the scanned side. A scan purl without it fails; one with an extra <code>tag</code> qualifier passes.</li>
<li><strong>Rows 6 to 8:</strong> pinned versions on both product and subcomponent mean a new service release (v1.5.0) or a dependency bump (x/net v0.18.0) silently drops the statement. That is often what you want, since the reachability analysis was for that build, but it means the document needs updating each release.</li>
</ul>
<h2 id="how-trivy-applies-it-to-a-dependency-tree">How Trivy applies it to a dependency tree</h2>
<p>Trivy&rsquo;s VEX guide and its <code>filterVulnerabilities</code> code show how it feeds this matcher. For each finding it first asks whether the vulnerable package itself is a not-affected product (no subcomponent). If not, it walks up the dependency graph from the vulnerable package and asks, for each parent, whether &ldquo;parent as product, vulnerable package as subcomponent&rdquo; is not affected. The finding is removed only if no path reaches the root without passing a not-affected node.</p>
<p>Two consequences:</p>
<ul>
<li><strong>A statement on an intermediate dependency can suppress a finding, but only if every path goes through it.</strong> Trivy&rsquo;s documentation gives the example of Module C reached through both B and D: a VEX statement for B does not suppress the finding, because D might still be affected.</li>
<li><strong>A statement on the image with no subcomponents suppresses the CVE in every package in that image.</strong> Trivy&rsquo;s docs call this out and recommend subcomponents for &ldquo;reducing the risk of incorrect filtering.&rdquo;</li>
</ul>
<p>Trivy needs a purl on the scanned side. Its OpenVEX matcher returns nothing when the product has no purl, so cpe-only or hash-only VEX products are not matched by Trivy even though the spec allows them. Use the purls Trivy prints in its JSON report.</p>
<h2 id="writing-statements-that-match">Writing statements that match</h2>
<p>These are our recommendations, based on the rules above:</p>
<ol>
<li><strong>Copy purls from the scanner or SBOM output.</strong> Do not type them by hand; the type, namespace and name must match exactly. Our guide to <a href="/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/">CVE packageURL and SBOM purls</a> covers common purl pitfalls.</li>
<li><strong>Always add subcomponents for image-level or application-level products.</strong> A product with no subcomponents is a blanket statement.</li>
<li><strong>Add a version only when the claim depends on the build.</strong> Reachability claims (<code>vulnerable_code_not_in_execute_path</code>) usually do. &ldquo;This package never ships the vulnerable file&rdquo; claims may not.</li>
<li><strong>Be sparing with qualifiers in the VEX purl.</strong> Each one is a requirement. <code>arch</code> is useful if only one architecture is unaffected; otherwise leave it out.</li>
<li><strong>Keep timestamps on old statements when you append new ones,</strong> and bump the document <code>version</code>. Use <code>status_notes</code> to record how the status was determined. Trivy does not read it, but reviewers do.</li>
<li><strong>Track the result.</strong> Our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">prioritization workflow</a> shows where a VEX decision sits between the CVE record and the ticket.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>Matching behaviour is from go-vex v0.2.9 and Trivy v0.76.2 source and docs. Grype, vexctl and commercial scanners have their own matchers and may differ.</li>
<li>Our test used a Python port of the go-vex rules, not the Go library or a Trivy scan. The port follows the source shown in the references; differences in purl parsing between <code>packageurl-python</code> and <code>packageurl-go</code> are possible for unusual purls.</li>
<li>We did not cover OpenVEX attestations attached to OCI images (<code>--vex oci</code>) or VEX repositories.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Product, subcomponent, Component fields, identifier types, inheritance and update rules</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification v0.2.0</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Example document is schema-valid</td>
          <td><a href="https://github.com/openvex/spec/blob/main/openvex_json_schema.json">openvex_json_schema.json</a>, checked with Python jsonschema</td>
          <td>Verified (tested)</td>
      </tr>
      <tr>
          <td>Trivy v0.76.2 uses go-vex v0.2.9; applies only not_affected and fixed; latest statement wins; requires a product purl</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/pkg/vex/openvex.go">Trivy pkg/vex/openvex.go</a> and go.mod</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Component, product, statement, vulnerability and purl matching rules; no version-range support; timestamp sort</td>
          <td><a href="https://github.com/openvex/go-vex/tree/v0.2.9/pkg/vex">go-vex v0.2.9 pkg/vex</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Dependency-tree traversal and the multiple-paths rule</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.2/docs/guide/supply-chain/vex/file.md">Trivy local VEX files guide</a> and pkg/vex/vex.go</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Match results in the example table</td>
          <td>Python port of go-vex v0.2.9 rules, run against the example document</td>
          <td>Qualified: port, not the Go library</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE packageURL: Join CVE JSON 5.2 Records to SBOM purls</title><link>https://firewallsync.com/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/</link><guid>https://firewallsync.com/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/</guid><pubDate>Thu, 05 Nov 2026 04:40:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>CVE Record Format 5.2.0 added an optional packageURL field to affected entries. We counted it across all 400,445 published records: 1.6% carry one, 12 CNAs produce them, and a third of GitHub purls are not in canonical lowercase. Here is what the field means, how to normalize it, and a tested Python join against a CycloneDX SBOM.</description><enclosure url="https://firewallsync.com/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/cover_hu10950611722794385026.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/cover_hu10950611722794385026.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-packageurl-purl-cve-json-5-2-affected-sbom-matching/cover_hu10950611722794385026.jpg" alt="Black-and-white close-up of a bank of grey metal mailboxes, each door marked with a box number and a four-digit unit number" width="1200" height="630"></p><p><strong>Scope.</strong> This is a specification explainer with measurements from public data. We read the CVE Record Format schema on the CVEProject/cve-schema main branch (default <code>dataVersion</code> 5.2.0), RFD 0002 in the same repository, and the Package URL specification in the package-url/purl-spec repository. We counted fields across every record in the cvelistV5 release <code>cve_2026-11-05_0400Z</code> (the &ldquo;all CVEs at midnight&rdquo; bundle for 5 November 2026) with Python 3.13, and ran the join script below against that data. The script is illustrative.</p>
<h2 id="the-short-answer">The short answer</h2>
<p>Since CVE Record Format 5.2.0, each entry in a record&rsquo;s <code>affected</code> array can carry a <code>packageURL</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vendor&#34;</span>: <span style="color:#e6db74">&#34;fast-uri&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;product&#34;</span>: <span style="color:#e6db74">&#34;fast-uri&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;packageURL&#34;</span>: <span style="color:#e6db74">&#34;pkg:npm/fast-uri&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;defaultStatus&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;versions&#34;</span>: [
</span></span><span style="display:flex;"><span>    {<span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;2.4.2&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;2.4.8&#34;</span>}
</span></span><span style="display:flex;"><span>  ]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>That fragment is from CVE-2026-110180 (trimmed to its first version range). Three rules decide how you use it:</p>
<ol>
<li><strong>No version in the purl.</strong> The schema says the Package URL &ldquo;MUST NOT include a version&rdquo;. Strip the version from your SBOM purl before you look it up.</li>
<li><strong>The versions array still decides.</strong> A purl match finds the records about your package. Whether your version is affected comes from <code>versions</code> and <code>defaultStatus</code>, read with the record&rsquo;s <code>versionType</code>.</li>
<li><strong>It is optional and still rare.</strong> 1.6% of published records carry it today, from 12 CNAs. Treat it as a fast, precise path where it exists, and keep your CPE and name-based matching for the rest.</li>
</ol>
<h2 id="what-the-schema-says">What the schema says</h2>
<p>The field definition in the 5.2.0 schema:</p>
<blockquote>
<p>&ldquo;A Package URL, a unified URL specification for identifying packages hosted by known package hosts. The Package URL MUST NOT include a version.&rdquo;</p>
</blockquote>
<p>The schema types it as a URI and lists examples such as <code>pkg:deb/debian/curl?arch=i386&amp;distro=jessie</code>, <code>pkg:npm/%40angular/animation</code> and <code>pkg:golang/google.golang.org/genproto#googleapis/api/annotations</code>. So qualifiers and subpaths are allowed; only the version is out.</p>
<p><code>packageURL</code> does not satisfy the identifier requirement on its own. The <code>product</code> definition still requires either <code>vendor</code> and <code>product</code> or <code>collectionURL</code> and <code>packageName</code>, plus either <code>versions</code> or <code>defaultStatus</code>. In our scan, every one of the 7,324 entries with a purl also had one of those identifier pairs.</p>
<h3 id="why-it-was-added">Why it was added</h3>
<p>RFD 0002, &ldquo;Supporting Package URLs in CVE&rdquo;, is the design record. It argues that the existing identifier fields &ldquo;are of limited value for automated applicability matching&rdquo; and that Package URLs &ldquo;are better suited than CPEs for capturing distinctions between different distributions of particular open source software&rdquo;.</p>
<p>Two design choices in the RFD matter to consumers:</p>
<ul>
<li><strong>Validation is mostly outside the schema.</strong> The JSON schema only checks that the value is a URI; the RFD says it &ldquo;would be further constrained by CVE Services&rdquo; to ensure the values are valid purls, because encoding purl rules in a regular expression &ldquo;would be excessive and difficult to maintain&rdquo;. Do not assume the schema has validated purl syntax for you.</li>
<li><strong>Versions live in one place.</strong> &ldquo;All version information should only be included within the existing <code>versions</code> field.&rdquo; That keeps one version-matching code path for consumers.</li>
</ul>
<p>The cve-schema repository tags <code>v5.2.0</code>; the tagged commit is dated 3 September 2025. Records that declare <code>dataVersion</code> 5.1 cannot use the field. In our snapshot, 144,619 published records (36.1%) declared 5.2 and 255,826 declared 5.1.</p>
<h2 id="what-the-data-shows">What the data shows</h2>
<p>We read every record in the 5 November 2026 bundle: 400,445 published and 18,890 rejected. Only CNA containers carried the field; no ADP container had a <code>packageURL</code>.</p>
<table>
  <thead>
      <tr>
          <th>Measure</th>
          <th>Count</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Published records with a CNA <code>packageURL</code></td>
          <td>6,284 (1.57%)</td>
      </tr>
      <tr>
          <td><code>affected</code> entries with a <code>packageURL</code></td>
          <td>7,324</td>
      </tr>
      <tr>
          <td>CNAs that use it</td>
          <td>12</td>
      </tr>
      <tr>
          <td>Purls that embed a version</td>
          <td>0</td>
      </tr>
      <tr>
          <td>Purls with qualifiers</td>
          <td>76</td>
      </tr>
      <tr>
          <td>Entries with both <code>packageURL</code> and <code>cpes</code></td>
          <td>863</td>
      </tr>
  </tbody>
</table>
<h3 id="who-writes-it">Who writes it</h3>
<table>
  <thead>
      <tr>
          <th>CNA (short name)</th>
          <th>Records</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>VulnCheck</td>
          <td>4,949</td>
      </tr>
      <tr>
          <td>apache</td>
          <td>455</td>
      </tr>
      <tr>
          <td>EEF</td>
          <td>387</td>
      </tr>
      <tr>
          <td>CPANSec</td>
          <td>199</td>
      </tr>
      <tr>
          <td>openjs</td>
          <td>147</td>
      </tr>
      <tr>
          <td>mitre</td>
          <td>61</td>
      </tr>
      <tr>
          <td>bcorg</td>
          <td>51</td>
      </tr>
      <tr>
          <td>RTI</td>
          <td>31</td>
      </tr>
      <tr>
          <td>Four others</td>
          <td>1 each</td>
      </tr>
  </tbody>
</table>
<p>VulnCheck accounts for 78.8% of records with a purl. If you build matching around this field, most of your hits will come from one CNA&rsquo;s conventions, so test against its records.</p>
<h3 id="which-purl-types">Which purl types</h3>
<p>The most common types across the 7,324 entries: <code>github</code> 2,371, <code>npm</code> 1,729, <code>pypi</code> 805, <code>maven</code> 533, <code>golang</code> 498, <code>composer</code> 314, <code>hex</code> 309, <code>cargo</code> 242, <code>cpan</code> 210, <code>otp</code> 82, <code>software-id</code> 72, <code>nuget</code> 40, <code>gem</code> 31, <code>generic</code> 31.</p>
<p>Two of those need care:</p>
<ul>
<li><strong><code>pkg:github</code></strong> identifies a source repository, not a registry package. VulnCheck wrote 1,917 of the GitHub entries and EEF 413. Your SBOM will usually describe the same software as <code>pkg:npm/...</code> or <code>pkg:pypi/...</code>, so a <code>pkg:github</code> record will not match it directly. You need a mapping from repository to registry package, or a fallback on <code>collectionURL</code> and <code>packageName</code>.</li>
<li><strong><code>pkg:software-id</code></strong> appears 72 times (for example <code>pkg:software-id/erlang.org/otp</code>), but the purl-spec decision record that registers it, dated 21 October 2026, still has the status &ldquo;proposed&rdquo;. Expect parsers to reject it until the type is registered.</li>
</ul>
<h3 id="how-fast-it-is-growing">How fast it is growing</h3>
<p>By the month each record was first published (as the records stand today, so later updates are included):</p>
<table>
  <thead>
      <tr>
          <th>Month</th>
          <th>Records with a purl</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>March 2026</td>
          <td>256</td>
      </tr>
      <tr>
          <td>June 2026</td>
          <td>447</td>
      </tr>
      <tr>
          <td>August 2026</td>
          <td>976</td>
      </tr>
      <tr>
          <td>September 2026</td>
          <td>1,155</td>
      </tr>
      <tr>
          <td>October 2026</td>
          <td>2,000</td>
      </tr>
  </tbody>
</table>
<p>October&rsquo;s 2,000 are 11.7% of the 17,058 records first published that month.</p>
<h2 id="normalize-before-you-join">Normalize before you join</h2>
<p>The purl specification makes the <code>type</code> case-insensitive with lowercase as the canonical form, and some types go further. The type definitions say a GitHub namespace and name are &ldquo;not case sensitive and shall be lowercased&rdquo;, and a PyPI name &ldquo;shall be lowercased and underscore _ replaced with a dash -&rdquo;. npm names are case-sensitive.</p>
<p>CVE records do not always follow those rules:</p>
<ul>
<li>756 of the 2,371 <code>pkg:github</code> entries (31.9%) contain uppercase letters, for example <code>pkg:github/Finsys/dockhand</code>.</li>
<li>9 <code>pkg:pypi</code> entries have uppercase letters and 24 contain an underscore.</li>
<li>GitPython is a concrete case. In the 34 records we found for it, 28 use <code>pkg:pypi/gitpython</code> and 6 use <code>pkg:pypi/GitPython</code>. An exact string join misses one set or the other, depending on how your SBOM tool spells it.</li>
</ul>
<p>So normalize both sides the same way before comparing: lowercase the type; lowercase <code>github</code> and <code>pypi</code> paths; replace <code>_</code> with <code>-</code> for <code>pypi</code>; drop the version and qualifiers from the SBOM purl. Leave npm, Maven and other case-sensitive types alone.</p>
<h2 id="a-tested-join">A tested join</h2>
<p>Two scripts. The first builds an index from the cvelistV5 bundle; the second joins a CycloneDX SBOM to it and applies the <code>versions</code> ranges.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;build_index.py: index CNA affected[] entries that carry packageURL,
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">keyed by a normalized purl (no version, no qualifiers).&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> zipfile
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">norm</span>(purl):
</span></span><span style="display:flex;"><span>    core <span style="color:#f92672">=</span> purl<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;#&#34;</span>)[<span style="color:#ae81ff">0</span>]<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;?&#34;</span>)[<span style="color:#ae81ff">0</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;@&#34;</span> <span style="color:#f92672">in</span> core<span style="color:#f92672">.</span>rsplit(<span style="color:#e6db74">&#34;/&#34;</span>, <span style="color:#ae81ff">1</span>)[<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>]:                 <span style="color:#75715e"># drop a version (SBOM side)</span>
</span></span><span style="display:flex;"><span>        core <span style="color:#f92672">=</span> core<span style="color:#f92672">.</span>rsplit(<span style="color:#e6db74">&#34;@&#34;</span>, <span style="color:#ae81ff">1</span>)[<span style="color:#ae81ff">0</span>]
</span></span><span style="display:flex;"><span>    scheme, rest <span style="color:#f92672">=</span> core<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;:&#34;</span>, <span style="color:#ae81ff">1</span>)
</span></span><span style="display:flex;"><span>    ptype, _, path <span style="color:#f92672">=</span> rest<span style="color:#f92672">.</span>partition(<span style="color:#e6db74">&#34;/&#34;</span>)
</span></span><span style="display:flex;"><span>    ptype <span style="color:#f92672">=</span> ptype<span style="color:#f92672">.</span>lower()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> ptype <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;github&#34;</span>, <span style="color:#e6db74">&#34;pypi&#34;</span>):                     <span style="color:#75715e"># case-insensitive types per purl-spec</span>
</span></span><span style="display:flex;"><span>        path <span style="color:#f92672">=</span> path<span style="color:#f92672">.</span>lower()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> ptype <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;pypi&#34;</span>:
</span></span><span style="display:flex;"><span>        path <span style="color:#f92672">=</span> path<span style="color:#f92672">.</span>replace(<span style="color:#e6db74">&#34;_&#34;</span>, <span style="color:#e6db74">&#34;-&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;pkg:</span><span style="color:#e6db74">{</span>ptype<span style="color:#e6db74">}</span><span style="color:#e6db74">/</span><span style="color:#e6db74">{</span>path<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>(zip_path, out_path):
</span></span><span style="display:flex;"><span>    z <span style="color:#f92672">=</span> zipfile<span style="color:#f92672">.</span>ZipFile(zip_path)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> open(out_path, <span style="color:#e6db74">&#34;w&#34;</span>) <span style="color:#66d9ef">as</span> out:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> n <span style="color:#f92672">in</span> z<span style="color:#f92672">.</span>namelist():
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> n<span style="color:#f92672">.</span>endswith(<span style="color:#e6db74">&#34;.json&#34;</span>) <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;/CVE-&#34;</span> <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> n:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>            rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>loads(z<span style="color:#f92672">.</span>read(n))
</span></span><span style="display:flex;"><span>            meta <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>]
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> meta<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;state&#34;</span>) <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;PUBLISHED&#34;</span>:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> a <span style="color:#f92672">in</span> rec[<span style="color:#e6db74">&#34;containers&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;cna&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;affected&#34;</span>, []):
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;packageURL&#34;</span>):
</span></span><span style="display:flex;"><span>                    out<span style="color:#f92672">.</span>write(json<span style="color:#f92672">.</span>dumps({<span style="color:#e6db74">&#34;key&#34;</span>: norm(a[<span style="color:#e6db74">&#34;packageURL&#34;</span>]), <span style="color:#e6db74">&#34;cve&#34;</span>: meta[<span style="color:#e6db74">&#34;cveId&#34;</span>],
</span></span><span style="display:flex;"><span>                        <span style="color:#e6db74">&#34;defaultStatus&#34;</span>: a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;defaultStatus&#34;</span>), <span style="color:#e6db74">&#34;versions&#34;</span>: a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versions&#34;</span>, [])}) <span style="color:#f92672">+</span> <span style="color:#e6db74">&#34;</span><span style="color:#ae81ff">\n</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;join_sbom.py: join CycloneDX component purls to the CVE packageURL index, then apply versions[] ranges.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Illustrative: only semver-like versionType values are compared; everything else is flagged for review.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> packaging.version <span style="color:#f92672">import</span> Version, InvalidVersion
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> build_index <span style="color:#f92672">import</span> norm
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">status_for</span>(ver, entry):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>        v <span style="color:#f92672">=</span> Version(ver)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">except</span> InvalidVersion:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;review: unparseable version&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> entry[<span style="color:#e6db74">&#34;versions&#34;</span>]:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> r<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versionType&#34;</span>) <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;semver&#34;</span>, <span style="color:#e6db74">&#34;maven&#34;</span>, <span style="color:#e6db74">&#34;python&#34;</span>, <span style="color:#66d9ef">None</span>):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;review: versionType=</span><span style="color:#e6db74">{</span>r<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;versionType&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>            lo <span style="color:#f92672">=</span> Version(r[<span style="color:#e6db74">&#34;version&#34;</span>]) <span style="color:#66d9ef">if</span> r[<span style="color:#e6db74">&#34;version&#34;</span>] <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;0&#34;</span>, <span style="color:#e6db74">&#34;*&#34;</span>) <span style="color:#66d9ef">else</span> Version(<span style="color:#e6db74">&#34;0&#34;</span>)
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lessThan&#34;</span> <span style="color:#f92672">in</span> r:
</span></span><span style="display:flex;"><span>                inside <span style="color:#f92672">=</span> lo <span style="color:#f92672">&lt;=</span> v <span style="color:#f92672">&lt;</span> Version(r[<span style="color:#e6db74">&#34;lessThan&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">elif</span> <span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span> <span style="color:#f92672">in</span> r:
</span></span><span style="display:flex;"><span>                inside <span style="color:#f92672">=</span> lo <span style="color:#f92672">&lt;=</span> v <span style="color:#f92672">&lt;=</span> Version(r[<span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>                inside <span style="color:#f92672">=</span> v <span style="color:#f92672">==</span> lo
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">except</span> InvalidVersion:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;review: unparseable range&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> inside:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> r[<span style="color:#e6db74">&#34;status&#34;</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> entry<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;defaultStatus&#34;</span>) <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;unknown&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>(index_path, sbom_path):
</span></span><span style="display:flex;"><span>    index <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> line <span style="color:#f92672">in</span> open(index_path):
</span></span><span style="display:flex;"><span>        e <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>loads(line)
</span></span><span style="display:flex;"><span>        index<span style="color:#f92672">.</span>setdefault(e[<span style="color:#e6db74">&#34;key&#34;</span>], [])<span style="color:#f92672">.</span>append(e)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> json<span style="color:#f92672">.</span>load(open(sbom_path))[<span style="color:#e6db74">&#34;components&#34;</span>]:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> index<span style="color:#f92672">.</span>get(norm(c[<span style="color:#e6db74">&#34;purl&#34;</span>]), []):
</span></span><span style="display:flex;"><span>            print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#39;</span><span style="color:#e6db74">{</span>c[<span style="color:#e6db74">&#34;purl&#34;</span>]<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>e[<span style="color:#e6db74">&#34;cve&#34;</span>]<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>status_for(c[<span style="color:#e6db74">&#34;version&#34;</span>], e)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#39;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span></code></pre></div><h3 id="test-run">Test run</h3>
<p><code>python3 build_index.py cves.zip purl-index.jsonl</code> took about 30 seconds and wrote 7,324 lines, one per entry. We then ran <code>python3 join_sbom.py purl-index.jsonl sbom.cdx.json</code> (packaging 26.3) with a three-component CycloneDX 1.6 SBOM:</p>
<table>
  <thead>
      <tr>
          <th>SBOM purl</th>
          <th>Records matched</th>
          <th>Affected</th>
          <th>Unaffected</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>pkg:npm/fast-uri@2.4.5</code></td>
          <td>17</td>
          <td>10</td>
          <td>7</td>
      </tr>
      <tr>
          <td><code>pkg:pypi/GitPython@3.1.62</code></td>
          <td>34</td>
          <td>6</td>
          <td>28</td>
      </tr>
      <tr>
          <td><code>pkg:maven/org.bouncycastle/bcprov-jdk18on@1.86?type=jar</code></td>
          <td>23</td>
          <td>0</td>
          <td>23</td>
      </tr>
  </tbody>
</table>
<p>GitPython 3.1.62 comes out affected by the six records whose ranges end at <code>lessThan: 3.2.0</code>, and unaffected by the 28 whose fixes landed at or before 3.1.62. Without normalization, the SBOM spelling <code>GitPython</code> would have matched 6 records and missed 28. The <code>?type=jar</code> qualifier on the Maven purl is dropped before lookup.</p>
<p>What this does not handle: <code>versionType</code> values such as <code>custom</code>, <code>git</code> or distribution schemes (it flags them for review), <code>changes</code> arrays inside version ranges, purl types that need a repository-to-package mapping, and the order sensitivity of overlapping ranges. The <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">CVE affected versions guide</a> covers those rules in detail.</p>
<h2 id="when-the-cve-has-no-purl">When the CVE has no purl</h2>
<p>For the other 98% of records, you still need the older paths. The <a href="/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/">cpeApplicability explainer</a> covers CPE-based matching, and <code>collectionURL</code> plus <code>packageName</code> is often enough to build a purl yourself: <code>https://www.npmjs.com/package/x</code> with <code>packageName</code> <code>x</code> is <code>pkg:npm/x</code>. Record which path produced each match, because a purl match and a name match deserve different confidence.</p>
<p>A match is only the start of the patch decision. In our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>, the purl join is the &ldquo;does this apply to us&rdquo; step; KEV, EPSS and SSVC decide how fast. If your supplier also ships VEX, apply it after the join; our <a href="/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/">CSAF VEX guide</a> shows how to resolve its product IDs to purls.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The counts describe one snapshot (the 5 November 2026 midnight bundle); CNAs update records, so the numbers will move. Month counts use each record&rsquo;s original <code>datePublished</code>, so a purl added later to an old record is counted in the month the record was first published. We checked purl spelling only for the <code>github</code> and <code>pypi</code> rules; other types have their own normalization rules we did not test. We did not check how CVE Services validates purls at submission time. The join script uses Python&rsquo;s <code>packaging</code> version ordering, which is close to, but not the same as, every ecosystem&rsquo;s own ordering.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>packageURL definition, &ldquo;MUST NOT include a version&rdquo;, examples; identifier and versions/defaultStatus requirements</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema (main)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Rationale, CVE Services validation, versions only in versions field</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/rfds/0002-supporting-package-urls.md">RFD 0002: Supporting Package URLs in CVE</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>v5.2.0 tag commit dated 3 September 2025</td>
          <td><a href="https://github.com/CVEProject/cve-schema/releases/tag/v5.2.0">cve-schema tag v5.2.0</a></td>
          <td>Qualified (commit date, not an announcement)</td>
      </tr>
      <tr>
          <td>All counts (records, CNAs, types, normalization issues, monthly figures)</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/releases/tag/cve_2026-11-05_0400Z">cvelistV5 release cve_2026-11-05_0400Z</a>, counted with Python</td>
          <td>Verified (our count)</td>
      </tr>
      <tr>
          <td>CVE-2026-110180 affected fragment</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/110xxx/CVE-2026-110180.json">CVE-2026-110180 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>GitHub and PyPI normalization rules; npm case sensitivity</td>
          <td><a href="https://github.com/package-url/purl-spec/tree/main/types">purl-spec type definitions</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>purl type is case-insensitive, lowercase canonical form</td>
          <td><a href="https://github.com/package-url/purl-spec/blob/main/docs/specification/standard/Clause-5-Package-URL-Specification.md">purl-spec, Clause 5</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>software-id type decision status &ldquo;proposed&rdquo;, dated 21 October 2026</td>
          <td><a href="https://github.com/package-url/purl-spec/blob/main/docs/decisions/001-Register-software-id-PURL-type.md">purl-spec decision 001</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CSAF VEX product_tree: Resolve Product IDs to SBOM purls</title><link>https://firewallsync.com/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/</link><guid>https://firewallsync.com/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/</guid><pubDate>Thu, 05 Nov 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>A CSAF VEX file never puts a purl next to a status. Product IDs in product_status point into the product_tree, often through relationships. Here is how to walk branches and relationships, collapse the eight status lists into four VEX answers, and join the result to a CycloneDX SBOM, with a tested Python script.</description><enclosure url="https://firewallsync.com/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/cover_hu13121586183071927737.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/cover_hu13121586183071927737.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/csaf-vex-product-tree-product-status-purl-sbom-matching/cover_hu13121586183071927737.jpg" alt="Thin bare tree branches crossing in a dense tangle against a flat grey-blue sky" width="1200" height="630"></p><p><strong>Scope.</strong> This is a field-level guide for engineers who consume vendor CSAF VEX files and need to apply them to their own SBOMs. We read the CSAF 2.0 prose and JSON schema in the OASIS CSAF TC&rsquo;s GitHub repository, Red Hat&rsquo;s July 2024 announcement of its CSAF VEX files, and Trivy&rsquo;s CSAF matcher and documentation at tag v0.76.1. The example document and the Python script below are ours; we validated the document against the CSAF 2.0 JSON schema and ran the script with Python 3.13. The script is illustrative, not a production matcher.</p>
<h2 id="the-short-answer">The short answer</h2>
<p>A CSAF VEX file never puts a purl next to a status. It says &ldquo;product ID <code>GW-4.2.1:LIB-1.8.0</code> is <code>known_not_affected</code>&rdquo;, and you have to work out what that ID is:</p>
<ol>
<li>Find the ID in the <code>product_tree</code>: at the end of a <code>branches</code> path, in <code>full_product_names</code>, or as the <code>full_product_name</code> of a <code>relationships</code> entry.</li>
<li>If it is a relationship, resolve both sides: <code>product_reference</code> (the component) and <code>relates_to_product_reference</code> (the product it sits in).</li>
<li>Read <code>product_identification_helper.purl</code> (or <code>cpe</code>) for each side.</li>
<li>Apply the status only when your SBOM contains the component <strong>and</strong> the parent matches the product you are assessing.</li>
<li>Collapse the eight <code>product_status</code> lists to the four VEX answers before you write anything to a ticket.</li>
</ol>
<p>The rest of this article goes through each step, shows a schema-valid example, and gives a script that does the join.</p>
<h2 id="what-the-vex-profile-requires">What the VEX profile requires</h2>
<p>CSAF 2.0 has been an OASIS Standard since 18 November 2022. The VEX profile is section 4.5. A document satisfies it when <code>/document/category</code> is <code>csaf_vex</code> and it contains:</p>
<ul>
<li>a <code>product_tree</code> &ldquo;which lists all products referenced later on in the CSAF document regardless of their state&rdquo;;</li>
<li><code>vulnerabilities</code>, each with <code>notes</code> and at least one of <code>cve</code> or <code>ids</code>;</li>
<li>at least one of the <code>fixed</code>, <code>known_affected</code>, <code>known_not_affected</code> or <code>under_investigation</code> lists.</li>
</ul>
<p>Two conditions are per product, and they are the ones consumers should check:</p>
<ul>
<li>Every product in <code>known_not_affected</code> needs an impact statement, either &ldquo;as machine readable flag in <code>/vulnerabilities[]/flags</code> or as human readable justification in <code>/vulnerabilities[]/threats</code>&rdquo; with <code>category</code> set to <code>impact</code>.</li>
<li>Every product in <code>known_affected</code> needs &ldquo;additional product specific information&rdquo; in <code>remediations</code> &ldquo;as an action statement&rdquo;. The categories <code>no_fix_planned</code> and <code>none_available</code> count.</li>
</ul>
<p>The five flag labels are the CISA justification codes: <code>component_not_present</code>, <code>vulnerable_code_not_present</code>, <code>vulnerable_code_cannot_be_controlled_by_adversary</code>, <code>vulnerable_code_not_in_execute_path</code> and <code>inline_mitigations_already_exist</code>. If you need the meaning of each one, the <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explainer</a> covers them. The <a href="/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/">CycloneDX VEX mapping guide</a> covers converting between CSAF and other VEX formats. This article is about something they do not cover: finding the products a CSAF statement actually refers to.</p>
<h2 id="step-1-walk-the-product-tree">Step 1: walk the product tree</h2>
<p>The <code>product_tree</code> can name products in three places. A consumer has to read all three.</p>
<table>
  <thead>
      <tr>
          <th>Where</th>
          <th>What it holds</th>
          <th>Product ID comes from</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>branches</code> (nested)</td>
          <td>A hierarchy such as <code>vendor</code> → <code>product_name</code> → <code>product_version</code>. A leaf branch has a <code>product</code> object.</td>
          <td><code>branches[]...product.product_id</code></td>
      </tr>
      <tr>
          <td><code>full_product_names</code></td>
          <td>Flat list of products with no hierarchy</td>
          <td><code>full_product_names[].product_id</code></td>
      </tr>
      <tr>
          <td><code>relationships</code></td>
          <td>A new product made of two existing ones, such as a library inside an image</td>
          <td><code>relationships[].full_product_name.product_id</code></td>
      </tr>
  </tbody>
</table>
<p>Every one of these is a &ldquo;full product name&rdquo;: <code>name</code> and <code>product_id</code> are required, and <code>product_identification_helper</code> is optional. That helper is where <code>purl</code>, <code>cpe</code>, hashes and SBOM URLs go. The CSAF schema only checks the purl against a loose pattern (<code>^pkg:[A-Za-z\.\-\+][A-Za-z0-9\.\-\+]*/.+</code>) and says it &ldquo;does not completely evaluate whether a PURL is valid&rdquo;, so validate purls yourself before matching on them.</p>
<p>There is also <code>product_groups</code>, which bundles product IDs under a group ID. Product status lists only hold product IDs, but <code>flags</code>, <code>threats</code> and <code>remediations</code> can reference groups through <code>group_ids</code>, so expand groups when you collect impact and action statements.</p>
<p>The spec recommends the <code>vendor</code> → <code>product_name</code> → <code>product_version</code> shape &ldquo;whenever possible to support the identification and matching of products on the consumer side&rdquo;. It also allows <code>product_version_range</code> branches, but recommends enumerating versions instead. If you meet a range branch, you need a version-range parser for whatever scheme it uses; the purl join below will not help.</p>
<h2 id="step-2-treat-relationships-as-component-in-product">Step 2: treat relationships as &ldquo;component in product&rdquo;</h2>
<p>A relationship has four required fields: <code>category</code>, <code>product_reference</code>, <code>relates_to_product_reference</code> and <code>full_product_name</code>. The categories are <code>default_component_of</code>, <code>external_component_of</code>, <code>installed_on</code>, <code>installed_with</code> and <code>optional_component_of</code>. The spec gives the reason relationships exist: a product may be &ldquo;vulnerable only when installed together with another&rdquo;, or you may need &ldquo;to describe operating system components&rdquo;.</p>
<p>This is the part consumers get wrong. In our example below, <code>LIB-1.8.0</code> is the library on its own and <code>GW-4.2.1:LIB-1.8.0</code> is &ldquo;libexample 1.8.0 as shipped in Example Gateway 4.2.1&rdquo;. A vendor that marks the second ID <code>known_not_affected</code> is saying the gateway does not reach the vulnerable code. It is not saying libexample 1.8.0 is safe everywhere. If your matcher resolves the relationship ID to the component purl alone, you will suppress the CVE for every other image that ships libexample 1.8.0. That is the false negative VEX is supposed to prevent.</p>
<p>The rule we use: for a relationship ID, apply the status only when the component purl matches a component in the SBOM <strong>and</strong> the parent purl matches the SBOM&rsquo;s subject (for CycloneDX, <code>metadata.component</code>).</p>
<p>Red Hat&rsquo;s VEX files are a real case of this structure. Red Hat publishes one CSAF VEX file per CVE. Its July 2024 announcement says each file has a <code>product_tree</code> object listing &ldquo;products, their components, and the relationships between them&rdquo;. In the general-availability release Red Hat &ldquo;extended the scope of the <code>purl</code> identifiers to unfixed components as well&rdquo; for rpm, oci and rpmmod components. In the October 2023 beta, only fixed components carried a purl. Check which one you are parsing.</p>
<h2 id="step-3-collapse-eight-status-lists-into-four-answers">Step 3: collapse eight status lists into four answers</h2>
<p>CSAF defines eight <code>product_status</code> lists. VEX consumers need four answers. The spec&rsquo;s own test for contradicting statuses (section 6.1.6) gives the grouping, and the sets &ldquo;MUST be pairwise disjoint&rdquo; within one vulnerability:</p>
<table>
  <thead>
      <tr>
          <th>VEX answer</th>
          <th>CSAF <code>product_status</code> lists</th>
          <th>What to do</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Affected</td>
          <td><code>first_affected</code>, <code>known_affected</code>, <code>last_affected</code></td>
          <td>Open or keep the ticket. Read <code>remediations</code> for the action statement.</td>
      </tr>
      <tr>
          <td>Not affected</td>
          <td><code>known_not_affected</code></td>
          <td>Suppress for that product only, and keep the flag or impact text as evidence.</td>
      </tr>
      <tr>
          <td>Fixed</td>
          <td><code>first_fixed</code>, <code>fixed</code></td>
          <td>Close if you run that version. The spec says these versions &ldquo;may not be the recommended fixed versions&rdquo;.</td>
      </tr>
      <tr>
          <td>Under investigation</td>
          <td><code>under_investigation</code></td>
          <td>Keep the scanner finding and re-pull the document later.</td>
      </tr>
      <tr>
          <td>(none)</td>
          <td><code>recommended</code></td>
          <td>Not part of any contradiction group. The spec notes an issuer might recommend a version &ldquo;from any group - also from the affected group&rdquo;. Use it to choose an upgrade target, not to decide status.</td>
      </tr>
  </tbody>
</table>
<p>If you find one product ID in two groups for the same CVE, the document fails the CSAF mandatory test. Reject the statement for that product rather than picking one.</p>
<p>For affected products, the remediation categories tell you what kind of action is available: <code>vendor_fix</code>, <code>workaround</code>, <code>mitigation</code>, <code>none_available</code> or <code>no_fix_planned</code>. <code>vendor_fix</code> cannot be combined with <code>none_available</code> or <code>no_fix_planned</code> for the same product.</p>
<h2 id="a-schema-valid-example">A schema-valid example</h2>
<p>This document is ours, written for this article. It has one product version where a library is not reachable and a later version where it was updated. It passes validation against the CSAF 2.0 JSON schema from the OASIS repository (Python <code>jsonschema</code>, Draft 2020-12, format checks on; 0 errors). Schema validation does not run CSAF&rsquo;s mandatory tests, so it does not prove the profile rules are met. We checked those by hand.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;document&#34;</span>: {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;csaf_vex&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;csaf_version&#34;</span>: <span style="color:#e6db74">&#34;2.0&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;publisher&#34;</span>: {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;vendor&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;Example Vendor&#34;</span>, <span style="color:#f92672">&#34;namespace&#34;</span>: <span style="color:#e6db74">&#34;https://vendor.example&#34;</span>},
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;title&#34;</span>: <span style="color:#e6db74">&#34;Example Gateway: CVE-2099-0001 in libexample&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;tracking&#34;</span>: {<span style="color:#f92672">&#34;id&#34;</span>: <span style="color:#e6db74">&#34;EXV-2099-0001&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;final&#34;</span>, <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;2&#34;</span>, <span style="color:#f92672">&#34;...&#34;</span>: <span style="color:#e6db74">&#34;...&#34;</span>}
</span></span><span style="display:flex;"><span>  },
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;product_tree&#34;</span>: {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;branches&#34;</span>: [{<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;vendor&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;Example Vendor&#34;</span>, <span style="color:#f92672">&#34;branches&#34;</span>: [
</span></span><span style="display:flex;"><span>      {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;product_name&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;Example Gateway&#34;</span>, <span style="color:#f92672">&#34;branches&#34;</span>: [
</span></span><span style="display:flex;"><span>        {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;product_version&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;4.2.1&#34;</span>,
</span></span><span style="display:flex;"><span>         <span style="color:#f92672">&#34;product&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;Example Gateway 4.2.1&#34;</span>, <span style="color:#f92672">&#34;product_id&#34;</span>: <span style="color:#e6db74">&#34;GW-4.2.1&#34;</span>,
</span></span><span style="display:flex;"><span>           <span style="color:#f92672">&#34;product_identification_helper&#34;</span>: {<span style="color:#f92672">&#34;purl&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/example-gateway@sha256%3A1111?tag=4.2.1&#34;</span>}}}
</span></span><span style="display:flex;"><span>      ]},
</span></span><span style="display:flex;"><span>      {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;product_name&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;libexample&#34;</span>, <span style="color:#f92672">&#34;branches&#34;</span>: [
</span></span><span style="display:flex;"><span>        {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;product_version&#34;</span>, <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;1.8.0&#34;</span>,
</span></span><span style="display:flex;"><span>         <span style="color:#f92672">&#34;product&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;libexample 1.8.0&#34;</span>, <span style="color:#f92672">&#34;product_id&#34;</span>: <span style="color:#e6db74">&#34;LIB-1.8.0&#34;</span>,
</span></span><span style="display:flex;"><span>           <span style="color:#f92672">&#34;product_identification_helper&#34;</span>: {<span style="color:#f92672">&#34;purl&#34;</span>: <span style="color:#e6db74">&#34;pkg:deb/debian/libexample@1.8.0-1&#34;</span>}}}
</span></span><span style="display:flex;"><span>      ]}
</span></span><span style="display:flex;"><span>    ]}],
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;relationships&#34;</span>: [
</span></span><span style="display:flex;"><span>      {<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;default_component_of&#34;</span>,
</span></span><span style="display:flex;"><span>       <span style="color:#f92672">&#34;product_reference&#34;</span>: <span style="color:#e6db74">&#34;LIB-1.8.0&#34;</span>,
</span></span><span style="display:flex;"><span>       <span style="color:#f92672">&#34;relates_to_product_reference&#34;</span>: <span style="color:#e6db74">&#34;GW-4.2.1&#34;</span>,
</span></span><span style="display:flex;"><span>       <span style="color:#f92672">&#34;full_product_name&#34;</span>: {<span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;libexample 1.8.0 in Example Gateway 4.2.1&#34;</span>,
</span></span><span style="display:flex;"><span>                             <span style="color:#f92672">&#34;product_id&#34;</span>: <span style="color:#e6db74">&#34;GW-4.2.1:LIB-1.8.0&#34;</span>}}
</span></span><span style="display:flex;"><span>    ]
</span></span><span style="display:flex;"><span>  },
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vulnerabilities&#34;</span>: [{
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;cve&#34;</span>: <span style="color:#e6db74">&#34;CVE-2099-0001&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;notes&#34;</span>: [{<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;description&#34;</span>, <span style="color:#f92672">&#34;text&#34;</span>: <span style="color:#e6db74">&#34;Illustrative vulnerability in libexample&#39;s XML parser.&#34;</span>}],
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;product_status&#34;</span>: {<span style="color:#f92672">&#34;known_not_affected&#34;</span>: [<span style="color:#e6db74">&#34;GW-4.2.1:LIB-1.8.0&#34;</span>], <span style="color:#f92672">&#34;fixed&#34;</span>: [<span style="color:#e6db74">&#34;GW-4.3.0:LIB-1.8.2&#34;</span>]},
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;flags&#34;</span>: [{<span style="color:#f92672">&#34;label&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span>, <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;GW-4.2.1:LIB-1.8.0&#34;</span>]}],
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;threats&#34;</span>: [{<span style="color:#f92672">&#34;category&#34;</span>: <span style="color:#e6db74">&#34;impact&#34;</span>, <span style="color:#f92672">&#34;details&#34;</span>: <span style="color:#e6db74">&#34;Example Gateway never calls the XML parser in libexample.&#34;</span>,
</span></span><span style="display:flex;"><span>                 <span style="color:#f92672">&#34;product_ids&#34;</span>: [<span style="color:#e6db74">&#34;GW-4.2.1:LIB-1.8.0&#34;</span>]}]
</span></span><span style="display:flex;"><span>  }]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>The excerpt trims the tracking block and the 4.3.0 / 1.8.2 branches and relationship, which follow the same pattern. The CVE ID is a placeholder.</p>
<h2 id="step-4-join-to-the-sbom">Step 4: join to the SBOM</h2>
<p>The script resolves every product ID in <code>product_status</code> to purls, follows relationships, and checks both sides against a CycloneDX SBOM. For purl comparison it uses the matching rules Trivy documents, which Trivy itself notes &ldquo;are not formally defined within the current official PURL specification&rdquo;: a VEX purl without a version matches every version; every qualifier in the VEX purl must be present in the SBOM purl; extra SBOM qualifiers are ignored.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Resolve CSAF VEX product_status entries to purls and join them to a CycloneDX SBOM.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Illustrative: purl matching follows the rules in purl_match(); no version-range logic.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> urllib.parse <span style="color:#f92672">import</span> unquote
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">parse</span>(p):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Split a purl into (type/namespace/name, version or None, qualifiers dict). Subpath ignored.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    p <span style="color:#f92672">=</span> p<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;#&#34;</span>)[<span style="color:#ae81ff">0</span>]
</span></span><span style="display:flex;"><span>    core, _, q <span style="color:#f92672">=</span> p<span style="color:#f92672">.</span>partition(<span style="color:#e6db74">&#34;?&#34;</span>)
</span></span><span style="display:flex;"><span>    quals <span style="color:#f92672">=</span> dict(kv<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;=&#34;</span>, <span style="color:#ae81ff">1</span>) <span style="color:#66d9ef">for</span> kv <span style="color:#f92672">in</span> q<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;&amp;&#34;</span>) <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;=&#34;</span> <span style="color:#f92672">in</span> kv)
</span></span><span style="display:flex;"><span>    path, _, ver <span style="color:#f92672">=</span> core<span style="color:#f92672">.</span>rpartition(<span style="color:#e6db74">&#34;@&#34;</span>) <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;@&#34;</span> <span style="color:#f92672">in</span> core<span style="color:#f92672">.</span>rsplit(<span style="color:#e6db74">&#34;/&#34;</span>, <span style="color:#ae81ff">1</span>)[<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>] <span style="color:#66d9ef">else</span> (core, <span style="color:#e6db74">&#34;&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> unquote(path)<span style="color:#f92672">.</span>lower(), unquote(ver) <span style="color:#f92672">or</span> <span style="color:#66d9ef">None</span>, {k<span style="color:#f92672">.</span>lower(): unquote(v) <span style="color:#66d9ef">for</span> k, v <span style="color:#f92672">in</span> quals<span style="color:#f92672">.</span>items()}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">purl_match</span>(vex_purl, sbom_purl):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Community matching rules (as documented by Trivy): a VEX purl without a version matches
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">    every version; VEX qualifiers must all be present in the SBOM purl; extra SBOM qualifiers are ignored.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    vp, vv, vq <span style="color:#f92672">=</span> parse(vex_purl)
</span></span><span style="display:flex;"><span>    sp, sv, sq <span style="color:#f92672">=</span> parse(sbom_purl)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> vp <span style="color:#f92672">==</span> sp <span style="color:#f92672">and</span> (vv <span style="color:#f92672">is</span> <span style="color:#66d9ef">None</span> <span style="color:#f92672">or</span> vv <span style="color:#f92672">==</span> sv) <span style="color:#f92672">and</span> all(sq<span style="color:#f92672">.</span>get(k) <span style="color:#f92672">==</span> v <span style="color:#66d9ef">for</span> k, v <span style="color:#f92672">in</span> vq<span style="color:#f92672">.</span>items())
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">walk_branches</span>(branches, out):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> b <span style="color:#f92672">in</span> branches <span style="color:#f92672">or</span> []:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;product&#34;</span> <span style="color:#f92672">in</span> b:
</span></span><span style="display:flex;"><span>            fpn <span style="color:#f92672">=</span> b[<span style="color:#e6db74">&#34;product&#34;</span>]
</span></span><span style="display:flex;"><span>            out[fpn[<span style="color:#e6db74">&#34;product_id&#34;</span>]] <span style="color:#f92672">=</span> fpn<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_identification_helper&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;purl&#34;</span>)
</span></span><span style="display:flex;"><span>        walk_branches(b<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;branches&#34;</span>), out)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_products</span>(doc):
</span></span><span style="display:flex;"><span>    tree <span style="color:#f92672">=</span> doc[<span style="color:#e6db74">&#34;product_tree&#34;</span>]
</span></span><span style="display:flex;"><span>    purls <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    walk_branches(tree<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;branches&#34;</span>), purls)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> fpn <span style="color:#f92672">in</span> tree<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;full_product_names&#34;</span>, []):
</span></span><span style="display:flex;"><span>        purls[fpn[<span style="color:#e6db74">&#34;product_id&#34;</span>]] <span style="color:#f92672">=</span> fpn<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_identification_helper&#34;</span>, {})<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;purl&#34;</span>)
</span></span><span style="display:flex;"><span>    rels <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> tree<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;relationships&#34;</span>, []):
</span></span><span style="display:flex;"><span>        rels[r[<span style="color:#e6db74">&#34;full_product_name&#34;</span>][<span style="color:#e6db74">&#34;product_id&#34;</span>]] <span style="color:#f92672">=</span> (
</span></span><span style="display:flex;"><span>            r[<span style="color:#e6db74">&#34;category&#34;</span>], r[<span style="color:#e6db74">&#34;product_reference&#34;</span>], r[<span style="color:#e6db74">&#34;relates_to_product_reference&#34;</span>])
</span></span><span style="display:flex;"><span>    groups <span style="color:#f92672">=</span> {g[<span style="color:#e6db74">&#34;group_id&#34;</span>]: g[<span style="color:#e6db74">&#34;product_ids&#34;</span>] <span style="color:#66d9ef">for</span> g <span style="color:#f92672">in</span> tree<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_groups&#34;</span>, [])}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> purls, rels, groups
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>(vex_path, sbom_path):
</span></span><span style="display:flex;"><span>    doc <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(vex_path))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">assert</span> doc[<span style="color:#e6db74">&#34;document&#34;</span>][<span style="color:#e6db74">&#34;category&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;csaf_vex&#34;</span>, <span style="color:#e6db74">&#34;not a VEX profile document&#34;</span>
</span></span><span style="display:flex;"><span>    purls, rels, groups <span style="color:#f92672">=</span> load_products(doc)
</span></span><span style="display:flex;"><span>    sbom <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sbom_path))
</span></span><span style="display:flex;"><span>    parent <span style="color:#f92672">=</span> sbom[<span style="color:#e6db74">&#34;metadata&#34;</span>][<span style="color:#e6db74">&#34;component&#34;</span>][<span style="color:#e6db74">&#34;purl&#34;</span>]
</span></span><span style="display:flex;"><span>    inventory <span style="color:#f92672">=</span> [c[<span style="color:#e6db74">&#34;purl&#34;</span>] <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> sbom<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;components&#34;</span>, []) <span style="color:#66d9ef">if</span> c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;purl&#34;</span>)]
</span></span><span style="display:flex;"><span>    in_sbom <span style="color:#f92672">=</span> <span style="color:#66d9ef">lambda</span> vex_purl: any(purl_match(vex_purl, s) <span style="color:#66d9ef">for</span> s <span style="color:#f92672">in</span> inventory)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> doc[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>        flags <span style="color:#f92672">=</span> {pid: f[<span style="color:#e6db74">&#34;label&#34;</span>] <span style="color:#66d9ef">for</span> f <span style="color:#f92672">in</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;flags&#34;</span>, [])
</span></span><span style="display:flex;"><span>                 <span style="color:#66d9ef">for</span> pid <span style="color:#f92672">in</span> f<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_ids&#34;</span>, []) <span style="color:#f92672">+</span> sum((groups[g] <span style="color:#66d9ef">for</span> g <span style="color:#f92672">in</span> f<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;group_ids&#34;</span>, [])), [])}
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> status, pids <span style="color:#f92672">in</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product_status&#34;</span>, {})<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> pid <span style="color:#f92672">in</span> pids:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> pid <span style="color:#f92672">in</span> rels:                      <span style="color:#75715e"># component-in-product: match BOTH sides</span>
</span></span><span style="display:flex;"><span>                    cat, comp, prod <span style="color:#f92672">=</span> rels[pid]
</span></span><span style="display:flex;"><span>                    comp_p, prod_p <span style="color:#f92672">=</span> purls<span style="color:#f92672">.</span>get(comp), purls<span style="color:#f92672">.</span>get(prod)
</span></span><span style="display:flex;"><span>                    hit <span style="color:#f92672">=</span> bool(comp_p <span style="color:#f92672">and</span> prod_p <span style="color:#f92672">and</span> purl_match(prod_p, parent) <span style="color:#f92672">and</span> in_sbom(comp_p))
</span></span><span style="display:flex;"><span>                    target <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>comp_p<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>cat<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>prod_p<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">else</span>:                                <span style="color:#75715e"># standalone product</span>
</span></span><span style="display:flex;"><span>                    p <span style="color:#f92672">=</span> purls<span style="color:#f92672">.</span>get(pid)
</span></span><span style="display:flex;"><span>                    hit <span style="color:#f92672">=</span> bool(p <span style="color:#f92672">and</span> (in_sbom(p) <span style="color:#f92672">or</span> purl_match(p, parent)))
</span></span><span style="display:flex;"><span>                    target <span style="color:#f92672">=</span> p
</span></span><span style="display:flex;"><span>                print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;cve&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>status<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>flags<span style="color:#f92672">.</span>get(pid, <span style="color:#e6db74">&#39;-&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span><span style="color:#e6db74">&#39;MATCH&#39;</span> <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#39;no match&#39;</span><span style="color:#e6db74">}</span><span style="color:#ae81ff">\t</span><span style="color:#e6db74">{</span>target<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main(<span style="color:#f92672">*</span>sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:<span style="color:#ae81ff">3</span>])
</span></span></code></pre></div><h3 id="test-run">Test run</h3>
<p>We ran it (<code>python3 -I csaf_vex_join.py vex.json sbom.cdx.json</code>) against the full example document and three small CycloneDX 1.6 SBOMs.</p>
<p><strong>SBOM 1</strong>, image 4.2.1 (<code>metadata.component.purl</code> = <code>pkg:oci/example-gateway@sha256%3A1111?tag=4.2.1</code>), containing <code>pkg:deb/debian/libexample@1.8.0-1?arch=amd64&amp;distro=debian-12</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2099-0001  known_not_affected  vulnerable_code_not_in_execute_path  MATCH     pkg:deb/debian/libexample@1.8.0-1 default_component_of pkg:oci/example-gateway@sha256%3A1111?tag=4.2.1
</span></span><span style="display:flex;"><span>CVE-2099-0001  fixed               -                                    no match  pkg:deb/debian/libexample@1.8.2-1 default_component_of pkg:oci/example-gateway@sha256%3A2222?tag=4.3.0
</span></span></code></pre></div><p>The SBOM&rsquo;s extra <code>arch</code> and <code>distro</code> qualifiers did not block the match, because the VEX purl has none.</p>
<p><strong>SBOM 2</strong>, image 4.3.0 with libexample 1.8.2: the results flip. The <code>fixed</code> statement matches and the <code>known_not_affected</code> one does not.</p>
<p><strong>SBOM 3</strong>, the 4.2.1 SBOM again, but with the VEX component purl changed to <code>...libexample@1.8.0-1?arch=arm64</code>: <code>no match</code>, because the VEX purl now requires a qualifier value the SBOM lacks.</p>
<p>What the script does not do: version ranges (<code>product_version_range</code>), CPE-only products, hashes, <code>product_groups</code> in status lists (the schema does not allow them there), or the mandatory-test checks. It also prints matches and nothing else; it does not decide what to do with a CVE your scanner found but the VEX file does not mention.</p>
<h2 id="how-trivy-applies-csaf-vex">How Trivy applies CSAF VEX</h2>
<p>Trivy is a common place engineers first meet CSAF VEX. At tag v0.76.1, its documentation says CSAF works for container images, filesystems, code repositories, VM images, Kubernetes and SBOM targets, passed with <code>--vex</code>. Its matcher (<code>pkg/vex/csaf.go</code>) shows three things worth knowing:</p>
<ul>
<li><strong>Only two statuses act.</strong> The status map holds <code>known_not_affected</code> and <code>fixed</code>. A match on either filters the finding. <code>known_affected</code> and <code>under_investigation</code> statements never change the result.</li>
<li><strong>Matching is by purl.</strong> If the detected package has no purl, or the vulnerability has no <code>product_status</code>, nothing is filtered.</li>
<li><strong>Relationship categories are not distinguished.</strong> The documentation says &ldquo;the specified relationship category is not taken into account&rdquo; and that <code>default_component_of</code>, <code>installed_on</code> and <code>installed_with</code> &ldquo;are treated internally as <code>depends_on</code>&rdquo;. For a relationship, the code requires the parent purl to match a product higher in the dependency tree (such as the scanned image) and the component purl to match the sub-component, which is the both-sides rule from Step 2. <code>external_component_of</code> and <code>optional_component_of</code> relationships are skipped. A status on a plain product ID, with no relationship, is matched against the package&rsquo;s own purl, so it applies wherever that package appears.</li>
</ul>
<p>So if a vendor&rsquo;s CSAF file uses <code>optional_component_of</code> for a component, Trivy will not apply that status through the relationship. That is safe in direction (no false suppression), but it can surprise you if you expect the finding to drop.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Matching the component alone.</strong> Covered above. It turns a statement about one product into a statement about all of them.</li>
<li><strong>Trusting <code>known_not_affected</code> with no impact statement.</strong> The profile requires a flag or an impact threat for each such product. If neither exists, the document does not satisfy the profile; send it back to the vendor rather than suppressing.</li>
<li><strong>Reading only <code>branches</code>.</strong> Products in <code>full_product_names</code> and relationships carry IDs too. A parser that only walks branches will report &ldquo;unknown product&rdquo; for most relationship-heavy files.</li>
<li><strong>Treating <code>recommended</code> as fixed.</strong> The spec allows recommending a version from the affected group.</li>
<li><strong>Ignoring <code>tracking.version</code> and <code>current_release_date</code>.</strong> VEX files are revised as investigations finish. Store the version you applied, and re-apply when it changes.</li>
</ul>
<h2 id="where-this-fits">Where this fits</h2>
<p>VEX is one input to a patch decision, not the decision. In our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>, a supplier VEX statement is the step that can close a ticket before KEV, EPSS or SSVC are weighed. For the SBOM side of the join, the <a href="/posts/sbom-minimum-elements-2026-what-cisa-and-partners-changed-from-the-2021-ntia-baseline/">SBOM minimum elements guide</a> covers which identifiers your SBOM should carry in the first place. Without purls in the SBOM, none of this matching works.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We read CSAF 2.0 from the OASIS TC&rsquo;s GitHub repository (the editor&rsquo;s draft prose, headed &ldquo;OASIS Standard, 18 November 2022&rdquo;, and the JSON schema) rather than the copy hosted on docs.oasis-open.org. CSAF 2.1 is still a draft in that repository and may change field names or add statuses. We did not download a Red Hat VEX file; Red Hat&rsquo;s structure is described from its own announcements. The purl matching rules are a community convention, not part of the purl standard. Our test document is synthetic.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CSAF 2.0 is an OASIS Standard dated 18 November 2022; VEX profile requirements (section 4.5)</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/prose/csaf-v2-editor-draft.md">CSAF 2.0 prose, OASIS TC repository</a></td>
          <td>Qualified (TC repository copy)</td>
      </tr>
      <tr>
          <td>Eight product_status lists and their definitions; flag labels; remediation categories; relationship categories</td>
          <td>Same document, sections 3.2.3.9, 3.2.3.5, 3.2.3.12.1, 3.2.2.4</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Contradiction groups must be pairwise disjoint; recommended may come from any group</td>
          <td>Same document, section 6.1.6</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>purl helper pattern and &ldquo;does not completely evaluate&rdquo; note</td>
          <td>Same document, section 3.1.3.3.4</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Example document is schema-valid</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/json_schema/csaf_json_schema.json">CSAF 2.0 JSON schema</a>, validated with Python jsonschema</td>
          <td>Verified (our test)</td>
      </tr>
      <tr>
          <td>Red Hat VEX files: per CVE, product_tree with relationships, purls extended to unfixed rpm/oci/rpmmod components at GA</td>
          <td><a href="https://www.redhat.com/en/blog/red-hat-vex-files-cves-are-now-generally-available">Red Hat blog, 10 July 2024</a>; <a href="https://www.redhat.com/en/blog/vulnerability-exploitability-exchange-vex-beta-files-now-available">beta announcement, 16 October 2023</a></td>
          <td>Qualified (vendor announcement; GA post read via a localized copy)</td>
      </tr>
      <tr>
          <td>Trivy CSAF matcher uses known_not_affected and fixed; three relationship categories</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.1/pkg/vex/csaf.go">Trivy v0.76.1 pkg/vex/csaf.go</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Trivy CSAF targets, relationship handling and purl matching rules</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/v0.76.1/docs/guide/supply-chain/vex/file.md">Trivy v0.76.1 VEX file documentation</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE cpeApplicability: CPE Match Nodes, Operators, Ranges</title><link>https://firewallsync.com/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/</link><guid>https://firewallsync.com/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/</guid><pubDate>Wed, 04 Nov 2026 04:30:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>Since CVE Record Format 5.1.1, a CNA can put CPE Applicability Language statements in a cpeApplicability block. Here is how nodes, operator, negate, vulnerable and the four version bounds combine, what a 2,000-record sample of 2026 CVEs contains, and a tested Python evaluator that shows where the block and the affected block disagree.</description><enclosure url="https://firewallsync.com/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/cover_hu3781550322919701848.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/cover_hu3781550322919701848.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-cpeapplicability-cpe-match-nodes-operators-version-ranges/cover_hu3781550322919701848.jpg" alt="Top-down view of a dense heap of loose jigsaw puzzle pieces, picture side and grey backs mixed together, in orange, yellow, green and red" width="1200" height="630"></p><p><strong>Scope.</strong> This is a specification explainer with measurements from public data. We read the CVE Record Format schema on the CVEProject/cve-schema main branch (default <code>dataVersion</code> 5.2.0) and its advanced example record, sampled 2,000 random 2026 records from the cvelistV5 repository at commit f0b25a9 (November 4, 2026), and ran the Python script below with Python 3.13 against real records. We did not test commercial scanners or NVD&rsquo;s own matching engine.</p>
<h2 id="the-short-answer">The short answer</h2>
<p><code>cpeApplicability</code> is where a CNA can say, in CPE terms, which configurations a CVE applies to. Each array element is one statement:</p>
<ol>
<li>A statement has <code>nodes</code>, and optionally an <code>operator</code> and <code>negate</code> for combining them.</li>
<li>Each node has a required <code>operator</code> (<code>AND</code> or <code>OR</code>), an optional <code>negate</code>, and a <code>cpeMatch</code> list.</li>
<li>Each <code>cpeMatch</code> has a CPE 2.3 <code>criteria</code> string, a required <code>vulnerable</code> boolean, and optional bounds: <code>versionStartIncluding</code>, <code>versionStartExcluding</code>, <code>versionEndIncluding</code>, <code>versionEndExcluding</code>.</li>
<li>A host is in scope when the logic evaluates true against its CPE inventory <strong>and</strong> at least one matched entry has <code>vulnerable: true</code>.</li>
</ol>
<p>If you already parse NVD configurations, the shape is familiar. The schema says so directly: the block is &ldquo;mostly copied/forked from the NIST NVD CVE API v2.0 schema&rdquo;.</p>
<h2 id="where-the-block-lives">Where the block lives</h2>
<p>The schema defines <code>cpeApplicability</code> as an optional array on both the published CNA container and the ADP container. It was added in CVE Record Format 5.1.1 (the <code>v5.1.1</code> tag in CVEProject/cve-schema dates from October 2024), which the release notes describe as adding &ldquo;support for Common Platform Enumeration (CPE) Identifiers using the CPE Applicability Language&rdquo;.</p>
<p>It is separate from the older <code>cpes</code> array inside each <code>affected[]</code> product. That array is a flat list of CPE names; its own schema description now ends: &ldquo;Consider using the newer cpeApplicability block for defining CPE data using the CPE Applicability Language which includes more options for defining CPE Names.&rdquo;</p>
<p>The schema adds one instruction that matters for consumers: &ldquo;it is recommended that it align with (as much as possible) the product data provided within the affected block.&rdquo; Recommended, not required. If you already read the <code>affected</code> block with the rules in <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">our guide to lessThan, defaultStatus and versionType</a>, treat <code>cpeApplicability</code> as a second encoding of the same claim, and expect the two to disagree sometimes.</p>
<h2 id="the-fields-one-by-one">The fields, one by one</h2>
<table>
  <thead>
      <tr>
          <th>Level</th>
          <th>Field</th>
          <th>Required</th>
          <th>Meaning</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Element</td>
          <td><code>nodes</code></td>
          <td>yes</td>
          <td>Array of nodes</td>
      </tr>
      <tr>
          <td>Element</td>
          <td><code>operator</code></td>
          <td>no</td>
          <td><code>AND</code> or <code>OR</code> between nodes</td>
      </tr>
      <tr>
          <td>Element</td>
          <td><code>negate</code></td>
          <td>no</td>
          <td>Invert the element&rsquo;s result</td>
      </tr>
      <tr>
          <td>Node</td>
          <td><code>operator</code></td>
          <td>yes</td>
          <td><code>AND</code> or <code>OR</code> between the node&rsquo;s cpeMatch entries</td>
      </tr>
      <tr>
          <td>Node</td>
          <td><code>negate</code></td>
          <td>no</td>
          <td>Invert the node&rsquo;s result</td>
      </tr>
      <tr>
          <td>Node</td>
          <td><code>cpeMatch</code></td>
          <td>yes</td>
          <td>Array of match entries</td>
      </tr>
      <tr>
          <td>cpeMatch</td>
          <td><code>criteria</code></td>
          <td>yes</td>
          <td>CPE 2.3 formatted string, e.g. <code>cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*</code></td>
      </tr>
      <tr>
          <td>cpeMatch</td>
          <td><code>vulnerable</code></td>
          <td>yes</td>
          <td><code>true</code> for the vulnerable product, <code>false</code> for a platform or context CPE</td>
      </tr>
      <tr>
          <td>cpeMatch</td>
          <td><code>versionStartIncluding</code> / <code>versionStartExcluding</code></td>
          <td>no</td>
          <td>Lower bound, inclusive or exclusive</td>
      </tr>
      <tr>
          <td>cpeMatch</td>
          <td><code>versionEndIncluding</code> / <code>versionEndExcluding</code></td>
          <td>no</td>
          <td>Upper bound, inclusive or exclusive</td>
      </tr>
      <tr>
          <td>cpeMatch</td>
          <td><code>matchCriteriaId</code></td>
          <td>no</td>
          <td>UUID, carried over from NVD&rsquo;s schema</td>
      </tr>
  </tbody>
</table>
<p>Two details trip people up.</p>
<p><strong>Version bounds apply only when the CPE&rsquo;s version field is a wildcard.</strong> A criteria string such as <code>cpe:2.3:a:ibm:mq_operator:3.4.1:*:...</code> names one exact version; a string with <code>*</code> in the version position plus bounds names a range. Our evaluator treats a specific version in the criteria as an exact match and ignores bounds in that case.</p>
<p><strong>There is no <code>versionType</code>.</strong> The <code>affected</code> block tells you whether versions are semver, git commits, or a custom scheme; <code>cpeMatch</code> does not. Whatever compares <code>6.18.9</code> with <code>6.19</code> has to guess the ordering, and for strings like <code>alpha_0.3</code> or <code>1.3-beta</code> there may be no safe guess.</p>
<h2 id="and-or-and-vulnerable-false">AND, OR and vulnerable: false</h2>
<p>The schema&rsquo;s advanced example record shows the pattern that makes this block more expressive than a flat list. Its single element has <code>&quot;operator&quot;: &quot;AND&quot;</code> and two nodes:</p>
<ul>
<li>Node 1 (<code>OR</code>): <code>example_org:example_enterprise</code> from 1.0.0 up to (excluding) 1.0.6, or from 2.1.6 up to 2.1.9, each with <code>vulnerable: true</code>.</li>
<li>Node 2 (<code>OR</code>): Windows, macOS or a hardware model <code>xt-4500</code>, each with <code>vulnerable: false</code>.</li>
</ul>
<p>Read it as: the application is vulnerable in those ranges <strong>when running on</strong> one of those platforms. The platform CPEs are conditions, not affected products. A tool that flags every host whose inventory contains <code>cpe:2.3:o:microsoft:windows:...</code> has misread the record.</p>
<h2 id="what-a-sample-of-2026-records-contains">What a sample of 2026 records contains</h2>
<p>We took a seeded random sample of 2,000 files from the 84,849 2026 records in cvelistV5 at commit f0b25a9; 1,969 were published. The counting script parsed each record&rsquo;s CNA and ADP containers.</p>
<table>
  <thead>
      <tr>
          <th>Measure</th>
          <th>Count</th>
          <th>Share</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Published records with a CNA <code>cpeApplicability</code> block</td>
          <td>538 of 1,969</td>
          <td>27.3%</td>
      </tr>
      <tr>
          <td>&hellip;of which from Linux, VulnCheck, Microsoft or Oracle (186, 110, 97, 84)</td>
          <td>477 of 538</td>
          <td>88.7%</td>
      </tr>
      <tr>
          <td>ADP containers carrying <code>cpeApplicability</code></td>
          <td>0</td>
          <td>0%</td>
      </tr>
      <tr>
          <td>Records still using the flat <code>affected[].cpes</code> list</td>
          <td>195 of 1,969</td>
          <td>9.9%</td>
      </tr>
      <tr>
          <td>Elements with no element-level <code>operator</code></td>
          <td>489 of 582</td>
          <td>84.0%</td>
      </tr>
      <tr>
          <td>Nodes, all with <code>operator: OR</code></td>
          <td>598</td>
          <td>100%</td>
      </tr>
      <tr>
          <td>cpeMatch entries</td>
          <td>2,667</td>
          <td></td>
      </tr>
      <tr>
          <td>&hellip;with <code>vulnerable: false</code></td>
          <td>10</td>
          <td>0.4%</td>
      </tr>
      <tr>
          <td>&hellip;with a specific version in <code>criteria</code> rather than <code>*</code></td>
          <td>170</td>
          <td>6.4%</td>
      </tr>
      <tr>
          <td>&hellip;with <code>versionStartIncluding</code> / <code>versionEndExcluding</code> / <code>versionEndIncluding</code></td>
          <td>2,397 / 2,333 / 113</td>
          <td>89.9% / 87.5% / 4.2%</td>
      </tr>
      <tr>
          <td>&hellip;with <code>matchCriteriaId</code></td>
          <td>0</td>
          <td>0%</td>
      </tr>
  </tbody>
</table>
<p>Three practical readings. First, most records have no CPE statement from the CNA at all, so a CPE-only matcher misses most of the corpus. Second, the real-world block is simpler than the spec allows: almost everything is one OR node of ranges, and <code>AND</code> with platform CPEs is rare. Third, coverage depends on a handful of CNAs, so your hit rate will swing with your vendor mix. An earlier 1,000-record sample we took for the affected-versions guide found 29.5%; treat the figure as a range, not a constant.</p>
<p>The absence in ADP containers fits CISA&rsquo;s stated policy. The Vulnrichment README says: &ldquo;as of December 10, 2024, CISA will no longer be adding CPE strings to the enriched dataset.&rdquo; See <a href="/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/">our explainer on CVE ADP containers</a> for what CISA-ADP does still add.</p>
<h2 id="a-tested-evaluator">A tested evaluator</h2>
<p>The script takes one CVE record and every CPE installed on one host, and answers <code>True</code>, <code>False</code> or <code>unknown</code>. It is illustrative: it compares only dotted numeric versions, handles the <code>\:</code>-style escaping in CPE 2.3 names only for splitting, and matches part, vendor and product but ignores the remaining CPE fields.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Evaluate a CVE record&#39;s cna.cpeApplicability against one host&#39;s CPE inventory.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Illustrative: numeric-dotted versions only; anything else is reported as &#39;unknown&#39;.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> re<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">split_cpe</span>(cpe):
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># split on &#39;:&#39; not preceded by a backslash (CPE 2.3 formatted-string escaping)</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> re<span style="color:#f92672">.</span>split(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#39;(?&lt;!</span><span style="color:#ae81ff">\\</span><span style="color:#e6db74">):&#39;</span>, cpe)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">vkey</span>(v):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> re<span style="color:#f92672">.</span>fullmatch(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#39;\d+(\.\d+)*&#39;</span>, v):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">raise</span> <span style="color:#a6e22e">ValueError</span>(v)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> tuple(int(x) <span style="color:#66d9ef">for</span> x <span style="color:#f92672">in</span> v<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#39;.&#39;</span>))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cmp</span>(a, b):
</span></span><span style="display:flex;"><span>    a, b <span style="color:#f92672">=</span> vkey(a), vkey(b)
</span></span><span style="display:flex;"><span>    n <span style="color:#f92672">=</span> max(len(a), len(b))
</span></span><span style="display:flex;"><span>    a, b <span style="color:#f92672">=</span> a <span style="color:#f92672">+</span> (<span style="color:#ae81ff">0</span>,) <span style="color:#f92672">*</span> (n <span style="color:#f92672">-</span> len(a)), b <span style="color:#f92672">+</span> (<span style="color:#ae81ff">0</span>,) <span style="color:#f92672">*</span> (n <span style="color:#f92672">-</span> len(b))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (a <span style="color:#f92672">&gt;</span> b) <span style="color:#f92672">-</span> (a <span style="color:#f92672">&lt;</span> b)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">match</span>(m, installed_cpe):
</span></span><span style="display:flex;"><span>    want, have <span style="color:#f92672">=</span> split_cpe(m[<span style="color:#e6db74">&#39;criteria&#39;</span>]), split_cpe(installed_cpe)
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># compare part, vendor, product (fields 2-4); &#39;*&#39; is ANY</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> i <span style="color:#f92672">in</span> (<span style="color:#ae81ff">2</span>, <span style="color:#ae81ff">3</span>, <span style="color:#ae81ff">4</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> want[i] <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#39;*&#39;</span> <span style="color:#f92672">and</span> want[i] <span style="color:#f92672">!=</span> have[i]:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#66d9ef">False</span>
</span></span><span style="display:flex;"><span>    ver <span style="color:#f92672">=</span> have[<span style="color:#ae81ff">5</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> want[<span style="color:#ae81ff">5</span>] <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#39;*&#39;</span>, <span style="color:#e6db74">&#39;-&#39;</span>):            <span style="color:#75715e"># exact version in the criteria</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> want[<span style="color:#ae81ff">5</span>] <span style="color:#f92672">==</span> ver
</span></span><span style="display:flex;"><span>    checks <span style="color:#f92672">=</span> [(<span style="color:#e6db74">&#39;versionStartIncluding&#39;</span>, <span style="color:#66d9ef">lambda</span> c: c <span style="color:#f92672">&gt;=</span> <span style="color:#ae81ff">0</span>), (<span style="color:#e6db74">&#39;versionStartExcluding&#39;</span>, <span style="color:#66d9ef">lambda</span> c: c <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">0</span>),
</span></span><span style="display:flex;"><span>              (<span style="color:#e6db74">&#39;versionEndIncluding&#39;</span>, <span style="color:#66d9ef">lambda</span> c: c <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">0</span>), (<span style="color:#e6db74">&#39;versionEndExcluding&#39;</span>, <span style="color:#66d9ef">lambda</span> c: c <span style="color:#f92672">&lt;</span> <span style="color:#ae81ff">0</span>)]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> all(test(cmp(ver, m[k])) <span style="color:#66d9ef">for</span> k, test <span style="color:#f92672">in</span> checks <span style="color:#66d9ef">if</span> k <span style="color:#f92672">in</span> m)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">present</span>(m, inventory):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> any(<span style="color:#66d9ef">match</span>(m, c) <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> inventory)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">node</span>(n, inv):
</span></span><span style="display:flex;"><span>    hits <span style="color:#f92672">=</span> [present(m, inv) <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> n[<span style="color:#e6db74">&#39;cpeMatch&#39;</span>]]
</span></span><span style="display:flex;"><span>    r <span style="color:#f92672">=</span> all(hits) <span style="color:#66d9ef">if</span> n[<span style="color:#e6db74">&#39;operator&#39;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#39;AND&#39;</span> <span style="color:#66d9ef">else</span> any(hits)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#f92672">not</span> r <span style="color:#66d9ef">if</span> n<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;negate&#39;</span>) <span style="color:#66d9ef">else</span> r
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">element</span>(el, inv):
</span></span><span style="display:flex;"><span>    res <span style="color:#f92672">=</span> [node(n, inv) <span style="color:#66d9ef">for</span> n <span style="color:#f92672">in</span> el[<span style="color:#e6db74">&#39;nodes&#39;</span>]]
</span></span><span style="display:flex;"><span>    r <span style="color:#f92672">=</span> all(res) <span style="color:#66d9ef">if</span> el<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;operator&#39;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#39;AND&#39;</span> <span style="color:#66d9ef">else</span> any(res)  <span style="color:#75715e"># no operator: OR</span>
</span></span><span style="display:flex;"><span>    r <span style="color:#f92672">=</span> <span style="color:#f92672">not</span> r <span style="color:#66d9ef">if</span> el<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;negate&#39;</span>) <span style="color:#66d9ef">else</span> r
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># the statement must also hit at least one entry flagged vulnerable</span>
</span></span><span style="display:flex;"><span>    vuln <span style="color:#f92672">=</span> any(m[<span style="color:#e6db74">&#39;vulnerable&#39;</span>] <span style="color:#f92672">and</span> present(m, inv)
</span></span><span style="display:flex;"><span>               <span style="color:#66d9ef">for</span> n <span style="color:#f92672">in</span> el[<span style="color:#e6db74">&#39;nodes&#39;</span>] <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> n[<span style="color:#e6db74">&#39;cpeMatch&#39;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> r <span style="color:#f92672">and</span> vuln
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">evaluate</span>(record, inventory):
</span></span><span style="display:flex;"><span>    cna <span style="color:#f92672">=</span> record[<span style="color:#e6db74">&#39;containers&#39;</span>][<span style="color:#e6db74">&#39;cna&#39;</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> any(element(el, inventory) <span style="color:#66d9ef">for</span> el <span style="color:#f92672">in</span> cna<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;cpeApplicability&#39;</span>, []))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">except</span> <span style="color:#a6e22e">ValueError</span> <span style="color:#66d9ef">as</span> e:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#39;unknown (cannot order version </span><span style="color:#e6db74">{</span>e<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#39;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#39;__main__&#39;</span>:
</span></span><span style="display:flex;"><span>    rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>    inventory <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>:]          <span style="color:#75715e"># every CPE installed on one host</span>
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">&#39; + &#39;</span><span style="color:#f92672">.</span>join(inventory), <span style="color:#e6db74">&#39;-&gt;&#39;</span>, evaluate(rec, inventory))
</span></span></code></pre></div><p>Saved as <code>cpe_eval.py</code>, we ran it against three records from the sample and the schema&rsquo;s example (output trimmed to the verdict):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 -I cpe_eval.py CVE-2026-74994.json &#39;cpe:2.3:a:erlang:erlang\/otp:28.5.0.5:*:*:*:*:*:*:*&#39;
</span></span><span style="display:flex;"><span>... -&gt; True
</span></span><span style="display:flex;"><span>$ python3 -I cpe_eval.py CVE-2026-74994.json &#39;cpe:2.3:a:erlang:erlang\/otp:28.5.0.6:*:*:*:*:*:*:*&#39;
</span></span><span style="display:flex;"><span>... -&gt; False
</span></span><span style="display:flex;"><span>$ python3 -I cpe_eval.py full-record-advanced-example.json cpe:2.3:a:example_org:example_enterprise:1.0.5:*:*:*:*:*:*:*
</span></span><span style="display:flex;"><span>... -&gt; False
</span></span><span style="display:flex;"><span>$ python3 -I cpe_eval.py full-record-advanced-example.json cpe:2.3:a:example_org:example_enterprise:1.0.5:*:*:*:*:*:*:* cpe:2.3:o:microsoft:windows:11:*:*:*:*:*:*:*
</span></span><span style="display:flex;"><span>... -&gt; True
</span></span><span style="display:flex;"><span>$ python3 -I cpe_eval.py CVE-2026-115975.json cpe:2.3:a:cervantessec:cervantes:v.1.4.0-beta:*:*:*:*:*:*:*
</span></span><span style="display:flex;"><span>... -&gt; unknown (cannot order version v.1.4.0-beta)
</span></span></code></pre></div><p>The Erlang/OTP record (CVE-2026-74994) is the common case: three ranges in one OR node, one per release line, with <code>versionEndExcluding</code> set to the fixed release. The example record shows the AND rule working: the application version alone is not enough; the platform must be present too. The Cervantes record (CVE-2026-115975) uses version strings like <code>alpha_0.3</code> and <code>1.3-beta</code>, and the honest answer is &ldquo;cannot decide&rdquo;.</p>
<h2 id="where-cpeapplicability-and-affected-disagree">Where cpeApplicability and affected disagree</h2>
<p>CVE-2026-23122 is a Linux kernel record (published February 14, 2026). Its <code>cpeApplicability</code> block has one OR node with two ranges on <code>cpe:2.3:o:linux:linux_kernel</code>: 6.16 up to (excluding) 6.18.8, and 6.16 up to (excluding) 6.19. Its <code>affected</code> block for the semver product says 6.16 is affected, below 6.16 is unaffected, <strong>6.18.8 through 6.18.* is unaffected</strong>, and 6.19 onwards is unaffected.</p>
<p>Our evaluator, run on kernel versions (output trimmed to the version and verdict):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>6.15.4 -&gt; False
</span></span><span style="display:flex;"><span>6.18.7 -&gt; True
</span></span><span style="display:flex;"><span>6.18.9 -&gt; True     # affected block: unaffected (6.18.8 to 6.18.*)
</span></span><span style="display:flex;"><span>6.19   -&gt; False
</span></span></code></pre></div><p>The second CPE range (6.16 to before 6.19) swallows the stable-branch fix, because a CPE range has no way to say &ldquo;fixed in the 6.18 branch at 6.18.8 but not in mainline until 6.19&rdquo;. The <code>affected</code> block can, through ordered entries and a <code>6.18.*</code> upper limit. A scanner that trusts only CPE here raises a false positive on every patched 6.18 kernel.</p>
<p>That is the general lesson: CPE ranges are a lossy projection of branch-aware version data. Use them for discovery (which records mention a product I run?) and as a cross-check, then decide with the <code>affected</code> block.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Treating every matched CPE as vulnerable.</strong> Entries with <code>vulnerable: false</code> are context. In our sample they were rare (10 of 2,667), which is exactly why code that ignores the flag passes tests and fails later.</li>
<li><strong>Applying bounds to an exact-version criteria string.</strong> If the CPE&rsquo;s version field is not <code>*</code>, compare it directly.</li>
<li><strong>Assuming the element operator.</strong> 84% of sampled elements omit it. With a single node it does not matter; with several, decide and document what your tool does. Our script treats a missing element operator as OR.</li>
<li><strong>Comparing versions as strings.</strong> <code>&quot;6.9&quot; &gt; &quot;6.18&quot;</code> is true as text. Use a scheme-aware comparator and return &ldquo;unknown&rdquo; when the scheme is unknown.</li>
<li><strong>Mixing sources silently.</strong> CNA <code>cpeApplicability</code>, the older <code>affected[].cpes</code>, and NVD&rsquo;s configurations are three different statements. Store which one produced a match.</li>
</ul>
<h2 id="where-this-fits-in-triage">Where this fits in triage</h2>
<p>CPE matching answers &ldquo;does this CVE apply to an asset I run?&rdquo;. It does not answer &ldquo;how urgently?&rdquo;. Once a record matches, the next questions are whether it is in CISA KEV, its EPSS probability and its SSVC decision values; our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a> walks that path from record to ticket, and <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">the KEV JSON feed guide</a> covers the exploitation check.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>The counts come from one random sample of 2026 records at one commit; CNA behaviour changes, so rerun the count on your own snapshot before relying on a share.</li>
<li>The evaluator is illustrative. It does not implement the full CPE 2.3 name-matching specification (NISTIR 7696), wildcard characters inside fields, or version schemes beyond dotted integers.</li>
<li>We describe NVD configurations only by the schema&rsquo;s own statement that this block was forked from the NVD CVE API v2.0 schema; we did not compare NVD&rsquo;s published configurations for these CVEs.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Field names, required fields, enums, &ldquo;mostly copied/forked from the NIST NVD CVE API v2.0 schema&rdquo;, alignment recommendation</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema (cve-schema main)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td><code>cpeApplicability</code> allowed in CNA and ADP containers; <code>cpes</code> description recommends the newer block</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Added in 5.1.1; tag dated October 2024</td>
          <td><a href="https://github.com/CVEProject/cve-schema/releases">cve-schema releases</a>; <code>v5.1.1</code> tag commit date</td>
          <td>Verified (release-note text read via a mirror; qualified)</td>
      </tr>
      <tr>
          <td>AND/OR example with vulnerable: false platforms</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/docs/full-record-advanced-example.json">full-record-advanced-example.json</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA stopped adding CPE strings on December 10, 2024</td>
          <td><a href="https://github.com/cisagov/vulnrichment">cisagov/vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Sample counts and shares</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> at commit f0b25a9, our script</td>
          <td>Verified (computed)</td>
      </tr>
      <tr>
          <td>CVE-2026-23122 cpeApplicability vs affected disagreement</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/23xxx/CVE-2026-23122.json">CVE-2026-23122 record</a></td>
          <td>Verified (tested)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CycloneDX VEX: analysis.state and justification to OpenVEX</title><link>https://firewallsync.com/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/</link><guid>https://firewallsync.com/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/</guid><pubDate>Tue, 03 Nov 2026 04:40:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>CycloneDX VEX uses six analysis states and nine justifications; OpenVEX and CSAF use four statuses and five justifications. Here is a field-by-field mapping, which pairs line up and which do not, how Trivy reads CycloneDX VEX, and a tested Python converter that refuses to invent justifications.</description><enclosure url="https://firewallsync.com/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/cover_hu15575583779999317849.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/cover_hu15575583779999317849.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cyclonedx-vex-analysis-state-justification-openvex-csaf-mapping/cover_hu15575583779999317849.jpg" alt="Close-up of a silver modular synthesizer panel with black knobs, patched with pink, green, orange, red and yellow cables" width="1200" height="630"></p><p><strong>Scope.</strong> This is a field-level mapping guide. We compared the CycloneDX JSON schemas for 1.6 and 1.7 (identical enums for the fields below), OpenVEX specification v0.2.0 and its JSON schema, and the CSAF 2.0 JSON schema and VEX profile text from the OASIS TC repository, then read how Trivy (main branch, commit <code>1d1ab074</code>, November 2, 2026) consumes CycloneDX VEX. The converter below was run with Python 3.13; its output validated against the OpenVEX 0.2.0 JSON schema and its inputs against the CycloneDX 1.6 schema. Our <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explainer</a> covers what the statuses mean and how far to trust them; this article is only about moving statements between formats.</p>
<h2 id="the-short-answer">The short answer</h2>
<p>If you receive CycloneDX VEX and your pipeline speaks OpenVEX or CSAF (or the reverse), the states map almost one-to-one, but the justifications do not.</p>
<table>
  <thead>
      <tr>
          <th>CycloneDX <code>analysis.state</code></th>
          <th>OpenVEX <code>status</code></th>
          <th>CSAF <code>product_status</code></th>
          <th>Trivy (code)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>not_affected</code></td>
          <td><code>not_affected</code></td>
          <td><code>known_not_affected</code></td>
          <td>not_affected</td>
      </tr>
      <tr>
          <td><code>false_positive</code></td>
          <td><code>not_affected</code> (no exact match)</td>
          <td><code>known_not_affected</code> (no exact match)</td>
          <td>not_affected</td>
      </tr>
      <tr>
          <td><code>exploitable</code></td>
          <td><code>affected</code></td>
          <td><code>known_affected</code></td>
          <td>affected</td>
      </tr>
      <tr>
          <td><code>in_triage</code></td>
          <td><code>under_investigation</code></td>
          <td><code>under_investigation</code></td>
          <td>under_investigation</td>
      </tr>
      <tr>
          <td><code>resolved</code></td>
          <td><code>fixed</code></td>
          <td><code>fixed</code></td>
          <td>fixed</td>
      </tr>
      <tr>
          <td><code>resolved_with_pedigree</code></td>
          <td><code>fixed</code> (pedigree evidence is lost)</td>
          <td><code>fixed</code></td>
          <td>fixed</td>
      </tr>
  </tbody>
</table>
<p>Of CycloneDX&rsquo;s nine justifications, only <code>code_not_present</code> and <code>code_not_reachable</code> have a clear counterpart among the five CISA-derived labels that OpenVEX and CSAF share. Convert those two; carry the other seven as text.</p>
<h2 id="where-vex-sits-in-a-cyclonedx-document">Where VEX sits in a CycloneDX document</h2>
<p>CycloneDX has no separate VEX document type. VEX is the <code>vulnerabilities</code> array of a BOM, and the fields that matter are:</p>
<ul>
<li><code>id</code>: the vulnerability identifier, such as a CVE ID.</li>
<li><code>affects[].ref</code>: what the statement is about. The schema allows a <code>bom-ref</code> in the same document or a BOM-Link element. A BOM-Link has the form <code>urn:cdx:serialNumber/version#bom-ref</code>, and the schema&rsquo;s pattern requires the serial number UUID, a version of 1 or more, and a fragment.</li>
<li><code>analysis.state</code>: one of <code>resolved</code>, <code>resolved_with_pedigree</code>, <code>exploitable</code>, <code>in_triage</code>, <code>false_positive</code>, <code>not_affected</code>.</li>
<li><code>analysis.justification</code>: one of nine reasons, described below.</li>
<li><code>analysis.response</code>: zero or more of <code>can_not_fix</code>, <code>will_not_fix</code>, <code>update</code>, <code>rollback</code>, <code>workaround_available</code>.</li>
<li><code>analysis.detail</code>: free text. The schema says that if a vulnerability is not exploitable, &ldquo;this field should include specific details on why the component or service is not impacted.&rdquo;</li>
</ul>
<p>A VEX statement can sit inside the SBOM itself or in a separate BOM that only has <code>vulnerabilities</code> and refers back to the SBOM by BOM-Link. That second pattern matters for tooling, below.</p>
<h2 id="states-four-clean-pairs-and-two-lossy-ones">States: four clean pairs and two lossy ones</h2>
<p>The CycloneDX schema defines the six states in its <code>meta:enum</code>. Four map directly:</p>
<ul>
<li><code>not_affected</code>: &ldquo;The component or service is not affected by the vulnerability. Justification should be specified for all not_affected cases.&rdquo; This is OpenVEX <code>not_affected</code> and CSAF <code>known_not_affected</code>.</li>
<li><code>exploitable</code>: &ldquo;The vulnerability may be directly or indirectly exploitable.&rdquo; This is OpenVEX <code>affected</code> and CSAF <code>known_affected</code>.</li>
<li><code>in_triage</code>: &ldquo;The vulnerability is being investigated.&rdquo; This is <code>under_investigation</code> in both.</li>
<li><code>resolved</code>: &ldquo;The vulnerability has been remediated.&rdquo; This is <code>fixed</code> in both.</li>
</ul>
<p>Two lose meaning:</p>
<ul>
<li><strong><code>false_positive</code></strong>: &ldquo;The vulnerability is not specific to the component or service and was falsely identified or associated.&rdquo; OpenVEX and CSAF have no status for &ldquo;the scanner was wrong&rdquo;. The safe target is <code>not_affected</code>, which is also what Trivy&rsquo;s <code>cdxStatus</code> function does (<code>case cdx.IASFalsePositive, cdx.IASNotAffected: return types.FindingStatusNotAffected</code>). Keep the word <code>false_positive</code> in the impact statement so a reviewer can see it.</li>
<li><strong><code>resolved_with_pedigree</code></strong>: remediated &ldquo;and evidence of the changes are provided in the affected components pedigree&rdquo;. It becomes <code>fixed</code>, and the pointer to pedigree evidence is lost unless you keep it as text.</li>
</ul>
<h2 id="justifications-two-of-nine-line-up">Justifications: two of nine line up</h2>
<p>OpenVEX&rsquo;s five justification labels come from CISA&rsquo;s June 2022 status-justification document, and the CSAF 2.0 schema&rsquo;s <code>flags[].label</code> uses the same five strings. CycloneDX defines a different set of nine. Here is each one with its CycloneDX definition and our reading of the closest OpenVEX/CSAF label.</p>
<table>
  <thead>
      <tr>
          <th>CycloneDX <code>justification</code></th>
          <th>CycloneDX definition</th>
          <th>OpenVEX / CSAF label</th>
          <th>Our call</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>code_not_present</code></td>
          <td>&ldquo;The code has been removed or tree-shaked.&rdquo;</td>
          <td><code>vulnerable_code_not_present</code></td>
          <td>Map</td>
      </tr>
      <tr>
          <td><code>code_not_reachable</code></td>
          <td>&ldquo;The vulnerable code is not invoked at runtime.&rdquo;</td>
          <td><code>vulnerable_code_not_in_execute_path</code></td>
          <td>Map</td>
      </tr>
      <tr>
          <td><code>requires_dependency</code></td>
          <td>&ldquo;Exploitability requires a dependency that is not present.&rdquo;</td>
          <td>none exact (<code>component_not_present</code> means the vulnerable component itself is absent)</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>requires_configuration</code></td>
          <td>&ldquo;Exploitability requires a configurable option to be set/unset.&rdquo;</td>
          <td>none</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>requires_environment</code></td>
          <td>&ldquo;Exploitability requires a certain environment which is not present.&rdquo;</td>
          <td>none</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>protected_by_compiler</code></td>
          <td>&ldquo;Exploitability requires a compiler flag to be set/unset.&rdquo;</td>
          <td>none exact</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>protected_at_runtime</code></td>
          <td>&ldquo;Exploits are prevented at runtime.&rdquo;</td>
          <td><code>inline_mitigations_already_exist</code> only if the protection cannot be disabled</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>protected_at_perimeter</code></td>
          <td>&ldquo;Attacks are blocked at physical, logical, or network perimeter.&rdquo;</td>
          <td>none</td>
          <td>Text</td>
      </tr>
      <tr>
          <td><code>protected_by_mitigating_control</code></td>
          <td>&ldquo;Preventative measures have been implemented that reduce the likelihood and/or impact of the vulnerability.&rdquo;</td>
          <td>none</td>
          <td>Text</td>
      </tr>
  </tbody>
</table>
<p>Why so strict? The OpenVEX spec defines <code>inline_mitigations_already_exist</code> as built-in protections that &ldquo;cannot be subverted by the attacker and cannot be configured or disabled by the user.&rdquo; A CycloneDX <code>protected_at_runtime</code> or <code>protected_by_compiler</code> claim does not tell you that. <code>protected_at_perimeter</code> and <code>protected_by_mitigating_control</code> describe the deployment, not the product, and a mitigating control that &ldquo;reduce[s] the likelihood and/or impact&rdquo; is weaker than &ldquo;not affected&rdquo; in the CISA sense.</p>
<p>This mapping is our editorial reading of the published definitions. We did not find a mapping table published by the CycloneDX project or CISA.</p>
<p>OpenVEX gives you a way out: for <code>not_affected</code>, &ldquo;a VEX statement MUST include either a status justification or an impact_statement.&rdquo; So you can emit a <code>not_affected</code> statement with no label and an impact statement that preserves the original CycloneDX reason. CSAF&rsquo;s VEX profile has the same escape hatch: each <code>known_not_affected</code> product needs either a machine-readable flag or a human-readable justification in <code>threats</code> with category <code>impact</code>.</p>
<h2 id="responses-and-action-statements">Responses and action statements</h2>
<p>For affected products, both targets require an action:</p>
<ul>
<li>OpenVEX: for <code>affected</code>, &ldquo;a VEX statement MUST include a statement that SHOULD describe actions to remediate or mitigate the vulnerability&rdquo; (<code>action_statement</code>).</li>
<li>CSAF VEX profile: for <code>known_affected</code>, &ldquo;additional product specific information SHALL be provided in <code>/vulnerabilities[]/remediations</code> as an action statement.&rdquo;</li>
</ul>
<p>CycloneDX&rsquo;s <code>response</code> values give you a starting point. <code>update</code> and <code>rollback</code> fit CSAF&rsquo;s <code>vendor_fix</code> remediation category, <code>workaround_available</code> fits <code>workaround</code>, and <code>will_not_fix</code> / <code>can_not_fix</code> fit <code>no_fix_planned</code> (or <code>none_available</code>), which the CSAF profile explicitly permits for an action statement. OpenVEX takes a free-text <code>action_statement</code>, so join the responses with <code>analysis.detail</code>. If an <code>exploitable</code> statement has neither a response nor detail, the converted statement will be valid but useless; send it back to the supplier.</p>
<h2 id="bom-link-the-part-that-silently-drops-statements">BOM-Link: the part that silently drops statements</h2>
<p>A CycloneDX VEX BOM that references components by BOM-Link is bound to one SBOM version. Trivy&rsquo;s docs (v0.67) say the tool &ldquo;only supports the Independent BOM and VEX BOM format, so you need to provide a separate VEX file alongside the SBOM,&rdquo; that &ldquo;BOM-Links must be placed in <code>affects.ref</code>,&rdquo; and that the input SBOM must be CycloneDX. The usage example is <code>trivy sbom trivy.sbom.cdx --vex trivy.vex.cdx</code>. The feature is marked experimental.</p>
<p>Trivy&rsquo;s code adds a detail the docs do not spell out: if the BOM-Link&rsquo;s serial number or version does not equal the scanned SBOM&rsquo;s, it logs <code>URN doesn't match with SBOM</code> and skips the statement. Regenerate the SBOM (new serial number or version) without regenerating the VEX, and every suppression stops applying. That fails safe, since findings come back rather than disappear, but it surprises teams.</p>
<p>OpenVEX has no BOM-Link. Products are identified by an <code>@id</code> such as a Package URL, so a converter has to resolve each BOM-Link to the referenced component&rsquo;s <code>purl</code> in the SBOM. That is why the converter below takes both files.</p>
<h2 id="a-tested-converter-cyclonedx-vex-to-openvex">A tested converter: CycloneDX VEX to OpenVEX</h2>
<p>This Python 3 script reads a CycloneDX SBOM and a separate CycloneDX VEX BOM and writes an OpenVEX 0.2.0 document. It maps states as in the first table, maps only the two clean justifications, and keeps everything else as text. It is illustrative, not production-ready: it does not handle version ranges in <code>affects[].versions</code>, services, or nested BOMs beyond <code>components</code>.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Convert a CycloneDX VEX BOM (independent VEX, BOM-Link refs) to OpenVEX v0.2.0.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 cdx2openvex.py sbom.cdx.json vex.cdx.json author &gt; out.openvex.json&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> hashlib
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> datetime <span style="color:#f92672">import</span> datetime, timezone
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>STATUS <span style="color:#f92672">=</span> {  <span style="color:#75715e"># CycloneDX analysis.state -&gt; OpenVEX status</span>
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;not_affected&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>, <span style="color:#e6db74">&#34;false_positive&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;exploitable&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#e6db74">&#34;in_triage&#34;</span>: <span style="color:#e6db74">&#34;under_investigation&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;resolved&#34;</span>: <span style="color:#e6db74">&#34;fixed&#34;</span>, <span style="color:#e6db74">&#34;resolved_with_pedigree&#34;</span>: <span style="color:#e6db74">&#34;fixed&#34;</span>,
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Only the pairs whose definitions line up; everything else is carried as text.</span>
</span></span><span style="display:flex;"><span>JUSTIFICATION <span style="color:#f92672">=</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;code_not_present&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_present&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;code_not_reachable&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span>,
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">purls_by_ref</span>(sbom):
</span></span><span style="display:flex;"><span>    serial, version <span style="color:#f92672">=</span> sbom[<span style="color:#e6db74">&#34;serialNumber&#34;</span>], sbom<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;version&#34;</span>, <span style="color:#ae81ff">1</span>)
</span></span><span style="display:flex;"><span>    out, stack <span style="color:#f92672">=</span> {}, list(sbom<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;components&#34;</span>, []))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">while</span> stack:
</span></span><span style="display:flex;"><span>        c <span style="color:#f92672">=</span> stack<span style="color:#f92672">.</span>pop()
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;bom-ref&#34;</span> <span style="color:#f92672">in</span> c <span style="color:#f92672">and</span> <span style="color:#e6db74">&#34;purl&#34;</span> <span style="color:#f92672">in</span> c:
</span></span><span style="display:flex;"><span>            out[<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;urn:cdx:</span><span style="color:#e6db74">{</span>serial<span style="color:#f92672">.</span>removeprefix(<span style="color:#e6db74">&#39;urn:uuid:&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">/</span><span style="color:#e6db74">{</span>version<span style="color:#e6db74">}</span><span style="color:#e6db74">#</span><span style="color:#e6db74">{</span>c[<span style="color:#e6db74">&#39;bom-ref&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>] <span style="color:#f92672">=</span> c[<span style="color:#e6db74">&#34;purl&#34;</span>]
</span></span><span style="display:flex;"><span>        stack<span style="color:#f92672">.</span>extend(c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;components&#34;</span>, []))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">convert</span>(sbom, vex, author):
</span></span><span style="display:flex;"><span>    refs, statements, problems <span style="color:#f92672">=</span> purls_by_ref(sbom), [], []
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> vex<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>, []):
</span></span><span style="display:flex;"><span>        a <span style="color:#f92672">=</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;analysis&#34;</span>, {})
</span></span><span style="display:flex;"><span>        state <span style="color:#f92672">=</span> a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;state&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> state <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> STATUS:
</span></span><span style="display:flex;"><span>            problems<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;id&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">: no analysis.state, skipped&#34;</span>); <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        products <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> af <span style="color:#f92672">in</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;affects&#34;</span>, []):
</span></span><span style="display:flex;"><span>            purl <span style="color:#f92672">=</span> refs<span style="color:#f92672">.</span>get(af[<span style="color:#e6db74">&#34;ref&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> purl: products<span style="color:#f92672">.</span>append({<span style="color:#e6db74">&#34;@id&#34;</span>: purl})
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">else</span>: problems<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;id&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>af[<span style="color:#e6db74">&#39;ref&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> not found in SBOM&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> products: <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        s <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;vulnerability&#34;</span>: {<span style="color:#e6db74">&#34;name&#34;</span>: v[<span style="color:#e6db74">&#34;id&#34;</span>]}, <span style="color:#e6db74">&#34;products&#34;</span>: products, <span style="color:#e6db74">&#34;status&#34;</span>: STATUS[state]}
</span></span><span style="display:flex;"><span>        detail <span style="color:#f92672">=</span> a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;detail&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)<span style="color:#f92672">.</span>strip()
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> s[<span style="color:#e6db74">&#34;status&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;not_affected&#34;</span>:
</span></span><span style="display:flex;"><span>            j <span style="color:#f92672">=</span> JUSTIFICATION<span style="color:#f92672">.</span>get(a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;justification&#34;</span>))
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> j: s[<span style="color:#e6db74">&#34;justification&#34;</span>] <span style="color:#f92672">=</span> j
</span></span><span style="display:flex;"><span>            note <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;CycloneDX state=</span><span style="color:#e6db74">{</span>state<span style="color:#e6db74">}</span><span style="color:#e6db74">, justification=</span><span style="color:#e6db74">{</span>a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;justification&#39;</span>, <span style="color:#e6db74">&#39;none&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>            s[<span style="color:#e6db74">&#34;impact_statement&#34;</span>] <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>note<span style="color:#e6db74">}</span><span style="color:#e6db74">. </span><span style="color:#e6db74">{</span>detail<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span><span style="color:#f92672">.</span>strip()
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">elif</span> s[<span style="color:#e6db74">&#34;status&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;affected&#34;</span>:
</span></span><span style="display:flex;"><span>            resp <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;, &#34;</span><span style="color:#f92672">.</span>join(a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;response&#34;</span>, [])) <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;none given&#34;</span>
</span></span><span style="display:flex;"><span>            s[<span style="color:#e6db74">&#34;action_statement&#34;</span>] <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;CycloneDX response: </span><span style="color:#e6db74">{</span>resp<span style="color:#e6db74">}</span><span style="color:#e6db74">. </span><span style="color:#e6db74">{</span>detail<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span><span style="color:#f92672">.</span>strip()
</span></span><span style="display:flex;"><span>        statements<span style="color:#f92672">.</span>append(s)
</span></span><span style="display:flex;"><span>    body <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>dumps(statements, sort_keys<span style="color:#f92672">=</span><span style="color:#66d9ef">True</span>)<span style="color:#f92672">.</span>encode()
</span></span><span style="display:flex;"><span>    doc <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;@context&#34;</span>: <span style="color:#e6db74">&#34;https://openvex.dev/ns/v0.2.0&#34;</span>,
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;https://example.com/vex/&#34;</span> <span style="color:#f92672">+</span> hashlib<span style="color:#f92672">.</span>sha256(body)<span style="color:#f92672">.</span>hexdigest()[:<span style="color:#ae81ff">16</span>],
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;author&#34;</span>: author,
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;timestamp&#34;</span>: datetime<span style="color:#f92672">.</span>now(timezone<span style="color:#f92672">.</span>utc)<span style="color:#f92672">.</span>isoformat(timespec<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;seconds&#34;</span>),
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;version&#34;</span>: <span style="color:#ae81ff">1</span>, <span style="color:#e6db74">&#34;statements&#34;</span>: statements}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> doc, problems
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    sbom, vex <span style="color:#f92672">=</span> (json<span style="color:#f92672">.</span>load(open(p)) <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:<span style="color:#ae81ff">3</span>])
</span></span><span style="display:flex;"><span>    doc, problems <span style="color:#f92672">=</span> convert(sbom, vex, sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">3</span>])
</span></span><span style="display:flex;"><span>    json<span style="color:#f92672">.</span>dump(doc, sys<span style="color:#f92672">.</span>stdout, indent<span style="color:#f92672">=</span><span style="color:#ae81ff">2</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> problems: print(<span style="color:#e6db74">&#34;WARN&#34;</span>, p, file<span style="color:#f92672">=</span>sys<span style="color:#f92672">.</span>stderr)
</span></span></code></pre></div><h3 id="test-run">Test run</h3>
<p>We fed it a three-component CycloneDX 1.6 SBOM (log4j-core 2.14.1, jackson-databind 2.13.0, snakeyaml 1.30) and a VEX BOM with four statements. The assessments in the VEX are invented test data, not real vendor findings:</p>
<ol>
<li>CVE-2021-44228 on log4j-core: <code>not_affected</code>, <code>code_not_reachable</code>.</li>
<li>CVE-2022-42003 on jackson-databind: <code>not_affected</code>, <code>requires_configuration</code>.</li>
<li>CVE-2022-1471 on snakeyaml: <code>exploitable</code>, response <code>update</code>.</li>
<li>CVE-2021-44228 again, but with a BOM-Link naming SBOM version <code>2</code> instead of <code>1</code>.</li>
</ol>
<p>Both input files passed <code>check-jsonschema</code> against the CycloneDX 1.6 schema. The run printed one warning to stderr and three statements; the output validated against <code>openvex_json_schema.json</code> (v0.2.0) with the Python <code>jsonschema</code> package:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>WARN CVE-2021-44228: urn:cdx:3e671687-395b-41f5-a30f-a58921a69b79/2#pkg-log4j not found in SBOM
</span></span></code></pre></div><p>The <code>statements</code> array of the output:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>[
</span></span><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vulnerability&#34;</span>: { <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2021-44228&#34;</span> },
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;products&#34;</span>: [ { <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:maven/org.apache.logging.log4j/log4j-core@2.14.1&#34;</span> } ],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;justification&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;impact_statement&#34;</span>: <span style="color:#e6db74">&#34;CycloneDX state=not_affected, justification=code_not_reachable. JNDI lookups are never invoked; the app logs through SLF4J with Logback.&#34;</span>
</span></span><span style="display:flex;"><span>},
</span></span><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vulnerability&#34;</span>: { <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2022-42003&#34;</span> },
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;products&#34;</span>: [ { <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:maven/com.fasterxml.jackson.core/jackson-databind@2.13.0&#34;</span> } ],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;impact_statement&#34;</span>: <span style="color:#e6db74">&#34;CycloneDX state=not_affected, justification=requires_configuration. The deserialization feature this issue depends on is disabled in our ObjectMapper configuration.&#34;</span>
</span></span><span style="display:flex;"><span>},
</span></span><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vulnerability&#34;</span>: { <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2022-1471&#34;</span> },
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;products&#34;</span>: [ { <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:maven/org.yaml/snakeyaml@1.30&#34;</span> } ],
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;action_statement&#34;</span>: <span style="color:#e6db74">&#34;CycloneDX response: update. Dependency upgrade ships in release 4.2.&#34;</span>
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>The second statement is the important one: it stays <code>not_affected</code> with no invented label, and the reason survives as text. The fourth input statement was dropped with a warning instead of being applied to the wrong SBOM.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Mapping every <code>protected_*</code> justification to <code>inline_mitigations_already_exist</code>.</strong> That label claims the protection cannot be disabled. A WAF rule or a compiler flag can.</li>
<li><strong>Dropping <code>false_positive</code> statements</strong> because the target format has no matching status. Map them to <code>not_affected</code> and say why.</li>
<li><strong>Dropping the <code>detail</code> text.</strong> It is often the only evidence a reviewer has, and both OpenVEX and CSAF have a place for it.</li>
<li><strong>Reusing a VEX BOM after regenerating the SBOM.</strong> BOM-Links name an exact serial number and version.</li>
<li><strong>Treating converted VEX as first-party.</strong> Keep the original author. OpenVEX <code>author</code> should be the organization that made the assessment, not the converter.</li>
</ul>
<h2 id="where-this-fits">Where this fits</h2>
<p>VEX is the &ldquo;is my product actually affected&rdquo; step in our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>, after the CVE record and before KEV, SSVC and EPSS. The SBOM that the VEX points at is covered in our guide to <a href="/posts/sbom-minimum-elements-2026-what-cisa-and-partners-changed-from-the-2021-ntia-baseline/">SBOM minimum elements in 2026</a>. If you want a signal for which <code>exploitable</code> statements to fix first, our <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV comparison</a> covers the trade-offs.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The justification mapping is our reading of the published definitions; no standards body has published it. We read CSAF 2.0 from the OASIS TC&rsquo;s GitHub repository (schema and editor&rsquo;s draft prose) rather than the OASIS-hosted standard. We read Trivy&rsquo;s main branch, not a tagged release, and its docs for v0.67; behaviour may differ in the version you run. We did not run Trivy itself or test the reverse conversion (OpenVEX or CSAF to CycloneDX), which is harder because CycloneDX <code>affects</code> needs a component reference the OpenVEX product may not resolve to.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CycloneDX states, justifications, responses and their definitions; detail field; BOM-Link pattern</td>
          <td><a href="https://github.com/CycloneDX/specification/tree/master/schema">CycloneDX bom-1.6 / bom-1.7 JSON schemas</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOM-Link syntax <code>urn:cdx:serialNumber/version#bom-ref</code></td>
          <td><a href="https://www.cyclonedx.org/capabilities/bomlink/">CycloneDX BOM-Link</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>OpenVEX statuses, five justifications, justification or impact_statement rule, action_statement rule</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX spec v0.2.0</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CSAF product_status categories, flag labels, remediation categories</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/json_schema/csaf_json_schema.json">CSAF 2.0 JSON schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CSAF VEX profile requirements for known_not_affected and known_affected</td>
          <td><a href="https://github.com/oasis-tcs/csaf/blob/master/csaf_2.0/prose/csaf-v2-editor-draft.md">CSAF 2.0 prose, section 4.5</a></td>
          <td>Qualified (TC repository draft text)</td>
      </tr>
      <tr>
          <td>Trivy maps false_positive to not_affected and skips BOM-Links that do not match the SBOM</td>
          <td><a href="https://github.com/aquasecurity/trivy/blob/main/pkg/vex/cyclonedx.go">Trivy pkg/vex/cyclonedx.go</a></td>
          <td>Verified (main branch, 2026-11-02)</td>
      </tr>
      <tr>
          <td>Trivy CycloneDX VEX: separate VEX BOM, BOM-Link in affects.ref, CycloneDX SBOM only, experimental</td>
          <td><a href="https://trivy.dev/docs/latest/supply-chain/vex/file">Trivy docs: VEX file</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Converter output and schema validation</td>
          <td>Our run, Python 3.13, jsonschema, check-jsonschema</td>
          <td>Verified (code)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE ADP Containers: CISA-ADP, CVE Program and Supplier ADPs</title><link>https://firewallsync.com/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/</link><guid>https://firewallsync.com/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/</guid><pubDate>Tue, 03 Nov 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>A CVE record has one CNA container and possibly several ADP containers. Here is what CISA-ADP, the CVE Program Container and supplier ADPs such as redhat-SADP and siemens-SADP add, how often each appears in a sample of 2,000 2026 records, and a tested script that shows who said what.</description><enclosure url="https://firewallsync.com/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/cover_hu6338534624026322280.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/cover_hu6338534624026322280.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-adp-containers-cisa-adp-cve-program-supplier-sadp/cover_hu6338534624026322280.jpg" alt="Grid of dark wooden card catalog drawers, each with a brass label holder showing an alphabetical range and a brass pull" width="1200" height="630"></p><p><strong>Scope.</strong> This is a specification explainer with measurements from public data. We read the CVE JSON schema (<code>CVE_Record_Format.json</code>, default <code>dataVersion</code> 5.2.0, CVEProject/cve-schema main branch), the cvelistV5 README and the CISA Vulnrichment README, then sampled 2,000 random 2026 record files from cvelistV5 at commit <code>1a44d59</code> (November 3, 2026). The script below was run with Python 3.13.</p>
<h2 id="the-short-answer">The short answer</h2>
<p>A CVE record is not one author&rsquo;s statement. Under <code>containers</code> there is exactly one <code>cna</code> object, written by the CVE Numbering Authority that assigned the ID, and an optional <code>adp</code> array. Each <code>adp</code> entry is written by an Authorized Data Publisher and names its author in <code>providerMetadata.shortName</code>.</p>
<p>In 2026 records you will mostly meet three kinds:</p>
<table>
  <thead>
      <tr>
          <th><code>shortName</code></th>
          <th>Title in the container</th>
          <th>What it adds</th>
          <th>Share of 1,963 published records in our sample</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>CISA-ADP</code></td>
          <td>CISA ADP Vulnrichment</td>
          <td>SSVC decision points; sometimes CVSS, CWE, KEV date</td>
          <td>93.9%</td>
      </tr>
      <tr>
          <td><code>CVE</code></td>
          <td>CVE Program Container</td>
          <td>References added by the CVE Program</td>
          <td>2.2%</td>
      </tr>
      <tr>
          <td><code>redhat-SADP</code>, <code>siemens-SADP</code></td>
          <td>Supplier&rsquo;s own titles (or none)</td>
          <td>The supplier&rsquo;s affected products, versions, fixes, severity</td>
          <td>1.2% combined</td>
      </tr>
  </tbody>
</table>
<p>The rule that matters for tooling: read the CNA container and the CVE Program Container every time, read CISA-ADP for prioritization signals, and treat supplier ADPs as extra affected-product data for that supplier&rsquo;s builds. Keep every value tagged with who published it.</p>
<h2 id="what-the-schema-says-about-adp-containers">What the schema says about ADP containers</h2>
<p>The schema describes <code>containers</code> as &ldquo;a set of structures (called containers) used to store vulnerability information related to a specific CVE ID provided by a specific organization participating in the CVE program.&rdquo; It goes on: &ldquo;There can only be one &lsquo;cna&rsquo; container, as there can only be one assigning CNA. However, there can be multiple &lsquo;adp&rsquo; containers.&rdquo;</p>
<p>Three details shape how you parse them:</p>
<ul>
<li><strong>Only <code>providerMetadata</code> is required.</strong> The <code>adpContainer</code> definition lists <code>providerMetadata</code> as required with <code>minProperties: 2</code>, so an ADP container can be just a publisher plus one other field. Code that assumes <code>affected</code>, <code>descriptions</code> or <code>metrics</code> will exist breaks.</li>
<li><strong>The fields mirror the CNA container.</strong> An ADP container can carry <code>affected</code>, <code>cpeApplicability</code>, <code>problemTypes</code>, <code>metrics</code>, <code>references</code>, <code>solutions</code>, <code>workarounds</code>, <code>exploits</code>, <code>timeline</code> and more. The schema notes that &ldquo;for the most part, the &lsquo;cna&rsquo; and &lsquo;adp&rsquo; containers contain the same properties. The main differences are the source of the information.&rdquo;</li>
<li><strong>Extensions are allowed.</strong> <code>adpContainer</code> permits properties matching <code>^x_[^.]*$</code>. Supplier ADPs use this: every redhat-SADP and siemens-SADP container in our sample had <code>&quot;x_adpType&quot;: &quot;supplier&quot;</code>, and Red Hat&rsquo;s also carried <code>x_generator</code>. These are not schema-defined fields, so treat them as hints.</li>
</ul>
<p><code>providerMetadata</code> holds <code>orgId</code> (a UUID, required), <code>shortName</code> and <code>dateUpdated</code>. Match publishers on <code>orgId</code> if you need certainty; <code>shortName</code> is what humans read.</p>
<h2 id="the-three-kinds-you-will-see">The three kinds you will see</h2>
<h3 id="cisa-adp-prioritization-signals">CISA-ADP: prioritization signals</h3>
<p>The cvelistV5 README says the CISA ADP provides three things: SSVC, KEV catalog data, and &ldquo;Vulnrichment&rdquo; updates such as &ldquo;missing CVSS, CWE, CPE information for CVE Records that meet specific threat characteristics, and for when CNAs do not provide it themselves.&rdquo;</p>
<p>What that looked like in our sample of 1,963 published records:</p>
<table>
  <thead>
      <tr>
          <th>CISA-ADP content</th>
          <th>Records</th>
          <th>Share of published</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>SSVC block (<code>metrics[].other.type = &quot;ssvc&quot;</code>)</td>
          <td>1,844</td>
          <td>93.9%</td>
      </tr>
      <tr>
          <td>CVSS (v3.1 in all but one)</td>
          <td>389</td>
          <td>19.8%</td>
      </tr>
      <tr>
          <td>CWE in <code>problemTypes</code></td>
          <td>336</td>
          <td>17.1%</td>
      </tr>
      <tr>
          <td>KEV block (<code>metrics[].other.type = &quot;kev&quot;</code>)</td>
          <td>5</td>
          <td>0.3%</td>
      </tr>
  </tbody>
</table>
<p>Every CISA-ADP container in the sample carried SSVC. CVSS and CWE were mostly gap-fills: 381 of the 389 CVSS entries (97.9%) and 326 of the 336 CWE entries (97.0%) sat on records whose CNA container had none. That matches the Vulnrichment README: &ldquo;For those CVEs that do not already have these fields populated by the originating CNA, CISA will populate the associated ADP container with those values when there is enough supporting evidence to do so. At no point will CISA overwrite the originating CNA&rsquo;s data in the original CNA container.&rdquo;</p>
<p>The exceptions matter. In 8 records both the CNA and CISA-ADP carried a CVSS score, and they did not always agree: in CVE-2026-70373 the CNA (TuranSec) published <code>CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H</code> (8.8) while CISA-ADP published <code>.../C:H/I:N/A:N</code> (6.5). We did not check which came first. The point is that a parser which picks &ldquo;the first CVSS it finds&rdquo; will produce different answers depending on container order.</p>
<p>The KEV block is small: <code>dateAdded</code> and a link to the catalog. Treat it as a pointer, and use the KEV catalog itself for <code>dueDate</code>, <code>requiredAction</code> and ransomware use. Our <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV JSON feed field guide</a> covers those fields, and the <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC guide</a> covers the three SSVC values.</p>
<h3 id="the-cve-program-container-references-you-must-not-skip">The CVE Program Container: references you must not skip</h3>
<p>The cvelistV5 README is explicit. The container has title &ldquo;CVE Program Container&rdquo; and <code>shortName</code> &ldquo;CVE&rdquo;, and &ldquo;all CVE Program-added references after 7/31/2024 for a CVE Record will be stored in the CVE Program Container of that Record. CNA-provided references will continue to be stored in the CNA Container.&rdquo;</p>
<p>It also says what tools must do: &ldquo;to retrieve all information about a reported vulnerability in the CVE Repository, tool vendors and community users will need to examine the CVE Record CNA Container and the CVE Program Container (if one exists). These two containers are minimally required to obtain the core information required by the Program. All other ADP containers remain optional from a Program perspective.&rdquo;</p>
<p>Two practical consequences:</p>
<ul>
<li>References tagged <code>x_transferred</code> were copied from the CNA container on July 31, 2024, so expect duplicates. The README says downstream users &ldquo;will have to determine the appropriate way to resolve potential reference duplications.&rdquo; Deduplicate on URL.</li>
<li>New records only get this container if the Program adds something. In our 2026 sample, 43 records (2.2%) had one.</li>
</ul>
<h3 id="supplier-adps-a-vendors-view-of-someone-elses-cve">Supplier ADPs: a vendor&rsquo;s view of someone else&rsquo;s CVE</h3>
<p>Supplier ADP containers let a vendor that ships an upstream component add its own product data to a CVE that a different CNA assigned. Siemens describes its SADP as useful when &ldquo;a supplier like Siemens wants to add information to a vulnerability&rdquo; in an upstream dependency, and says it was introduced in April 2026 after a pilot. Its stated aim is better scanner matching for affected Siemens products, with &ldquo;known-not-affected&rdquo; product data planned later.</p>
<p>In our sample, 24 published records (1.2%) carried one: 18 from <code>redhat-SADP</code>, 6 from <code>siemens-SADP</code>. Their shapes differ:</p>
<ul>
<li><strong>redhat-SADP</strong> containers looked like a full advisory: <code>affected</code> entries per Red Hat product stream with CPEs, package names and RPM <code>versionType</code>, plus Red Hat&rsquo;s severity rating, CVSS, <code>solutions</code>, <code>workarounds</code> and references.</li>
<li><strong>siemens-SADP</strong> containers were lean: <code>affected</code> entries for Siemens products, references to the Siemens advisory, and nothing else. All had <code>defaultStatus</code> set to <code>unknown</code>.</li>
</ul>
<p>Two traps. First, supplier <code>affected</code> data describes the supplier&rsquo;s product names and version strings, not the upstream project&rsquo;s. A Red Hat entry such as <code>defaultStatus: affected</code> with <code>lessThan: &quot;*&quot;</code> and <code>status: unaffected</code> from a fixed RPM version is a statement about that Red Hat package, which is how our <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">affected-versions guide</a> explains these encodings. Second, supplier severity ratings and CVSS are the supplier&rsquo;s assessment for its builds and can legitimately differ from the CNA&rsquo;s.</p>
<h2 id="one-record-four-publishers">One record, four publishers</h2>
<p>CVE-2026-46300 is a Linux kernel flaw assigned by the <code>Linux</code> CNA and added to the KEV catalog on October 28, 2026. Its record shows why you need a per-provider view. This is the output of the script in the next section against the cvelistV5 file at commit <code>1a44d59</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2026-46300 PUBLISHED dataVersion 5.2
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>[cna] Linux (updated 2026-08-05)
</span></span><span style="display:flex;"><span>  cvss: [&#39;3.1 7.8 CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H&#39;]
</span></span><span style="display:flex;"><span>  affected_products: 2
</span></span><span style="display:flex;"><span>  references: 8
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>[adp] CVE (updated 2026-05-23)
</span></span><span style="display:flex;"><span>  references: 4
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>[adp] redhat-SADP (updated 2026-09-01)
</span></span><span style="display:flex;"><span>  other: [&#39;Red Hat severity rating: Important&#39;]
</span></span><span style="display:flex;"><span>  cvss: [&#39;3.1 7.8 CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H&#39;]
</span></span><span style="display:flex;"><span>  cwe: [&#39;CWE-123&#39;]
</span></span><span style="display:flex;"><span>  affected_products: 39
</span></span><span style="display:flex;"><span>  references: 36
</span></span><span style="display:flex;"><span>  solutions: 33
</span></span><span style="display:flex;"><span>  workarounds: 1
</span></span><span style="display:flex;"><span>  x_adpType: supplier
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>[adp] siemens-SADP (updated 2026-10-15)
</span></span><span style="display:flex;"><span>  affected_products: 10
</span></span><span style="display:flex;"><span>  references: 2
</span></span><span style="display:flex;"><span>  x_adpType: supplier
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>[adp] CISA-ADP (updated 2026-10-29)
</span></span><span style="display:flex;"><span>  ssvc: Exploitation=active, Automatable=no, Technical Impact=total
</span></span><span style="display:flex;"><span>  kev: dateAdded=2026-10-28
</span></span><span style="display:flex;"><span>  cwe: [&#39;CWE-456&#39;]
</span></span><span style="display:flex;"><span>  references: 1
</span></span></code></pre></div><p>Read it as five separate statements:</p>
<ul>
<li>The <strong>CNA</strong> says what the kernel project considers affected and scores it 7.8. It gives no CWE.</li>
<li>The <strong>CVE Program</strong> added four references a tool reading only the CNA container would miss.</li>
<li><strong>Red Hat</strong> lists 39 affected product entries for its own streams with 33 solutions, rates it Important, and maps it to CWE-123.</li>
<li><strong>Siemens</strong> lists 10 affected product entries for its own devices.</li>
<li><strong>CISA</strong> says exploitation is active, records the KEV date, and maps it to CWE-456, a different CWE from Red Hat&rsquo;s.</li>
</ul>
<p>None of these overrides the others. A kernel team, a RHEL fleet owner and an OT engineer running Siemens PLCs each need a different slice of the same record.</p>
<h2 id="a-tested-script-who-said-what">A tested script: who said what</h2>
<p>This Python script prints one block per container. It uses only the standard library and was run with Python 3.13 against the record above and against the 2,000-record sample.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Print who said what in a CVE JSON 5 record: one line per container and field.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">summarize</span>(container):
</span></span><span style="display:flex;"><span>    out <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> container<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> k, v <span style="color:#f92672">in</span> m<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> k<span style="color:#f92672">.</span>startswith(<span style="color:#e6db74">&#34;cvss&#34;</span>):
</span></span><span style="display:flex;"><span>                out<span style="color:#f92672">.</span>setdefault(<span style="color:#e6db74">&#34;cvss&#34;</span>, [])<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;version&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;baseScore&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>v[<span style="color:#e6db74">&#39;vectorString&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">elif</span> k <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;other&#34;</span>:
</span></span><span style="display:flex;"><span>                t, c <span style="color:#f92672">=</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;type&#34;</span>), v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;content&#34;</span>, {})
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> t <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;ssvc&#34;</span>:
</span></span><span style="display:flex;"><span>                    out[<span style="color:#e6db74">&#34;ssvc&#34;</span>] <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;, &#34;</span><span style="color:#f92672">.</span>join(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>k2<span style="color:#e6db74">}</span><span style="color:#e6db74">=</span><span style="color:#e6db74">{</span>v2<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span> <span style="color:#66d9ef">for</span> o <span style="color:#f92672">in</span> c[<span style="color:#e6db74">&#34;options&#34;</span>] <span style="color:#66d9ef">for</span> k2, v2 <span style="color:#f92672">in</span> o<span style="color:#f92672">.</span>items())
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">elif</span> t <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;kev&#34;</span>:
</span></span><span style="display:flex;"><span>                    out[<span style="color:#e6db74">&#34;kev&#34;</span>] <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;dateAdded=</span><span style="color:#e6db74">{</span>c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;dateAdded&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>                    out<span style="color:#f92672">.</span>setdefault(<span style="color:#e6db74">&#34;other&#34;</span>, [])<span style="color:#f92672">.</span>append(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>t<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;value&#39;</span>, c)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    cwes <span style="color:#f92672">=</span> [d[<span style="color:#e6db74">&#34;cweId&#34;</span>] <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> container<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;problemTypes&#34;</span>, []) <span style="color:#66d9ef">for</span> d <span style="color:#f92672">in</span> p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;descriptions&#34;</span>, []) <span style="color:#66d9ef">if</span> d<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;cweId&#34;</span>)]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> cwes: out[<span style="color:#e6db74">&#34;cwe&#34;</span>] <span style="color:#f92672">=</span> cwes
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> container<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;affected&#34;</span>): out[<span style="color:#e6db74">&#34;affected_products&#34;</span>] <span style="color:#f92672">=</span> len(container[<span style="color:#e6db74">&#34;affected&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> container<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;references&#34;</span>): out[<span style="color:#e6db74">&#34;references&#34;</span>] <span style="color:#f92672">=</span> len(container[<span style="color:#e6db74">&#34;references&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> k <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;solutions&#34;</span>, <span style="color:#e6db74">&#34;workarounds&#34;</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> container<span style="color:#f92672">.</span>get(k): out[k] <span style="color:#f92672">=</span> len(container[k])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;x_adpType&#34;</span> <span style="color:#f92672">in</span> container: out[<span style="color:#e6db74">&#34;x_adpType&#34;</span>] <span style="color:#f92672">=</span> container[<span style="color:#e6db74">&#34;x_adpType&#34;</span>]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>c <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;containers&#34;</span>]
</span></span><span style="display:flex;"><span>rows <span style="color:#f92672">=</span> [(<span style="color:#e6db74">&#34;cna&#34;</span>, c[<span style="color:#e6db74">&#34;cna&#34;</span>])] <span style="color:#f92672">+</span> [(<span style="color:#e6db74">&#34;adp&#34;</span>, a) <span style="color:#66d9ef">for</span> a <span style="color:#f92672">in</span> c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, [])]
</span></span><span style="display:flex;"><span>print(rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;cveId&#34;</span>], rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;state&#34;</span>], <span style="color:#e6db74">&#34;dataVersion&#34;</span>, rec<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;dataVersion&#34;</span>))
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> kind, cont <span style="color:#f92672">in</span> rows:
</span></span><span style="display:flex;"><span>    pm <span style="color:#f92672">=</span> cont[<span style="color:#e6db74">&#34;providerMetadata&#34;</span>]
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#ae81ff">\n</span><span style="color:#e6db74">[</span><span style="color:#e6db74">{</span>kind<span style="color:#e6db74">}</span><span style="color:#e6db74">] </span><span style="color:#e6db74">{</span>pm<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;shortName&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74"> (updated </span><span style="color:#e6db74">{</span>pm<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;dateUpdated&#39;</span>, <span style="color:#e6db74">&#39;?&#39;</span>)[:<span style="color:#ae81ff">10</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> k, v <span style="color:#f92672">in</span> summarize(cont)<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;  </span><span style="color:#e6db74">{</span>k<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>Run it with <code>python3 containers.py CVE-2026-46300.json</code>, using a record file from the cvelistV5 repository (<code>cves/2026/46xxx/CVE-2026-46300.json</code>). It is a reading aid, not a merge engine; the next section is about merging.</p>
<h2 id="how-to-merge-adp-data-without-lying-to-yourself">How to merge ADP data without lying to yourself</h2>
<p>There is no Program-defined precedence between ADPs. The following policy is our editorial recommendation, built on the rules quoted above:</p>
<ol>
<li><strong>Always read <code>cna</code> and the <code>CVE</code> container.</strong> The README calls them the minimum. Union and deduplicate references across both.</li>
<li><strong>Store values with their provider.</strong> Keep <code>cvss_cna</code>, <code>cvss_cisa_adp</code> and <code>cvss_supplier</code> side by side rather than one <code>cvss</code> column. The same goes for CWE.</li>
<li><strong>Prefer the CNA for scoring, fall back to CISA-ADP.</strong> Use the CNA&rsquo;s CVSS when present; use CISA-ADP&rsquo;s when the CNA gave none (the common case in our sample); flag disagreements for a human when both exist.</li>
<li><strong>Take SSVC and the KEV date from CISA-ADP.</strong> No other container in our sample carried them. Then confirm KEV details against the catalog.</li>
<li><strong>Scope supplier data to the supplier.</strong> Use redhat-SADP <code>affected</code> entries only to match Red Hat packages, and siemens-SADP entries only for Siemens products. Never use them to clear or confirm an upstream version.</li>
<li><strong>Do not ignore unknown publishers.</strong> Log any new <code>shortName</code> so a person can decide how to treat it, rather than dropping it silently.</li>
<li><strong>Check the record state first.</strong> A rejected record has no ADP data to act on; see our guide to <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">CVE record states</a>.</li>
</ol>
<p>These steps fit into the record-reading stage of our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a>, which takes a CVE from record to patch ticket.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Reading only <code>containers.cna</code>.</strong> You lose CISA&rsquo;s SSVC and KEV signals, about one in five CVSS scores in our 2026 sample, and Program-added references.</li>
<li><strong>Taking <code>adp[0]</code>.</strong> Container order is not a precedence order. In CVE-2026-46300, CISA-ADP came last.</li>
<li><strong>Merging CWE lists without provenance.</strong> Red Hat and CISA mapped the same CVE to different CWEs. A merged list hides that they disagree.</li>
<li><strong>Matching upstream versions against supplier entries.</strong> RPM release strings and Siemens firmware versions are not kernel versions.</li>
<li><strong>Hard-coding today&rsquo;s publisher list.</strong> Supplier ADP is new. More publishers will appear.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The sample covers 2,000 of 84,066 2026 record files in cvelistV5 on November 3, 2026, chosen at random, and 2026 records only. Shares will differ for older years, since the README says the CISA-ADP container was launched on June 4 (2024) and applied back to February 2024. We could not load cve.org&rsquo;s ADP pages (they render with JavaScript), so the description of supplier ADPs relies on Siemens&rsquo; own page and on the records themselves. <code>x_adpType</code> is an extension field, not part of the schema. We did not check how NVD or commercial scanners consume ADP data.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>One cna container, multiple adp containers; ADP requires only providerMetadata; x_ extensions allowed</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE JSON schema, CVE_Record_Format.json</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE Program Container: shortName CVE, Program references after 7/31/2024, CNA + Program container are the minimum</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/README.md">cvelistV5 README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA-ADP provides SSVC, KEV data and Vulnrichment; launched June 4</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/README.md">cvelistV5 README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA fills CVSS/CWE where the CNA has not and does not overwrite CNA data</td>
          <td><a href="https://github.com/cisagov/vulnrichment/blob/develop/README.md">CISA Vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Supplier ADP purpose and April 2026 introduction</td>
          <td><a href="https://www.siemens.com/en-gb/content/cert-services-sadp/">Siemens: CERT services, SADP</a></td>
          <td>Qualified (vendor page)</td>
      </tr>
      <tr>
          <td>Sample shares (93.9%, 2.2%, 1.2%, 19.8%, 17.1%, 5 KEV blocks; 381/389, 326/336)</td>
          <td>Our count over 2,000 random files from <a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a> at commit 1a44d59</td>
          <td>Verified (code)</td>
      </tr>
      <tr>
          <td>CVE-2026-46300 container contents</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/46xxx/CVE-2026-46300.json">cvelistV5 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-46300 KEV dateAdded 2026-10-28</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE Affected Versions: lessThan, defaultStatus, versionType</title><link>https://firewallsync.com/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/</link><guid>https://firewallsync.com/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/</guid><pubDate>Mon, 02 Nov 2026 04:40:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>AppSec &amp; APIs</category><description>The affected block in a CVE record is a small algorithm, not a list. Here is how version, lessThan, lessThanOrEqual, changes, versionType and defaultStatus combine, what a sample of 1,000 2026 CVE records actually contains, and a tested Python evaluator that refuses version types it cannot order.</description><enclosure url="https://firewallsync.com/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/cover_hu16366087011500977870.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/cover_hu16366087011500977870.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/cover_hu16366087011500977870.jpg" alt="Macro close-up of the steel scales of a vernier caliper, with inch and centimetre graduations and a sliding vernier scale" width="1200" height="630"></p><p><strong>Scope.</strong> This is a specification explainer with measurements from public data. We read the CVE JSON schema (version 5.2.0, CVEProject/cve-schema main branch) and its version-encoding guide, sampled 1,000 random 2026 records from the cvelistV5 repository at commit 451e97a (November 2, 2026), and ran the Python script below against real records. We did not test commercial scanners.</p>
<h2 id="the-short-answer">The short answer</h2>
<p>A CVE record&rsquo;s <code>affected</code> block does not hand you a list of bad versions. For each product it gives a small decision procedure:</p>
<ol>
<li>Walk the <code>versions</code> list in order.</li>
<li>An entry with only <code>version</code> and <code>status</code> matches that exact version.</li>
<li>An entry with <code>versionType</code> plus <code>lessThan</code> or <code>lessThanOrEqual</code> matches a range, and an optional <code>changes</code> list can flip the status partway through it.</li>
<li>The first matching entry wins. If nothing matches, the answer is <code>defaultStatus</code>, and if that is missing, the answer is <code>unknown</code>.</li>
</ol>
<p>The possible answers are <code>affected</code>, <code>unaffected</code> and <code>unknown</code>. Your tooling needs a fourth, &ldquo;I cannot order these versions&rdquo;, because the record does not always give you enough to decide.</p>
<h2 id="where-the-fields-live">Where the fields live</h2>
<p>The CNA (CVE Numbering Authority) writes <code>containers.cna.affected</code>, an array of product objects. The schema requires each product to be identified by either <code>vendor</code> and <code>product</code> or <code>collectionURL</code> and <code>packageName</code>, and to carry <code>versions</code>, <code>defaultStatus</code> or both. A 5.2 record can also carry a <code>packageURL</code> (a purl, which the schema says must not include a version).</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;vendor&#34;</span>: <span style="color:#e6db74">&#34;GitoxideLabs&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;product&#34;</span>: <span style="color:#e6db74">&#34;gitoxide&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;defaultStatus&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;packageURL&#34;</span>: <span style="color:#e6db74">&#34;pkg:cargo/gix-fs&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;versions&#34;</span>: [
</span></span><span style="display:flex;"><span>    {<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;0&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;0.23.0&#34;</span>},
</span></span><span style="display:flex;"><span>    {<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;0.23.0&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}
</span></span><span style="display:flex;"><span>  ]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>That is the product entry from CVE-2026-100419 (a gix-fs symlink worktree escape), shown in full. Read it as: semantic versions from 0 up to but not including 0.23.0 are affected, 0.23.0 is unaffected, and anything else is unaffected by default.</p>
<h2 id="the-fields-one-by-one">The fields, one by one</h2>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>What the schema says</th>
          <th>What it means in practice</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>version</code></td>
          <td>The single version, or the start of a range. &ldquo;By convention, typically 0 denotes the earliest possible version.&rdquo;</td>
          <td><code>&quot;version&quot;: &quot;0&quot;</code> means &ldquo;no lower bound&rdquo;</td>
      </tr>
      <tr>
          <td><code>status</code></td>
          <td>One of <code>affected</code>, <code>unaffected</code>, <code>unknown</code></td>
          <td><code>unknown</code> can mean nobody investigated or the vendor did not disclose</td>
      </tr>
      <tr>
          <td><code>versionType</code></td>
          <td>&ldquo;The version numbering system used for specifying the range&rdquo;; examples include <code>semver</code>, <code>maven</code>, <code>python</code>, <code>rpm</code>, <code>git</code>, <code>custom</code></td>
          <td>Picks the comparison rules. Required for a range</td>
      </tr>
      <tr>
          <td><code>lessThan</code></td>
          <td>&ldquo;The non-inclusive upper limit of the range. This is the least version NOT in the range.&rdquo;</td>
          <td>Usually the first fixed release</td>
      </tr>
      <tr>
          <td><code>lessThanOrEqual</code></td>
          <td>&ldquo;The inclusive upper limit of the range&rdquo;</td>
          <td>Usually the last vulnerable release</td>
      </tr>
      <tr>
          <td><code>changes</code></td>
          <td>Status changes inside a range, each with <code>at</code> and <code>status</code>; &ldquo;clients must re-sort the list themselves rather than assume it is sorted&rdquo;</td>
          <td>Lets one entry say &ldquo;affected, then fixed from X&rdquo;</td>
      </tr>
      <tr>
          <td><code>defaultStatus</code></td>
          <td>&ldquo;The default status for versions that are not otherwise listed&hellip; If not specified, defaultStatus defaults to &lsquo;unknown&rsquo;.&rdquo;</td>
          <td>The answer for every version no entry matched</td>
      </tr>
  </tbody>
</table>
<p>The upper limit can end in <code>*</code>, which the schema defines as &ldquo;an arbitrarily large number in the version ordering&rdquo;. <code>{version: 1.0, lessThan: 1.*}</code> describes the whole 1.x branch, and <code>lessThan: *</code> means &ldquo;and everything after&rdquo;.</p>
<h3 id="why-versiontype-is-not-optional">Why versionType is not optional</h3>
<p>The schema guide explains that &ldquo;there is no single definition of &rsquo;less than&rsquo; for versions&rdquo; and gives the example that in semantic versioning <code>1.0.0-cr1</code> sorts before <code>1.0.0-m1</code>, while in Maven the opposite is true. If your code compares every version string with one generic routine, it will be wrong for some ecosystems.</p>
<p>The guide is blunt about <code>custom</code>: it &ldquo;means that the version status algorithm cannot be executed, so its use is discouraged&rdquo;. It exists to convert older data.</p>
<h2 id="three-records-three-encodings">Three records, three encodings</h2>
<p><strong>Allow-list style (gix-fs, CVE-2026-100419).</strong> The range is affected and <code>defaultStatus</code> is <code>unaffected</code>. Version 0.22.1 is affected, 0.23.0 is unaffected by exact match, and 0.25.4 is unaffected by default.</p>
<p><strong>List without a default (quickjs, CVE-2026-0821).</strong> The CNA lists eleven single versions from 0.1 to 0.11.0 as affected and sets no <code>defaultStatus</code>. Version 0.10 is affected. Version 0.12.0 is <code>unknown</code>: the record makes no claim about it. A naive matcher that treats &ldquo;not listed&rdquo; as &ldquo;not affected&rdquo; would clear it.</p>
<p><strong>Deny-list style (Linux, CVE-2026-100075).</strong> The second product entry sets <code>&quot;defaultStatus&quot;: &quot;affected&quot;</code> and lists unaffected ranges per stable branch:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;4.7&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;affected&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;0&#34;</span>, <span style="color:#f92672">&#34;lessThan&#34;</span>: <span style="color:#e6db74">&#34;4.7&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;6.6.157&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;6.6.*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;semver&#34;</span>}<span style="color:#960050;background-color:#1e0010">,</span>
</span></span><span style="display:flex;"><span>{<span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;7.3&#34;</span>, <span style="color:#f92672">&#34;lessThanOrEqual&#34;</span>: <span style="color:#e6db74">&#34;*&#34;</span>, <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;unaffected&#34;</span>, <span style="color:#f92672">&#34;versionType&#34;</span>: <span style="color:#e6db74">&#34;original_commit_for_fix&#34;</span>}
</span></span></code></pre></div><p>(Excerpt; the record also lists 5.10, 5.15, 6.1, 6.12, 6.18 and 7.2 branches.) Kernels before 4.7 are unaffected, 6.6.157 and later 6.6 releases are unaffected, and anything not covered, such as 6.6.150 or a 5.4 kernel, falls through to <code>affected</code>. The first product entry in the same record describes the same flaw with <code>versionType: git</code> commit ranges, which you cannot evaluate without the kernel&rsquo;s commit history.</p>
<h2 id="what-a-sample-of-2026-records-contains">What a sample of 2026 records contains</h2>
<p>We drew 1,000 CVE IDs at random (seeded) from the 83,665 record files under <code>cves/2026</code> at cvelistV5 commit 451e97a. 16 were <code>REJECTED</code>; the 984 published records held 2,061 product entries, 4,555 version entries and 2,831 ranges. Counts were computed with a Python script over the JSON.</p>
<table>
  <thead>
      <tr>
          <th>Measure</th>
          <th>Count</th>
          <th>Share</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Product entries with no <code>defaultStatus</code> (so <code>unknown</code> applies)</td>
          <td>1,130 of 2,061</td>
          <td>54.8%</td>
      </tr>
      <tr>
          <td><code>defaultStatus: unaffected</code></td>
          <td>585</td>
          <td>28.4%</td>
      </tr>
      <tr>
          <td><code>defaultStatus: affected</code></td>
          <td>311</td>
          <td>15.1%</td>
      </tr>
      <tr>
          <td>Product entries with only single versions and no <code>defaultStatus</code></td>
          <td>315</td>
          <td>15.3%</td>
      </tr>
      <tr>
          <td>Version entries that are single versions / <code>lessThan</code> ranges / <code>lessThanOrEqual</code> ranges</td>
          <td>1,724 / 1,929 / 902 of 4,555</td>
          <td>37.8% / 42.3% / 19.8%</td>
      </tr>
      <tr>
          <td>Ranges with <code>versionType: custom</code></td>
          <td>1,006 of 2,831</td>
          <td>35.5%</td>
      </tr>
      <tr>
          <td>Ranges with <code>semver</code> / <code>git</code> / <code>rpm</code></td>
          <td>971 / 560 / 171</td>
          <td>34.3% / 19.8% / 6.0%</td>
      </tr>
      <tr>
          <td>Ranges with a <code>*</code> in the upper limit</td>
          <td>717</td>
          <td>25.3%</td>
      </tr>
      <tr>
          <td>Version entries using <code>changes</code></td>
          <td>48</td>
          <td>1.1%</td>
      </tr>
      <tr>
          <td>Published records with a <code>cpeApplicability</code> block</td>
          <td>290 of 984</td>
          <td>29.5%</td>
      </tr>
      <tr>
          <td>Published records with a <code>packageURL</code> on any product</td>
          <td>61</td>
          <td>6.2%</td>
      </tr>
  </tbody>
</table>
<p>Two numbers matter most for automation. More than half of product entries leave unlisted versions as <code>unknown</code>, and about a third of ranges use <code>custom</code>, so a generic comparator cannot decide them. This is one sample at one point in time; the mix varies by CNA, and a different year or sample would give different shares.</p>
<h2 id="a-tested-evaluator">A tested evaluator</h2>
<p>The script implements the schema&rsquo;s algorithm. It handles dotted numeric versions (so 1.2.10 sorts after 1.2.9) and <code>*</code>, and it refuses any range whose <code>versionType</code> it has not been told how to order. It does not implement semver pre-release ordering, RPM epochs or Maven qualifiers; treat it as a reference, not a drop-in matcher. Python 3.10 or later, standard library only.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Decide affected / unaffected / unknown for one product version from a CVE record,
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">following the algorithm in the CVE JSON 5 schema (product.versions + defaultStatus).
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: affected_status.py RECORD.json VERSION [vendor/product filter] [--numeric TYPE,...]
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>SUPPORTED <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;semver&#34;</span>}  <span style="color:#75715e"># extend with real comparators (rpm, maven, python...) before use</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">key</span>(v):
</span></span><span style="display:flex;"><span>    parts <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> v<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;.&#34;</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> p <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;*&#34;</span>:
</span></span><span style="display:flex;"><span>            parts<span style="color:#f92672">.</span>append(float(<span style="color:#e6db74">&#34;inf&#34;</span>))
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">elif</span> p<span style="color:#f92672">.</span>isdigit():
</span></span><span style="display:flex;"><span>            parts<span style="color:#f92672">.</span>append(int(p))
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">raise</span> <span style="color:#a6e22e">ValueError</span>(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;non-numeric version part </span><span style="color:#e6db74">{</span>p<span style="color:#e6db74">!r}</span><span style="color:#e6db74"> in </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">!r}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> parts
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cmp</span>(a, b):
</span></span><span style="display:flex;"><span>    ka, kb <span style="color:#f92672">=</span> key(a), key(b)
</span></span><span style="display:flex;"><span>    n <span style="color:#f92672">=</span> max(len(ka), len(kb))
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># pad: a missing part is 0, except after a wildcard, which stays infinite</span>
</span></span><span style="display:flex;"><span>    ka <span style="color:#f92672">+=</span> [float(<span style="color:#e6db74">&#34;inf&#34;</span>) <span style="color:#66d9ef">if</span> ka <span style="color:#f92672">and</span> ka[<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>] <span style="color:#f92672">==</span> float(<span style="color:#e6db74">&#34;inf&#34;</span>) <span style="color:#66d9ef">else</span> <span style="color:#ae81ff">0</span>] <span style="color:#f92672">*</span> (n <span style="color:#f92672">-</span> len(ka))
</span></span><span style="display:flex;"><span>    kb <span style="color:#f92672">+=</span> [float(<span style="color:#e6db74">&#34;inf&#34;</span>) <span style="color:#66d9ef">if</span> kb <span style="color:#f92672">and</span> kb[<span style="color:#f92672">-</span><span style="color:#ae81ff">1</span>] <span style="color:#f92672">==</span> float(<span style="color:#e6db74">&#34;inf&#34;</span>) <span style="color:#66d9ef">else</span> <span style="color:#ae81ff">0</span>] <span style="color:#f92672">*</span> (n <span style="color:#f92672">-</span> len(kb))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (ka <span style="color:#f92672">&gt;</span> kb) <span style="color:#f92672">-</span> (ka <span style="color:#f92672">&lt;</span> kb)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">status</span>(product, v):
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> e <span style="color:#f92672">in</span> product<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versions&#34;</span>, []):
</span></span><span style="display:flex;"><span>        is_range <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;lessThan&#34;</span> <span style="color:#f92672">in</span> e <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span> <span style="color:#f92672">in</span> e
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> is_range:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> v <span style="color:#f92672">==</span> e[<span style="color:#e6db74">&#34;version&#34;</span>]:
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">return</span> e[<span style="color:#e6db74">&#34;status&#34;</span>], <span style="color:#e6db74">&#34;exact match&#34;</span>
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        vt <span style="color:#f92672">=</span> e<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versionType&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> vt <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> SUPPORTED:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;undetermined&#34;</span>, <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;range uses versionType </span><span style="color:#e6db74">{</span>vt<span style="color:#e6db74">!r}</span><span style="color:#e6db74">; needs that ordering&#34;</span>
</span></span><span style="display:flex;"><span>        lo_ok <span style="color:#f92672">=</span> cmp(e[<span style="color:#e6db74">&#34;version&#34;</span>], v) <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">0</span>
</span></span><span style="display:flex;"><span>        hi_ok <span style="color:#f92672">=</span> (cmp(v, e[<span style="color:#e6db74">&#34;lessThan&#34;</span>]) <span style="color:#f92672">&lt;</span> <span style="color:#ae81ff">0</span>) <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;lessThan&#34;</span> <span style="color:#f92672">in</span> e <span style="color:#66d9ef">else</span> (cmp(v, e[<span style="color:#e6db74">&#34;lessThanOrEqual&#34;</span>]) <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">0</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> lo_ok <span style="color:#f92672">and</span> hi_ok:
</span></span><span style="display:flex;"><span>            s <span style="color:#f92672">=</span> e[<span style="color:#e6db74">&#34;status&#34;</span>]
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> sorted(e<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;changes&#34;</span>, []), key<span style="color:#f92672">=</span><span style="color:#66d9ef">lambda</span> c: key(c[<span style="color:#e6db74">&#34;at&#34;</span>])):  <span style="color:#75715e"># schema: re-sort yourself</span>
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">if</span> cmp(c[<span style="color:#e6db74">&#34;at&#34;</span>], v) <span style="color:#f92672">&lt;=</span> <span style="color:#ae81ff">0</span>:
</span></span><span style="display:flex;"><span>                    s <span style="color:#f92672">=</span> c[<span style="color:#e6db74">&#34;status&#34;</span>]
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> s, <span style="color:#e6db74">&#34;range match&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> product<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;defaultStatus&#34;</span>, <span style="color:#e6db74">&#34;unknown&#34;</span>), <span style="color:#e6db74">&#34;defaultStatus&#34;</span> <span style="color:#f92672">+</span> (<span style="color:#e6db74">&#34;&#34;</span> <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;defaultStatus&#34;</span> <span style="color:#f92672">in</span> product <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34; (absent, so unknown)&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    args <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#e6db74">&#34;--numeric&#34;</span> <span style="color:#f92672">in</span> args:  <span style="color:#75715e"># opt-in: treat these extra versionTypes as dotted numeric</span>
</span></span><span style="display:flex;"><span>        i <span style="color:#f92672">=</span> args<span style="color:#f92672">.</span>index(<span style="color:#e6db74">&#34;--numeric&#34;</span>)
</span></span><span style="display:flex;"><span>        SUPPORTED<span style="color:#f92672">.</span>update(args[i <span style="color:#f92672">+</span> <span style="color:#ae81ff">1</span>]<span style="color:#f92672">.</span>split(<span style="color:#e6db74">&#34;,&#34;</span>))
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">del</span> args[i:i <span style="color:#f92672">+</span> <span style="color:#ae81ff">2</span>]
</span></span><span style="display:flex;"><span>    sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:] <span style="color:#f92672">=</span> args
</span></span><span style="display:flex;"><span>    rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>]))
</span></span><span style="display:flex;"><span>    version <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>]
</span></span><span style="display:flex;"><span>    want <span style="color:#f92672">=</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">3</span>]<span style="color:#f92672">.</span>lower() <span style="color:#66d9ef">if</span> len(sys<span style="color:#f92672">.</span>argv) <span style="color:#f92672">&gt;</span> <span style="color:#ae81ff">3</span> <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;state&#34;</span>] <span style="color:#f92672">!=</span> <span style="color:#e6db74">&#34;PUBLISHED&#34;</span>:
</span></span><span style="display:flex;"><span>        sys<span style="color:#f92672">.</span>exit(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>rec[<span style="color:#e6db74">&#39;cveMetadata&#39;</span>][<span style="color:#e6db74">&#39;cveId&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> is </span><span style="color:#e6db74">{</span>rec[<span style="color:#e6db74">&#39;cveMetadata&#39;</span>][<span style="color:#e6db74">&#39;state&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> i, p <span style="color:#f92672">in</span> enumerate(rec[<span style="color:#e6db74">&#34;containers&#34;</span>][<span style="color:#e6db74">&#34;cna&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;affected&#34;</span>, [])):
</span></span><span style="display:flex;"><span>        name <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;vendor&#39;</span>, <span style="color:#e6db74">&#39;&#39;</span>)<span style="color:#e6db74">}</span><span style="color:#e6db74">/</span><span style="color:#e6db74">{</span>p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;product&#39;</span>, p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#39;packageName&#39;</span>, <span style="color:#e6db74">&#39;&#39;</span>))<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> want <span style="color:#f92672">and</span> want <span style="color:#f92672">not</span> <span style="color:#f92672">in</span> name<span style="color:#f92672">.</span>lower():
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>rec[<span style="color:#e6db74">&#39;cveMetadata&#39;</span>][<span style="color:#e6db74">&#39;cveId&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74"> [</span><span style="color:#e6db74">{</span>i<span style="color:#e6db74">}</span><span style="color:#e6db74">] </span><span style="color:#e6db74">{</span>name<span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>version<span style="color:#e6db74">}</span><span style="color:#e6db74">: %s (%s)&#34;</span> <span style="color:#f92672">%</span> status(p, version))
</span></span></code></pre></div><p>We ran it with Python 3.13 on November 2, 2026 against records downloaded from cvelistV5 (<code>raw.githubusercontent.com/CVEProject/cvelistV5/&lt;commit&gt;/cves/2026/&lt;block&gt;/&lt;CVE&gt;.json</code>). Selected output:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2026-100419 [0] GitoxideLabs/gitoxide 0.22.1: affected (range match)
</span></span><span style="display:flex;"><span>CVE-2026-100419 [0] GitoxideLabs/gitoxide 0.23.0: unaffected (exact match)
</span></span><span style="display:flex;"><span>CVE-2026-100419 [0] GitoxideLabs/gitoxide 0.25.4: unaffected (defaultStatus)
</span></span><span style="display:flex;"><span>CVE-2026-0821 [0] quickjs-ng/quickjs 0.12.0: unknown (defaultStatus (absent, so unknown))
</span></span><span style="display:flex;"><span>CVE-2026-100075 [0] Linux/Linux 6.6.150: undetermined (range uses versionType &#39;git&#39;; needs that ordering)
</span></span><span style="display:flex;"><span>CVE-2026-100075 [1] Linux/Linux 6.6.150: undetermined (range uses versionType &#39;original_commit_for_fix&#39;; needs that ordering)
</span></span><span style="display:flex;"><span>CVE-2026-103345 [0] Shamim Rajani/Pie Register 3.8.4.13: undetermined (range uses versionType &#39;custom&#39;; needs that ordering)
</span></span><span style="display:flex;"><span>CVE-2026-23915 is REJECTED
</span></span></code></pre></div><p>The Linux result is the instructive one. For 6.6.150 the evaluator reaches the <code>7.3 ... *</code> entry, whose <code>versionType</code> is <code>original_commit_for_fix</code>, a value that is not among the schema&rsquo;s examples. A strict implementation stops there. If you decide, for this CNA&rsquo;s data, that those values are ordinary kernel version numbers, you can opt in with <code>--numeric original_commit_for_fix</code>, and the second product entry then gives 4.4.302 unaffected, 4.7 affected, 6.6.150 affected (by <code>defaultStatus</code>), 6.6.157 unaffected and 7.3 unaffected. That opt-in is our judgment about one CNA&rsquo;s encoding, not something the schema states.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Treating &ldquo;not listed&rdquo; as &ldquo;not affected&rdquo;.</strong> Without <code>defaultStatus: unaffected</code>, unlisted versions are <code>unknown</code>.</li>
<li><strong>Off-by-one on the upper bound.</strong> <code>lessThan: 0.23.0</code> excludes 0.23.0; <code>lessThanOrEqual: 3.8.4.13</code> includes 3.8.4.13.</li>
<li><strong>Comparing strings.</strong> As text, &ldquo;0.10&rdquo; sorts before &ldquo;0.9&rdquo;. Parse versions with the ecosystem&rsquo;s rules.</li>
<li><strong>Assuming <code>changes</code> is sorted.</strong> The schema tells clients to re-sort.</li>
<li><strong>Reading only the first product entry.</strong> The same record can describe a product twice, as the Linux record does with git ranges and version ranges.</li>
<li><strong>Ignoring the record state.</strong> A <code>REJECTED</code> record has no <code>affected</code> block to evaluate; see <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">CVE record states</a>.</li>
</ul>
<h2 id="where-this-fits-in-triage">Where this fits in triage</h2>
<p>Version matching answers &ldquo;do we run an affected version?&rdquo;, which comes before any priority signal. If the answer is <code>unknown</code> or undetermined, check the vendor advisory, or a supplier VEX statement that says whether the product is affected (see <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">how to read a VEX not-affected statement</a>). An SBOM with accurate component versions is what makes this lookup possible at scale; our summary of the <a href="/posts/sbom-minimum-elements-2026-what-cisa-and-partners-changed-from-the-2021-ntia-baseline/">2026 SBOM minimum elements</a> covers what to ask suppliers for. Once a finding is confirmed as affected, our <a href="/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/">vulnerability prioritization workflow</a> takes it from record to ticket using KEV, EPSS and SSVC.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The sample covers 2026 records only and reflects the repository on November 2, 2026; CNAs update records, so shares will drift. Our evaluator is a reference implementation of the schema&rsquo;s algorithm for numeric versions, not a full comparator for semver pre-releases, RPM, Maven, Python or git ranges. We did not examine ADP containers, which can carry their own <code>affected</code> data (15 of the 984 sampled records had one), or NVD&rsquo;s separate CPE configurations.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Field definitions, the status algorithm, <code>defaultStatus</code> defaults to unknown, <code>*</code> wildcard, <code>changes</code> must be re-sorted, <code>packageURL</code> must not include a version</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE JSON schema, CVE_Record_Format.json (v5.2.0)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td><code>version: 0</code> convention, Maven vs semver ordering example, <code>custom</code> means the algorithm cannot be executed</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/docs/versions.md">CVE schema version-encoding guide (versions.md)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Example records CVE-2026-100419, CVE-2026-0821, CVE-2026-100075, CVE-2026-103345, CVE-2026-23915</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">cvelistV5 repository</a>, commit 451e97a</td>
          <td>Verified (read November 2, 2026)</td>
      </tr>
      <tr>
          <td>Sample counts and percentages</td>
          <td>Our analysis of 1,000 randomly sampled 2026 records from cvelistV5 at commit 451e97a</td>
          <td>Verified (computed with code); a sample, not a census</td>
      </tr>
      <tr>
          <td>Evaluator output</td>
          <td>Script above, run with Python 3.13</td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Vulnerability Prioritization Workflow: CVE Record to Ticket</title><link>https://firewallsync.com/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/</link><guid>https://firewallsync.com/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/</guid><pubDate>Mon, 02 Nov 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>A six-step patch-decision flow that takes a scanner finding from CVE record to ticket: check the record state, confirm you are affected, apply supplier VEX, read KEV and CISA's SSVC values, add EPSS, then let your exposure decide. Includes a tested Python script that writes the ticket fields.</description><enclosure url="https://firewallsync.com/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/cover_hu7197188011340061090.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/cover_hu7197188011340061090.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/vulnerability-prioritization-workflow-cve-record-to-patch-ticket/cover_hu7197188011340061090.jpg" alt="View from the side of a passenger train of a railway switch where the tracks split, with a red point indicator, dry golden grassland and brown mountains behind" width="1200" height="630"></p><p><strong>Scope.</strong> This is a workflow guide built on public specifications and feeds, plus one script we ran. On November 2, 2026 we read the CVE JSON schema (5.2.0), CISA&rsquo;s Vulnrichment README, CERT/CC&rsquo;s SSVC deployer documentation and decision table, the OpenVEX specification and the EPSS repository README, and ran the script below against the KEV catalog (version 2026.10.30), the EPSS file scored 2026-11-01 and CVE records from cvelistV5. The assets, exposure ratings and the VEX document in the test are invented for the demonstration. The policy choices in the script are ours, not a standard.</p>
<h2 id="the-workflow-in-one-table">The workflow in one table</h2>
<p>Our comparison of <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS, EPSS and KEV</a> explains what each signal measures. This guide is the plumbing: the order in which a finding from a scanner or SBOM passes through the data, and what lands in the ticket.</p>
<table>
  <thead>
      <tr>
          <th>Step</th>
          <th>Question</th>
          <th>Data you read</th>
          <th>Possible outcome</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>1</td>
          <td>Is the CVE record valid?</td>
          <td><code>cveMetadata.state</code> in the CVE record</td>
          <td>Close if <code>REJECTED</code></td>
      </tr>
      <tr>
          <td>2</td>
          <td>Do we run an affected version?</td>
          <td><code>containers.cna.affected</code>: <code>versions</code>, <code>defaultStatus</code></td>
          <td>Close if unaffected; escalate if unknown</td>
      </tr>
      <tr>
          <td>3</td>
          <td>Has the supplier said we are not affected?</td>
          <td>VEX statement for this product</td>
          <td>Close with the VEX justification</td>
      </tr>
      <tr>
          <td>4</td>
          <td>Is it being exploited?</td>
          <td>KEV entry; CISA ADP SSVC <code>Exploitation</code></td>
          <td>Sets the Exploitation input</td>
      </tr>
      <tr>
          <td>5</td>
          <td>How likely is exploitation?</td>
          <td>EPSS <code>epss</code>, <code>percentile</code></td>
          <td>Re-orders the non-exploited queue</td>
      </tr>
      <tr>
          <td>6</td>
          <td>How exposed and important is this asset?</td>
          <td>Your inventory: System Exposure, Human Impact</td>
          <td>Priority: Defer, Scheduled, Out-of-cycle, Immediate</td>
      </tr>
  </tbody>
</table>
<p>Steps 1 to 3 remove work. Steps 4 to 6 order what remains.</p>
<h2 id="step-1-check-the-record-state">Step 1: check the record state</h2>
<p>A CVE record is either <code>PUBLISHED</code> or <code>REJECTED</code> (the schema&rsquo;s two states), and published records can carry tags such as <code>disputed</code>. A rejected ID has no affected products to evaluate. In our test run, CVE-2026-23915 is a Fortinet-assigned ID that was rejected on January 20, 2026 with the reason &ldquo;Not used&rdquo;, so the finding closes. Our guide to <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">CVE record states</a> covers the tags and edge cases.</p>
<h2 id="step-2-confirm-you-are-affected">Step 2: confirm you are affected</h2>
<p>The CNA&rsquo;s <code>affected</code> block is an algorithm, not a list: walk <code>versions</code>, take the first match, otherwise use <code>defaultStatus</code>, and if that is missing the schema says the status is <code>unknown</code>. Scanner results that only match a product name skip this. Our explainer on <a href="/posts/cve-affected-versions-lessthan-defaultstatus-versiontype/">CVE affected versions</a> shows how <code>lessThan</code>, <code>defaultStatus</code> and <code>versionType</code> combine, with a tested evaluator. Keep three outcomes apart in the ticket: affected, unaffected, and unknown or undetermined. The last needs a person or the vendor advisory.</p>
<h2 id="step-3-apply-supplier-vex">Step 3: apply supplier VEX</h2>
<p>If the vulnerable component ships inside someone else&rsquo;s product, the supplier may publish a VEX (Vulnerability Exploitability eXchange) statement. OpenVEX defines four statuses: <code>not_affected</code> (&ldquo;No remediation is required regarding this vulnerability&rdquo;), <code>affected</code>, <code>fixed</code> and <code>under_investigation</code>. A <code>not_affected</code> statement must carry a justification or an impact statement.</p>
<p>Two cautions. A VEX statement is about a specific product and version, so match on the product identifier, not on the CVE alone; a <code>not_affected</code> statement from an appliance vendor says nothing about your own services that use the same library. And read the justification before closing. Our guide to <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">reading a VEX not-affected statement</a> covers the five justifications and what to question.</p>
<h2 id="step-4-read-the-exploitation-evidence">Step 4: read the exploitation evidence</h2>
<p>Two public sources answer &ldquo;is this being exploited?&rdquo;:</p>
<ul>
<li><strong>KEV.</strong> The <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">CISA KEV JSON feed</a> lists CVEs with evidence of exploitation, with <code>dateAdded</code>, <code>dueDate</code>, <code>knownRansomwareCampaignUse</code> and, on newer entries, <code>forensicTriage</code>. The catalog we used (version 2026.10.30, released October 30, 2026) had 1,766 entries.</li>
<li><strong>CISA&rsquo;s SSVC values in the CVE record.</strong> CISA&rsquo;s Vulnrichment project writes three SSVC decision points (Exploitation, Automatable, Technical Impact) into a CISA-ADP container. Its README says CISA is &ldquo;assessing new and recent CVEs&rdquo;, and that it will not overwrite the CNA&rsquo;s data. Each value set carries a timestamp. Our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment guide</a> shows where they sit and how to pull them with jq.</li>
</ul>
<p>The SSVC documentation describes Exploitation as &ldquo;the present state of exploitation&rdquo;, not a prediction. A <code>none</code> from months ago can be stale, so a KEV listing should outrank an older ADP value. That is the rule our script applies.</p>
<h2 id="step-5-add-epss-for-likelihood">Step 5: add EPSS for likelihood</h2>
<p>For findings with no exploitation evidence, EPSS estimates &ldquo;the probability [0-1] of exploitation in the wild in the next 30 days&rdquo;, and <code>percentile</code> gives the share of scored CVEs with the same or a lower score. The file we used (model v2026.06.15, scored 2026-11-01) held 398,784 CVEs. EPSS is a ranking aid within the non-exploited queue. It does not replace step 4: in our run the KEV-listed FortiMail flaw scored 0.04573. See the <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed guide</a> for the columns, model-version breaks and a scanner join.</p>
<h2 id="step-6-let-exposure-and-impact-decide">Step 6: let exposure and impact decide</h2>
<p>CERT/CC&rsquo;s SSVC deployer model maps four inputs to a patch priority:</p>
<table>
  <thead>
      <tr>
          <th>Input</th>
          <th>Values</th>
          <th>Who supplies it</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Exploitation</td>
          <td>none, public poc, active</td>
          <td>CISA ADP or KEV, or your own research</td>
      </tr>
      <tr>
          <td>System Exposure</td>
          <td>small, controlled, open</td>
          <td>You, per asset</td>
      </tr>
      <tr>
          <td>Automatable</td>
          <td>no, yes</td>
          <td>CISA ADP, or your own analysis</td>
      </tr>
      <tr>
          <td>Human Impact</td>
          <td>low, medium, high, very high</td>
          <td>You, per asset</td>
      </tr>
  </tbody>
</table>
<p>The decision table (<code>deployer_patch_application_priority_1_0_0.csv</code>, 72 rows) returns one of four outcomes, which the deployer documentation defines as: Defer (&ldquo;Do not act at present&rdquo;), Scheduled (&ldquo;Act during regularly scheduled maintenance time&rdquo;), Out-of-cycle (&ldquo;Act more quickly than usual&hellip; during the next available opportunity&rdquo;) and Immediate (&ldquo;Act immediately; focus all resources on applying the fix&rdquo;). In the table, Immediate appears in 3 rows, all with active exploitation and open exposure. Our <a href="/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/">SSVC deployer explainer</a> walks through the inputs, including where Human Impact is easy to get wrong.</p>
<p>One mapping detail: CISA&rsquo;s ADP data writes <code>poc</code>, while the deployer table uses <code>public poc</code>. They are not identical definitions. CISA&rsquo;s tree uses the older PoC value, which also covers privately attested exploitation and widespread hearsay; the newer Public PoC value covers only public exploit code or a well-known method. Our script maps one to the other, which can understate cases CISA marked <code>poc</code> on private evidence.</p>
<h2 id="a-tested-script-that-writes-the-ticket-fields">A tested script that writes the ticket fields</h2>
<p>The script reads a findings CSV, the KEV JSON, the EPSS daily file, the SSVC deployer CSV, a folder of CVE records and an optional OpenVEX file, and prints one ticket row per finding. Python 3.10 or later, standard library only.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Turn scanner findings into patch-decision ticket rows.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Inputs : findings.csv (asset,product,cve,exposure,human_impact)
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">         KEV JSON, EPSS daily CSV (.csv.gz), SSVC deployer table CSV,
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">         a directory of cvelistV5 records, optional OpenVEX file.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Output : tickets.csv on stdout. Python 3.10+, standard library only.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> gzip<span style="color:#f92672">,</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> argparse<span style="color:#f92672">,</span> pathlib
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_kev</span>(path):
</span></span><span style="display:flex;"><span>    data <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(path))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> {v[<span style="color:#e6db74">&#34;cveID&#34;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#f92672">.</span>upper(): v <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> data[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_epss</span>(path):
</span></span><span style="display:flex;"><span>    scores <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> gzip<span style="color:#f92672">.</span>open(path, <span style="color:#e6db74">&#34;rt&#34;</span>) <span style="color:#66d9ef">as</span> fh:
</span></span><span style="display:flex;"><span>        rows <span style="color:#f92672">=</span> (line <span style="color:#66d9ef">for</span> line <span style="color:#f92672">in</span> fh <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> line<span style="color:#f92672">.</span>startswith(<span style="color:#e6db74">&#34;#&#34;</span>))
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(rows):
</span></span><span style="display:flex;"><span>            scores[r[<span style="color:#e6db74">&#34;cve&#34;</span>]] <span style="color:#f92672">=</span> (float(r[<span style="color:#e6db74">&#34;epss&#34;</span>]), float(r[<span style="color:#e6db74">&#34;percentile&#34;</span>]))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> scores
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_deployer</span>(path):
</span></span><span style="display:flex;"><span>    rows <span style="color:#f92672">=</span> list(csv<span style="color:#f92672">.</span>reader(open(path)))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> {tuple(r[<span style="color:#ae81ff">1</span>:<span style="color:#ae81ff">5</span>]): r[<span style="color:#ae81ff">5</span>] <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> rows[<span style="color:#ae81ff">1</span>:]}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_vex</span>(path):
</span></span><span style="display:flex;"><span>    out <span style="color:#f92672">=</span> {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> path:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> s <span style="color:#f92672">in</span> json<span style="color:#f92672">.</span>load(open(path))<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;statements&#34;</span>, []):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> s<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;products&#34;</span>, []):
</span></span><span style="display:flex;"><span>                out[(s[<span style="color:#e6db74">&#34;vulnerability&#34;</span>][<span style="color:#e6db74">&#34;name&#34;</span>]<span style="color:#f92672">.</span>upper(), p[<span style="color:#e6db74">&#34;@id&#34;</span>])] <span style="color:#f92672">=</span> s
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">cisa_ssvc</span>(record):
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;&#34;&#34;Return CISA ADP SSVC values (Exploitation, Automatable, Technical Impact, timestamp).&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> adp <span style="color:#f92672">in</span> record[<span style="color:#e6db74">&#34;containers&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, []):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> m <span style="color:#f92672">in</span> adp<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;metrics&#34;</span>, []):
</span></span><span style="display:flex;"><span>            other <span style="color:#f92672">=</span> m<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;other&#34;</span>, {})
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> other<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;type&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;ssvc&#34;</span>:
</span></span><span style="display:flex;"><span>                c <span style="color:#f92672">=</span> other[<span style="color:#e6db74">&#34;content&#34;</span>]
</span></span><span style="display:flex;"><span>                opts <span style="color:#f92672">=</span> {k: v <span style="color:#66d9ef">for</span> o <span style="color:#f92672">in</span> c[<span style="color:#e6db74">&#34;options&#34;</span>] <span style="color:#66d9ef">for</span> k, v <span style="color:#f92672">in</span> o<span style="color:#f92672">.</span>items()}
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">return</span> (opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Exploitation&#34;</span>), opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Automatable&#34;</span>),
</span></span><span style="display:flex;"><span>                        opts<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;Technical Impact&#34;</span>), c<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;timestamp&#34;</span>))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (<span style="color:#66d9ef">None</span>, <span style="color:#66d9ef">None</span>, <span style="color:#66d9ef">None</span>, <span style="color:#66d9ef">None</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>():
</span></span><span style="display:flex;"><span>    ap <span style="color:#f92672">=</span> argparse<span style="color:#f92672">.</span>ArgumentParser()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> a <span style="color:#f92672">in</span> (<span style="color:#e6db74">&#34;findings&#34;</span>, <span style="color:#e6db74">&#34;kev&#34;</span>, <span style="color:#e6db74">&#34;epss&#34;</span>, <span style="color:#e6db74">&#34;deployer&#34;</span>, <span style="color:#e6db74">&#34;records&#34;</span>):
</span></span><span style="display:flex;"><span>        ap<span style="color:#f92672">.</span>add_argument(a)
</span></span><span style="display:flex;"><span>    ap<span style="color:#f92672">.</span>add_argument(<span style="color:#e6db74">&#34;--vex&#34;</span>)
</span></span><span style="display:flex;"><span>    args <span style="color:#f92672">=</span> ap<span style="color:#f92672">.</span>parse_args()
</span></span><span style="display:flex;"><span>    kev, epss <span style="color:#f92672">=</span> load_kev(args<span style="color:#f92672">.</span>kev), load_epss(args<span style="color:#f92672">.</span>epss)
</span></span><span style="display:flex;"><span>    table, vex <span style="color:#f92672">=</span> load_deployer(args<span style="color:#f92672">.</span>deployer), load_vex(args<span style="color:#f92672">.</span>vex)
</span></span><span style="display:flex;"><span>    out <span style="color:#f92672">=</span> csv<span style="color:#f92672">.</span>writer(sys<span style="color:#f92672">.</span>stdout)
</span></span><span style="display:flex;"><span>    out<span style="color:#f92672">.</span>writerow([<span style="color:#e6db74">&#34;asset&#34;</span>, <span style="color:#e6db74">&#34;cve&#34;</span>, <span style="color:#e6db74">&#34;record_state&#34;</span>, <span style="color:#e6db74">&#34;vex_status&#34;</span>, <span style="color:#e6db74">&#34;kev_dateAdded&#34;</span>,
</span></span><span style="display:flex;"><span>                  <span style="color:#e6db74">&#34;kev_dueDate&#34;</span>, <span style="color:#e6db74">&#34;kev_ransomware&#34;</span>, <span style="color:#e6db74">&#34;cisa_exploitation&#34;</span>, <span style="color:#e6db74">&#34;cisa_automatable&#34;</span>,
</span></span><span style="display:flex;"><span>                  <span style="color:#e6db74">&#34;cisa_technical_impact&#34;</span>, <span style="color:#e6db74">&#34;cisa_ssvc_timestamp&#34;</span>, <span style="color:#e6db74">&#34;epss&#34;</span>, <span style="color:#e6db74">&#34;epss_percentile&#34;</span>,
</span></span><span style="display:flex;"><span>                  <span style="color:#e6db74">&#34;deployer_exploitation&#34;</span>, <span style="color:#e6db74">&#34;deployer_priority&#34;</span>, <span style="color:#e6db74">&#34;next_step&#34;</span>])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> f <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(open(args<span style="color:#f92672">.</span>findings)):
</span></span><span style="display:flex;"><span>        cve <span style="color:#f92672">=</span> f[<span style="color:#e6db74">&#34;cve&#34;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#f92672">.</span>upper()
</span></span><span style="display:flex;"><span>        rec_path <span style="color:#f92672">=</span> pathlib<span style="color:#f92672">.</span>Path(args<span style="color:#f92672">.</span>records) <span style="color:#f92672">/</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">}</span><span style="color:#e6db74">.json&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> rec_path<span style="color:#f92672">.</span>exists():
</span></span><span style="display:flex;"><span>            out<span style="color:#f92672">.</span>writerow([f[<span style="color:#e6db74">&#34;asset&#34;</span>], cve, <span style="color:#e6db74">&#34;NO RECORD&#34;</span>] <span style="color:#f92672">+</span> [<span style="color:#e6db74">&#34;&#34;</span>] <span style="color:#f92672">*</span> <span style="color:#ae81ff">12</span> <span style="color:#f92672">+</span> [<span style="color:#e6db74">&#34;check the ID&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(open(rec_path))
</span></span><span style="display:flex;"><span>        state <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>][<span style="color:#e6db74">&#34;state&#34;</span>]
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> state <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;REJECTED&#34;</span>:
</span></span><span style="display:flex;"><span>            out<span style="color:#f92672">.</span>writerow([f[<span style="color:#e6db74">&#34;asset&#34;</span>], cve, state] <span style="color:#f92672">+</span> [<span style="color:#e6db74">&#34;&#34;</span>] <span style="color:#f92672">*</span> <span style="color:#ae81ff">12</span> <span style="color:#f92672">+</span> [<span style="color:#e6db74">&#34;close: CVE ID rejected&#34;</span>])
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>        v <span style="color:#f92672">=</span> vex<span style="color:#f92672">.</span>get((cve, f<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;product&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)), {})  <span style="color:#75715e"># VEX applies to one product, not every asset</span>
</span></span><span style="display:flex;"><span>        k <span style="color:#f92672">=</span> kev<span style="color:#f92672">.</span>get(cve, {})
</span></span><span style="display:flex;"><span>        exp, auto, ti, ts <span style="color:#f92672">=</span> cisa_ssvc(rec)
</span></span><span style="display:flex;"><span>        e, pct <span style="color:#f92672">=</span> epss<span style="color:#f92672">.</span>get(cve, (<span style="color:#e6db74">&#34;&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>))
</span></span><span style="display:flex;"><span>        <span style="color:#75715e"># Our policy choice: a KEV listing counts as active exploitation, whatever</span>
</span></span><span style="display:flex;"><span>        <span style="color:#75715e"># the (possibly older) CISA ADP snapshot says. CISA writes &#34;poc&#34;; the</span>
</span></span><span style="display:flex;"><span>        <span style="color:#75715e"># deployer table writes &#34;public poc&#34;.</span>
</span></span><span style="display:flex;"><span>        dexp <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;active&#34;</span> <span style="color:#66d9ef">if</span> k <span style="color:#66d9ef">else</span> {<span style="color:#e6db74">&#34;poc&#34;</span>: <span style="color:#e6db74">&#34;public poc&#34;</span>}<span style="color:#f92672">.</span>get(exp, exp)
</span></span><span style="display:flex;"><span>        prio <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> dexp <span style="color:#f92672">and</span> auto:
</span></span><span style="display:flex;"><span>            prio <span style="color:#f92672">=</span> table<span style="color:#f92672">.</span>get((dexp, f[<span style="color:#e6db74">&#34;exposure&#34;</span>], auto, f[<span style="color:#e6db74">&#34;human_impact&#34;</span>]), <span style="color:#e6db74">&#34;&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;status&#34;</span>) <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;not_affected&#34;</span>:
</span></span><span style="display:flex;"><span>            step <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;close: supplier VEX not_affected (&#34;</span> <span style="color:#f92672">+</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;justification&#34;</span>, <span style="color:#e6db74">&#34;no justification&#34;</span>) <span style="color:#f92672">+</span> <span style="color:#e6db74">&#34;)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">elif</span> <span style="color:#f92672">not</span> prio:
</span></span><span style="display:flex;"><span>            step <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;manual triage: no CISA SSVC data&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>            step <span style="color:#f92672">=</span> prio
</span></span><span style="display:flex;"><span>        out<span style="color:#f92672">.</span>writerow([f[<span style="color:#e6db74">&#34;asset&#34;</span>], cve, state, v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;status&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>), k<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;dateAdded&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>),
</span></span><span style="display:flex;"><span>                      k<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;dueDate&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>), k<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;knownRansomwareCampaignUse&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>),
</span></span><span style="display:flex;"><span>                      exp <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;&#34;</span>, auto <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;&#34;</span>, ti <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;&#34;</span>, ts <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;&#34;</span>, e, pct, dexp <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;&#34;</span>, prio, step])
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main()
</span></span></code></pre></div><p>The test input (<code>findings.csv</code>) used invented assets with real CVE IDs:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>asset,product,cve,exposure,human_impact
</span></span><span style="display:flex;"><span>edge-mail-01,fortimail,CVE-2026-104286,open,high
</span></span><span style="display:flex;"><span>k8s-node-17,linux-kernel,CVE-2026-46300,controlled,medium
</span></span><span style="display:flex;"><span>build-runner-03,pkg:cargo/gix-fs@0.22.1,CVE-2026-100419,small,medium
</span></span><span style="display:flex;"><span>billing-api,pkg:maven/org.apache.logging.log4j/log4j-core@2.14.1,CVE-2021-44228,open,very high
</span></span><span style="display:flex;"><span>vendor-appliance-2,pkg:generic/example-appliance@4.2,CVE-2021-44228,open,very high
</span></span><span style="display:flex;"><span>legacy-scan,unknown,CVE-2026-23915,open,low
</span></span></code></pre></div><p>The VEX file is an invented OpenVEX document from an &ldquo;Example Appliance Vendor&rdquo; stating <code>not_affected</code> with justification <code>vulnerable_code_not_present</code> for CVE-2021-44228 in <code>pkg:generic/example-appliance@4.2</code>. We ran:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>python3 patch_decision.py findings.csv known_exploited_vulnerabilities.json <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  epss_scores-2026-11-01.csv.gz deployer_patch_application_priority_1_0_0.csv <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  records/ --vex vendor.openvex.json &gt; tickets.csv
</span></span></code></pre></div><p>The data came from <code>raw.githubusercontent.com/cisagov/kev-data</code>, the <code>empiricalsec/epss_scores</code> repository (2026/epss_scores-2026-11-01.csv.gz), <code>CERTCC/SSVC</code> (data/csv/ssvc/) and <code>CVEProject/cvelistV5</code> at commit 451e97a. Output, with some columns dropped for width:</p>
<table>
  <thead>
      <tr>
          <th>Asset</th>
          <th>CVE</th>
          <th>KEV added</th>
          <th>CISA SSVC (E / A / TI)</th>
          <th>EPSS (percentile)</th>
          <th>Deployer priority</th>
          <th>Next step</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>edge-mail-01</td>
          <td>CVE-2026-104286</td>
          <td>2026-10-01</td>
          <td>active / yes / total</td>
          <td>0.04573 (0.916)</td>
          <td>immediate</td>
          <td>immediate</td>
      </tr>
      <tr>
          <td>k8s-node-17</td>
          <td>CVE-2026-46300</td>
          <td>2026-10-28</td>
          <td>active / no / total</td>
          <td>0.1613 (0.969)</td>
          <td>scheduled</td>
          <td>scheduled</td>
      </tr>
      <tr>
          <td>build-runner-03</td>
          <td>CVE-2026-100419</td>
          <td>-</td>
          <td>poc / no / total</td>
          <td>0.00222 (0.129)</td>
          <td>scheduled</td>
          <td>scheduled</td>
      </tr>
      <tr>
          <td>billing-api</td>
          <td>CVE-2021-44228</td>
          <td>2021-12-10</td>
          <td>active / yes / total</td>
          <td>0.99999 (1.0)</td>
          <td>immediate</td>
          <td>immediate</td>
      </tr>
      <tr>
          <td>vendor-appliance-2</td>
          <td>CVE-2021-44228</td>
          <td>2021-12-10</td>
          <td>active / yes / total</td>
          <td>0.99999 (1.0)</td>
          <td>immediate</td>
          <td>close: supplier VEX not_affected (vulnerable_code_not_present)</td>
      </tr>
      <tr>
          <td>legacy-scan</td>
          <td>CVE-2026-23915</td>
          <td>-</td>
          <td>-</td>
          <td>-</td>
          <td>-</td>
          <td>close: CVE ID rejected</td>
      </tr>
  </tbody>
</table>
<p>Three things in that output are worth noticing:</p>
<ul>
<li><strong>The same CVE got two different answers</strong> (Log4Shell rows) because the VEX statement applies to one product only.</li>
<li><strong>A KEV-listed kernel flaw came out Scheduled.</strong> CVE-2026-46300 is in KEV with <code>forensicTriage: Yes</code> and a three-day federal due date, but with controlled exposure, no automatability and medium Human Impact the deployer table says Scheduled. If you follow CISA&rsquo;s directives, the due date wins. If you do not, the table is telling you the exposure is what drives urgency. Either way, keep both fields in the ticket, and consider a compromise check when <code>forensicTriage</code> is Yes; our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04 explainer</a> covers what that flag asks agencies to do.</li>
<li><strong>EPSS and KEV disagreed</strong> on FortiMail, as expected: one estimates future observed activity, the other records evidence already seen.</li>
</ul>
<h2 id="what-the-ticket-should-hold">What the ticket should hold</h2>
<p>Store the inputs, not just the priority, so a reviewer can see why. A suggested field list (ours, not a standard):</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Why</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Asset, product identifier (purl or CPE), version</td>
          <td>Ties VEX and affected checks to the right thing</td>
      </tr>
      <tr>
          <td>CVE ID, record state, record <code>dateUpdated</code></td>
          <td>Detects rejected IDs and record changes</td>
      </tr>
      <tr>
          <td>Affected status and how it was decided</td>
          <td>&ldquo;unknown&rdquo; needs a different owner from &ldquo;affected&rdquo;</td>
      </tr>
      <tr>
          <td>VEX status, justification, document author and timestamp</td>
          <td>Audit trail for closures</td>
      </tr>
      <tr>
          <td>KEV <code>dateAdded</code>, <code>dueDate</code>, <code>knownRansomwareCampaignUse</code>, <code>forensicTriage</code></td>
          <td>Exploitation evidence and any federal deadline</td>
      </tr>
      <tr>
          <td>CISA SSVC Exploitation, Automatable, Technical Impact, timestamp</td>
          <td>Shows how fresh the assessment is</td>
      </tr>
      <tr>
          <td>EPSS score, percentile, model version, score date</td>
          <td>Scores shift when the model changes</td>
      </tr>
      <tr>
          <td>Your System Exposure and Human Impact</td>
          <td>The inputs only you can supply</td>
      </tr>
      <tr>
          <td>Priority and data pull date</td>
          <td>Lets you re-run and compare</td>
      </tr>
  </tbody>
</table>
<h2 id="keeping-it-current">Keeping it current</h2>
<p>The feeds move at different speeds: CISA updates KEV on weekdays, EPSS publishes daily, and CVE records and VEX documents change when their authors revise them. Re-run the flow on a schedule and on any change to a ticket&rsquo;s inputs, and alert when a finding&rsquo;s priority rises. Set your service-level targets per priority (for example, how many days &ldquo;Out-of-cycle&rdquo; means for you) in policy, not in the script; the deployer model does not define days.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>The script is a working example, not a product. It reads CVE records from a local folder (it does not fetch them), does not evaluate affected versions itself (pair it with a version check such as the evaluator in our affected-versions guide), handles OpenVEX JSON only, and uses one policy choice (KEV overrides an older CISA Exploitation value) that you may want to change. The four-value deployer inputs simplify real decisions: SSVC expects organizations to adapt trees to their context. We did not test it against commercial scanner exports.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE record states, <code>affected</code> algorithm, <code>defaultStatus</code> defaults to unknown</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE JSON schema (CVE_Record_Format.json, 5.2.0)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-23915 rejected, reason &ldquo;Not used&rdquo;; SSVC values and timestamps for the test CVEs</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">cvelistV5</a>, commit 451e97a</td>
          <td>Verified (read November 2, 2026)</td>
      </tr>
      <tr>
          <td>Vulnrichment assesses &ldquo;new and recent CVEs&rdquo;, adds SSVC decision points, never overwrites CNA data</td>
          <td><a href="https://github.com/cisagov/vulnrichment">cisagov/vulnrichment README</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV catalog 2026.10.30, 1,766 entries; dates and flags for the test CVEs</td>
          <td><a href="https://github.com/cisagov/kev-data">cisagov/kev-data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>EPSS field definitions; file model v2026.06.15, 398,784 CVEs; scores for the test CVEs</td>
          <td><a href="https://github.com/empiricalsec/epss_scores">empiricalsec/epss_scores</a></td>
          <td>Verified (computed from the file)</td>
      </tr>
      <tr>
          <td>Deployer inputs, values, 72-row table, outcome definitions, active never Defer</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/docs/howto/deployer_tree.md">CERT/CC SSVC deployer tree</a> and <a href="https://github.com/CERTCC/SSVC/blob/main/data/csv/ssvc/deployer_patch_application_priority_1_0_0.csv">decision table CSV</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Exploitation is &ldquo;the present state of exploitation&rdquo;; PoC vs Public PoC definitions</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/docs/reference/decision_points/exploitation.md">SSVC Exploitation decision point</a> and its source definitions</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>OpenVEX statuses and the justification requirement</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Script output</td>
          <td>Script above, Python 3.13, run November 2, 2026</td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Ivanti CVE-2026-111357 and CVE-2026-118527: What to Do</title><link>https://firewallsync.com/posts/ivanti-connect-secure-cve-2026-111357-cve-2026-118527-kev/</link><guid>https://firewallsync.com/posts/ivanti-connect-secure-cve-2026-111357-cve-2026-118527-kev/</guid><pubDate>Sat, 31 Oct 2026 05:00:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CISA added two more Ivanti Connect Secure flaws to KEV on 30 October: CVE-2026-111357, a privilege escalation to root, and CVE-2026-118527, a predictable-identifier bug that lets an authenticated attacker write files to a known path. Both chain with CVE-2026-104935, the authentication bypass Ivanti patched on 13 October. Here is what the public records confirm, what they do not, and what to do before the 2 November KEV deadline.</description><enclosure url="https://firewallsync.com/posts/ivanti-connect-secure-cve-2026-111357-cve-2026-118527-kev/cover_hu10585840409955398963.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/ivanti-connect-secure-cve-2026-111357-cve-2026-118527-kev/cover_hu10585840409955398963.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/ivanti-connect-secure-cve-2026-111357-cve-2026-118527-kev/cover_hu10585840409955398963.jpg" alt="Close-up of three thick, rusty steel chain links, the nearest in sharp focus and the rest fading into a blurred grey background" width="1200" height="630"></p><p>If you run Ivanti Connect Secure, the job this weekend is the same as two weeks ago, only with less room for doubt. On 30 October 2026 CISA added two more Connect Secure flaws to its <a href="https://github.com/cisagov/kev-data">Known Exploited Vulnerabilities catalog</a>: <strong>CVE-2026-111357</strong>, a privilege escalation to root, and <strong>CVE-2026-118527</strong>, a predictable-identifier bug that lets an authenticated attacker write files to a known location on disk. KEV says both chain with <strong>CVE-2026-104935</strong>, the authentication bypass Ivanti patched on 13 October.</p>
<p>Upgrade to a release that fixes CVE-2026-104935 if you have not already. Then treat any appliance that was exposed on 22.8R2.4 or earlier during October as a possible compromise. As of 31 October we could not find a public fixed version for either of the two new CVEs, and the sections below explain why.</p>
<h2 id="the-three-cves-in-the-chain">The three CVEs in the chain</h2>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-104935</th>
          <th>CVE-2026-118527</th>
          <th>CVE-2026-111357</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>What KEV calls it</td>
          <td>Authentication bypass by spoofing</td>
          <td>Generation of predictable numbers or identifiers</td>
          <td>Missing authentication for critical function</td>
      </tr>
      <tr>
          <td>Attacker position (KEV text)</td>
          <td>Remote, unauthenticated</td>
          <td>Remote, authenticated</td>
          <td>Local, authenticated</td>
      </tr>
      <tr>
          <td>Effect (KEV / CVE text)</td>
          <td>Execute arbitrary commands</td>
          <td>Upload files with attacker-controlled content to a known location on disk</td>
          <td>Privilege escalation; the CVE record says &ldquo;to root&rdquo;</td>
      </tr>
      <tr>
          <td>CWE</td>
          <td>CWE-290</td>
          <td>CWE-340</td>
          <td>KEV: CWE-306, CWE-862, CWE-345. CVE record: CWE-829</td>
      </tr>
      <tr>
          <td>Added to KEV</td>
          <td>2026-10-13</td>
          <td>2026-10-30</td>
          <td>2026-10-30</td>
      </tr>
      <tr>
          <td>KEV due date</td>
          <td>2026-10-16</td>
          <td>2026-11-13</td>
          <td>2026-11-02</td>
      </tr>
      <tr>
          <td>KEV forensic triage</td>
          <td>Yes</td>
          <td>No</td>
          <td>Yes</td>
      </tr>
      <tr>
          <td>CVE record</td>
          <td>Published by Ivanti, CVSS 3.1 8.6</td>
          <td>No published record found on 31 Oct</td>
          <td>Published by MITRE, CVSS 3.1 7.0</td>
      </tr>
  </tbody>
</table>
<p>The KEV entries for CVE-2026-118527 and CVE-2026-111357 both say the flaw &ldquo;can be chained with&rdquo; the other two. Read together, the descriptions fit a familiar pattern: an unauthenticated foothold, a way to place attacker-controlled content where the appliance will find it, then a jump to root. That ordering is our reading of the descriptions. Neither Ivanti nor CISA has published the step-by-step mechanics.</p>
<p>The MITRE record for CVE-2026-111357 adds one technical detail. It says Connect Secure 22.8R2.4 &ldquo;mishandles the manner in which the dsinvoked broker trusts DSCache configuration data, leading to privilege escalation to root, as exploited in the wild in October 2026.&rdquo;</p>
<p>If you track these programmatically, this pulls the three entries from CISA&rsquo;s KEV JSON. We ran it with Python 3.13 on 31 October; it uses only the standard library:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>URL <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://raw.githubusercontent.com/cisagov/kev-data/main/known_exploited_vulnerabilities.json&#34;</span>
</span></span><span style="display:flex;"><span>WANT <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;CVE-2026-104935&#34;</span>, <span style="color:#e6db74">&#34;CVE-2026-118527&#34;</span>, <span style="color:#e6db74">&#34;CVE-2026-111357&#34;</span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(URL, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">30</span>) <span style="color:#66d9ef">as</span> r:
</span></span><span style="display:flex;"><span>    kev <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(r)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> kev[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> v[<span style="color:#e6db74">&#34;cveID&#34;</span>] <span style="color:#f92672">in</span> WANT:
</span></span><span style="display:flex;"><span>        print(v[<span style="color:#e6db74">&#34;cveID&#34;</span>], v[<span style="color:#e6db74">&#34;dateAdded&#34;</span>], v[<span style="color:#e6db74">&#34;dueDate&#34;</span>],
</span></span><span style="display:flex;"><span>              <span style="color:#e6db74">&#34;triage=&#34;</span> <span style="color:#f92672">+</span> v<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;forensicTriage&#34;</span>, <span style="color:#e6db74">&#34;?&#34;</span>), <span style="color:#e6db74">&#34;,&#34;</span><span style="color:#f92672">.</span>join(v[<span style="color:#e6db74">&#34;cwes&#34;</span>]))
</span></span></code></pre></div><p>Output on 31 October:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>CVE-2026-118527 2026-10-30 2026-11-13 triage=No CWE-340
</span></span><span style="display:flex;"><span>CVE-2026-111357 2026-10-30 2026-11-02 triage=Yes CWE-306,CWE-862,CWE-345
</span></span><span style="display:flex;"><span>CVE-2026-104935 2026-10-13 2026-10-16 triage=Yes CWE-290
</span></span></code></pre></div><p>For what each field means and how to match the feed against a scanner export, see our guide to the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">CISA KEV JSON feed</a>.</p>
<h2 id="what-is-fixed-and-what-is-not">What is fixed and what is not</h2>
<p><strong>CVE-2026-104935 has fixes.</strong> The CVE record Ivanti published says the bug affects &ldquo;ICS before version 25.1.3.0&rdquo; and &ldquo;does not impact ICS version 22.7R2.14 or prior&rdquo;. It lists 25.1.3.0, 25.1.3.1, 22.8R2.5 and 22.7R2.14 as unaffected.</p>
<p>Rapid7&rsquo;s write-up of Ivanti&rsquo;s advisory breaks this down by branch: 25.1.1.1 and earlier fixed in 25.1.1.2, 25.1.2.2 and earlier fixed in 25.1.3.0 and 25.1.3.1, and 22.8R2.4 and earlier fixed in 22.8R2.5. The 25.1.1.2 entry does not appear in the CVE record&rsquo;s unaffected list, so confirm it against <a href="https://hub.ivanti.com/s/article/Security-Advisory-Ivanti-Connect-Secure-CVE-2026-104935">Ivanti&rsquo;s advisory</a> before you rely on it.</p>
<p><strong>CVE-2026-111357 and CVE-2026-118527 have no fixed version that we could confirm.</strong> When Ivanti disclosed CVE-2026-104935 on 13 October, it fixed two stages of the chain. <a href="https://www.theregister.com/a/5303095">The Register</a> and <a href="https://thehackernews.com/2026/10/attacker-chains-three-ivanti-connect.html">The Hacker News</a> both reported that a third flaw, the privilege escalation, had no fix yet, and quoted Volexity saying Ivanti planned to fix it &ldquo;in the coming weeks&rdquo;.</p>
<p>Since then:</p>
<ul>
<li>The only CVE record for CVE-2026-111357 is MITRE&rsquo;s. It lists 22.8R2.4 as affected and gives no fixed version. Its references are Volexity&rsquo;s report and Ivanti&rsquo;s CVE-2026-104935 advisory, not a separate Ivanti advisory.</li>
<li>CVE-2026-118527 had no record in the CVE List repository when we checked on 31 October. Its KEV entry is the only primary description we found.</li>
<li>Both KEV entries list Ivanti&rsquo;s CVE-2026-104935 advisory as the vendor reference, and both carry CISA&rsquo;s standard required action: apply vendor mitigations, or &ldquo;discontinue use of the product if mitigations are unavailable&rdquo;.</li>
<li>Canada&rsquo;s Cyber Centre updated its advisory on 16 October to say open-source reporting indicates CVE-2026-111357 is exploited. Singapore&rsquo;s CSA alert, last updated 30 October, still names only CVE-2026-104935 and advises updating &ldquo;to the latest versions&rdquo;.</li>
</ul>
<p>We could not read Ivanti&rsquo;s advisory directly during this run, and Ivanti may have published a fix that these records do not yet reflect. Check the advisory and Ivanti&rsquo;s support portal before you decide that your current version is enough.</p>
<h2 id="why-228r24-matters">Why 22.8R2.4 matters</h2>
<p>According to The Hacker News&rsquo;s summary of Volexity&rsquo;s report, Volexity found the chain on a customer appliance running 22.8R2.4, which was fully patched at the time. That is the exact version MITRE&rsquo;s CVE-2026-111357 record lists as affected. Exploitation began on 3 October, ten days before disclosure.</p>
<p>Volexity tracks the actor as UTA0576. The same reports say the attacker:</p>
<ul>
<li>stole LDAP domain credentials from the appliance, searched Active Directory for passwords and ran password-spraying attacks;</li>
<li>used no malware or web shells, only commands already on the appliance (per The Register);</li>
<li>was assessed by Volexity, with low confidence, as criminally rather than espionage motivated.</li>
</ul>
<p>Two vendor statements limit how much you can learn from the appliance itself. Rapid7 quotes Ivanti saying &ldquo;reliable indicators of compromise are not available&rdquo;. The Hacker News reports Volexity saying the Ivanti Integrity Checker Tool does not detect this chain, even though Ivanti still recommends running it regularly.</p>
<h2 id="triage-checklist">Triage checklist</h2>
<ol>
<li><strong>Inventory.</strong> List every Connect Secure appliance and its exact version, including standby units and lab boxes that face the internet.</li>
<li><strong>Close the entry point.</strong> Upgrade anything below the fixed versions for CVE-2026-104935 (see above). This is the unauthenticated step. KEV describes the other two flaws as needing an authenticated or local attacker.</li>
<li><strong>Assume compromise on exposed 22.8R2.4-and-earlier appliances.</strong> If an appliance ran a vulnerable version on the internet during October, patching does not undo what an attacker already did. Neither Ivanti nor Volexity has said that upgrading removes existing access.</li>
<li><strong>Rotate what the appliance knows.</strong> Volexity saw LDAP credentials stolen. Rotate the directory bind account and any other service credentials stored on the appliance, and review VPN user sessions and tokens. This is our recommendation, based on the reported attacker behavior. Our explainer on <a href="/posts/what-is-token-theft-and-why-mfa-doesnt-always-stop-it/">why MFA does not always stop token theft</a> covers the session side.</li>
<li><strong>Hunt outside the appliance.</strong> With no reliable on-box IOCs, the useful signals are in your other logs:
<ul>
<li>password spraying or bursts of failed authentications from the appliance&rsquo;s internal IP against Active Directory;</li>
<li>LDAP queries from the appliance that go beyond its normal user lookups;</li>
<li>in web or WAF logs in front of the appliance, large numbers of requests to <code>/dana-na/auth/url_default/welcome.cgi</code>, and requests whose <code>X-DSCGI-PATH</code> header does not match the URL path (both indicators Volexity published, as reported by The Hacker News);</li>
<li>traffic to or from <code>108.165.147[.]64</code>, the IP Volexity listed.</li>
</ul>
</li>
<li><strong>Escalate to forensics if you find anything.</strong> Volexity says confirming compromise needs network and memory forensics. Our guide to <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">CISA BOD 26-04</a> explains what the KEV &ldquo;forensic triage&rdquo; flag asks of federal agencies and what private teams can borrow from it.</li>
<li><strong>Decide on exposure while you wait.</strong> Until a fix for CVE-2026-111357 is confirmed, limit who can reach the appliance where you can. KEV&rsquo;s own fallback is to stop using the product if mitigations are unavailable. Our post on the <a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">FortiMail zero-day with no patch</a> walks through the same decision for another edge device.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>We did not test any of these flaws, and we have not seen exploit code.</li>
<li>We could not fetch Ivanti&rsquo;s advisory or Volexity&rsquo;s report directly. Version and fix details come from the CVE records, CISA KEV data and Rapid7&rsquo;s reproduction of the advisory. Attacker behavior and indicators come from press summaries of Volexity&rsquo;s findings, which we attribute throughout.</li>
<li>CVE-2026-118527 had no CVE record when we checked, so its description rests on one source, CISA&rsquo;s KEV entry.</li>
<li>Fix status can change quickly. Treat &ldquo;no confirmed fix&rdquo; as true for 31 October 2026 only.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV additions, dates, due dates, CWEs, forensic triage, &ldquo;can be chained with&rdquo;, required action</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities.json">CISA KEV catalog (kev-data)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-104935 description, unaffected versions, CVSS 8.6, CISA SSVC &ldquo;active&rdquo;</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/104xxx/CVE-2026-104935.json">CVE-2026-104935 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-111357 description (dsinvoked, DSCache, root), affected 22.8R2.4, CVSS 7.0, MITRE as CNA</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/111xxx/CVE-2026-111357.json">CVE-2026-111357 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Per-branch fixed versions including 25.1.1.2; Ivanti: &ldquo;reliable indicators of compromise are not available&rdquo;</td>
          <td><a href="https://www.rapid7.com/blog/post/etr-ivanti-connect-secure-zero-day-exploited-in-the-wild-cve-2026-104935/">Rapid7</a></td>
          <td>Qualified (secondary)</td>
      </tr>
      <tr>
          <td>Third flaw unfixed at disclosure; Volexity on 22.8R2.4, UTA0576, LDAP theft, indicators, Integrity Checker gap</td>
          <td><a href="https://thehackernews.com/2026/10/attacker-chains-three-ivanti-connect.html">The Hacker News</a>, <a href="https://www.theregister.com/a/5303095">The Register</a></td>
          <td>Qualified (press, citing Volexity)</td>
      </tr>
      <tr>
          <td>CVE-2026-111357 reported exploited (update of 16 Oct)</td>
          <td><a href="https://cyber.gc.ca/en/alerts-advisories/ivanti-security-advisory-av26-1035">Canadian Centre for Cyber Security AV26-1035</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Singapore alert names only CVE-2026-104935 (updated 30 Oct)</td>
          <td><a href="https://www.csa.gov.sg/alerts-and-advisories/alerts/al-2026-147/">CSA Singapore AL-2026-147</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>No public fix for CVE-2026-111357 or CVE-2026-118527</td>
          <td>Absence across the sources above</td>
          <td>Qualified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>GitLab CVE-2026-97229: Patch the Regex RCE, Then Hunt</title><link>https://firewallsync.com/posts/gitlab-cve-2026-97229-ci-regex-rce-patch-and-hunt/</link><guid>https://firewallsync.com/posts/gitlab-cve-2026-97229-ci-regex-rce-patch-and-hunt/</guid><pubDate>Fri, 30 Oct 2026 04:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CISA added GitLab CVE-2026-97229, an unauthenticated remote code execution bug in how GitLab compiles regular expressions from CI/CD configuration, to KEV on October 29 with a November 1 deadline. The 19.4.2 fix is not the end of it: a sibling bug, CVE-2026-110264, needs 19.4.3 or 19.5.1. Here are the versions that close all four related CVEs, a script to check yours, GitLab's detections, and what Ruby shops outside GitLab should check.</description><enclosure url="https://firewallsync.com/posts/gitlab-cve-2026-97229-ci-regex-rce-patch-and-hunt/cover_hu18432601525673025422.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/gitlab-cve-2026-97229-ci-regex-rce-patch-and-hunt/cover_hu18432601525673025422.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/gitlab-cve-2026-97229-ci-regex-rce-patch-and-hunt/cover_hu18432601525673025422.jpg" alt="Black-and-white photo of rows of large valve wheels on pipes along an underground corridor, with ducting overhead and a metal grate walkway" width="1200" height="630"></p><p>CVE-2026-97229 is a remote code execution bug in self-managed GitLab Community and Enterprise Edition. GitLab&rsquo;s <a href="https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-4-2-released/">19.4.2 patch release</a> says it &ldquo;could have allowed an unauthenticated user to execute arbitrary code on the GitLab server via a specially crafted regular expression in a CI/CD configuration, due to insufficient regular expression validation.&rdquo; GitLab scores it CVSS 3.1 10.0. CISA added it to the <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-97229">Known Exploited Vulnerabilities catalog</a> on October 29, 2026, with a due date of November 1.</p>
<p>If you run self-managed GitLab, do two things today: upgrade to the October 21 patch release for your branch (not just the October 12 one), and check for signs that someone got there first.</p>
<h2 id="what-is-affected-and-what-fixes-it">What is affected and what fixes it</h2>
<p>CVE-2026-97229 is one of four GitLab CVEs that trace back to bugs in Onigmo, the regular expression engine inside Ruby. GitLab fixed them in three patch releases between September and October. Ruby&rsquo;s own <a href="https://www.ruby-lang.org/en/news/2026/10/23/regexp-fixes-for-gitlab-cves/">advisory of October 23</a> ties them together.</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>GitLab&rsquo;s description</th>
          <th>Auth needed</th>
          <th>CVSS 3.1</th>
          <th>Affected (from the CVE record)</th>
          <th>First fixed</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-89078</td>
          <td>Double free parsing a regex in CI/CD configuration</td>
          <td>Authenticated</td>
          <td>9.9</td>
          <td>19.2 before 19.2.7, 19.3 before 19.3.3, 19.4 before 19.4.1</td>
          <td>19.4.1, 19.3.3, 19.2.7</td>
      </tr>
      <tr>
          <td>CVE-2026-93577</td>
          <td>Integer overflow compiling a regex in CI/CD configuration</td>
          <td>Authenticated</td>
          <td>9.9</td>
          <td>Same as above</td>
          <td>19.4.1, 19.3.3, 19.2.7</td>
      </tr>
      <tr>
          <td><strong>CVE-2026-97229</strong></td>
          <td>Integer overflow, CI/CD regex compiler</td>
          <td><strong>Unauthenticated</strong></td>
          <td><strong>10.0</strong></td>
          <td>19.2 before 19.2.8, 19.3 before 19.3.5, 19.4 before 19.4.2</td>
          <td>19.4.2, 19.3.5, 19.2.8 (Oct 12)</td>
      </tr>
      <tr>
          <td>CVE-2026-110264</td>
          <td>Integer overflow compiling a regex</td>
          <td>Unauthenticated</td>
          <td>10.0</td>
          <td>7.0 before 18.11.13, 19.0 before 19.0.10, 19.1 before 19.1.9, 19.2 before 19.2.9, 19.3 before 19.3.6, 19.4 before 19.4.3, 19.5 before 19.5.1</td>
          <td>19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10, 18.11.13 (Oct 21)</td>
      </tr>
  </tbody>
</table>
<p>Only CVE-2026-97229 is in KEV today. CISA&rsquo;s SSVC entries in the CVE records list Exploitation as &ldquo;active&rdquo; for CVE-2026-97229 and &ldquo;none&rdquo; for the other three, as of the dates CISA recorded them. CVE-2026-110264 is also unauthenticated and rated 10.0, and its range reaches back to GitLab 7.0, so treat it with the same urgency.</p>
<p><strong>Upgrade target:</strong> 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10 or 18.11.13, or anything later on the same branch. If you are on a branch older than 18.11, there is no backport; move to a supported branch. GitLab&rsquo;s <a href="https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-5-1-released/">19.5.1 release post</a> lists no workaround, only &ldquo;upgrade to the latest version as soon as possible.&rdquo;</p>
<p>GitLab.com and GitLab Dedicated are already patched, according to both release posts. This is a self-managed problem.</p>
<h3 id="check-a-version-quickly">Check a version quickly</h3>
<p>The <a href="https://docs.gitlab.com/ee/api/version">Version API</a> returns the running version (<code>GET /api/v4/version</code> with a token). The script below compares a version string against the affected ranges in the four CVE records. We ran it with Python 3.13; the output shown is real.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Check a GitLab version against the four Onigmo regex CVEs.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Fixed versions come from each CVE record&#39;s &#39;affected&#39; ranges (read 2026-10-30).&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> re<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>FIXED <span style="color:#f92672">=</span> {  <span style="color:#75715e"># CVE -&gt; {(major, minor): first fixed patch}</span>
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;CVE-2026-89078&#34;</span>:  {(<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">2</span>): <span style="color:#ae81ff">7</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">3</span>): <span style="color:#ae81ff">3</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">4</span>): <span style="color:#ae81ff">1</span>},
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;CVE-2026-93577&#34;</span>:  {(<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">2</span>): <span style="color:#ae81ff">7</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">3</span>): <span style="color:#ae81ff">3</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">4</span>): <span style="color:#ae81ff">1</span>},
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;CVE-2026-97229&#34;</span>:  {(<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">2</span>): <span style="color:#ae81ff">8</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">3</span>): <span style="color:#ae81ff">5</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">4</span>): <span style="color:#ae81ff">2</span>},
</span></span><span style="display:flex;"><span>    <span style="color:#e6db74">&#34;CVE-2026-110264&#34;</span>: {(<span style="color:#ae81ff">18</span>, <span style="color:#ae81ff">11</span>): <span style="color:#ae81ff">13</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">0</span>): <span style="color:#ae81ff">10</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">1</span>): <span style="color:#ae81ff">9</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">2</span>): <span style="color:#ae81ff">9</span>,
</span></span><span style="display:flex;"><span>                        (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">3</span>): <span style="color:#ae81ff">6</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">4</span>): <span style="color:#ae81ff">3</span>, (<span style="color:#ae81ff">19</span>, <span style="color:#ae81ff">5</span>): <span style="color:#ae81ff">1</span>},
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>OLD_RANGE_START <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;CVE-2026-110264&#34;</span>: (<span style="color:#ae81ff">7</span>, <span style="color:#ae81ff">0</span>)}  <span style="color:#75715e"># affected from 7.0, no fix before 18.11</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">check</span>(ver):
</span></span><span style="display:flex;"><span>    major, minor, patch <span style="color:#f92672">=</span> map(int, re<span style="color:#f92672">.</span><span style="color:#66d9ef">match</span>(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;(\d+)\.(\d+)\.(\d+)&#34;</span>, ver)<span style="color:#f92672">.</span>groups())
</span></span><span style="display:flex;"><span>    branch, out <span style="color:#f92672">=</span> (major, minor), {}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> cve, fixes <span style="color:#f92672">in</span> FIXED<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> branch <span style="color:#f92672">in</span> fixes:
</span></span><span style="display:flex;"><span>            out[cve] <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;FIXED&#34;</span> <span style="color:#66d9ef">if</span> patch <span style="color:#f92672">&gt;=</span> fixes[branch] <span style="color:#66d9ef">else</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;VULNERABLE (fixed in </span><span style="color:#e6db74">{</span>major<span style="color:#e6db74">}</span><span style="color:#e6db74">.</span><span style="color:#e6db74">{</span>minor<span style="color:#e6db74">}</span><span style="color:#e6db74">.</span><span style="color:#e6db74">{</span>fixes[branch]<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">elif</span> branch <span style="color:#f92672">&gt;</span> max(fixes):
</span></span><span style="display:flex;"><span>            out[cve] <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;FIXED (newer branch)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">elif</span> cve <span style="color:#f92672">in</span> OLD_RANGE_START <span style="color:#f92672">and</span> branch <span style="color:#f92672">&gt;=</span> OLD_RANGE_START[cve]:
</span></span><span style="display:flex;"><span>            out[cve] <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;VULNERABLE (no fix on this branch)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>            out[cve] <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;not in affected range&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]:
</span></span><span style="display:flex;"><span>    print(v)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> cve, status <span style="color:#f92672">in</span> check(v)<span style="color:#f92672">.</span>items():
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;  </span><span style="color:#e6db74">{</span>cve<span style="color:#e6db74">:</span><span style="color:#e6db74">16</span><span style="color:#e6db74">}</span><span style="color:#e6db74"> </span><span style="color:#e6db74">{</span>status<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 check_gitlab_regex_cves.py 19.4.2-ee 18.9.4
</span></span><span style="display:flex;"><span>19.4.2-ee
</span></span><span style="display:flex;"><span>  CVE-2026-89078   FIXED
</span></span><span style="display:flex;"><span>  CVE-2026-93577   FIXED
</span></span><span style="display:flex;"><span>  CVE-2026-97229   FIXED
</span></span><span style="display:flex;"><span>  CVE-2026-110264  VULNERABLE (fixed in 19.4.3)
</span></span><span style="display:flex;"><span>18.9.4
</span></span><span style="display:flex;"><span>  CVE-2026-89078   not in affected range
</span></span><span style="display:flex;"><span>  CVE-2026-93577   not in affected range
</span></span><span style="display:flex;"><span>  CVE-2026-97229   not in affected range
</span></span><span style="display:flex;"><span>  CVE-2026-110264  VULNERABLE (no fix on this branch)
</span></span></code></pre></div><p>The first result is the trap: an instance patched promptly on October 12 for the CVE that is now in KEV is still exposed to the sibling bug.</p>
<h2 id="how-the-bug-works-as-far-as-the-sources-say">How the bug works, as far as the sources say</h2>
<p>GitLab lets users supply regular expressions in places such as CI/CD configuration and compiles them with Ruby. Ruby&rsquo;s advisory explains: &ldquo;GitLab compiles regular expressions supplied by its users, such as patterns in CI configuration, with <code>Regexp.new</code>. GitLab found that specially crafted patterns could corrupt memory while Ruby compiles them, and that this could be turned into remote code execution on GitLab servers.&rdquo;</p>
<p>The Ruby change behind CVE-2026-97229 (and CVE-2026-93577) is described as an &ldquo;Integer overflow in the size of a compiled pattern&rdquo;, fixed in <a href="https://github.com/ruby/ruby/pull/19020">ruby/ruby#19020</a>. GitLab patched the Ruby it ships rather than waiting for a Ruby release.</p>
<p>GitLab says the bug is reachable by an unauthenticated user &ldquo;under certain conditions&rdquo; and does not spell those conditions out. The names of GitLab&rsquo;s own detections (below) refer to the CI Lint feature and its GraphQL endpoint, which suggests where to look first, but that is our reading of the detection names, not a statement from GitLab. No public write-up of the exploitation we could find gives more detail; treat any blog that claims a precise attack path without citing GitLab or the reporter with caution.</p>
<h2 id="hunt-gitlabs-detections">Hunt: GitLab&rsquo;s detections</h2>
<p>The 19.4.2 release post lists six detections for self-managed instances, each linked to a write-up in GitLab&rsquo;s <a href="https://gitlab.com/gitlab-security-oss/tldr"><code>gitlab-security-oss/tldr</code> repository</a> under <code>Detections/platforms/GitLab/Vulnerability Detections/</code>:</p>
<table>
  <thead>
      <tr>
          <th>Detection name</th>
          <th>File</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Onigmo native worker crash</td>
          <td><code>onigmo_native_worker_crash.md</code></td>
      </tr>
      <tr>
          <td>Graphql cilint hostile regexp</td>
          <td><code>graphql_cilint_hostile_regexp.md</code></td>
      </tr>
      <tr>
          <td>ci regexp oversized non ascii path</td>
          <td><code>ci_regexp_oversized_non_ascii_path.md</code></td>
      </tr>
      <tr>
          <td>ci regexp guard rejection</td>
          <td><code>ci_regexp_guard_rejection.md</code></td>
      </tr>
      <tr>
          <td>ci lint workhorse worker crash</td>
          <td><code>ci_lint_workhorse_worker_crash.md</code></td>
      </tr>
      <tr>
          <td>ci lint scripted client error burst</td>
          <td><code>ci_lint_scripted_client_error_burst.md</code></td>
      </tr>
  </tbody>
</table>
<p>The 19.5.1 post adds two for CVE-2026-110264: &ldquo;Onigmo native worker crash&rdquo; and &ldquo;Org mode render error burst&rdquo;.</p>
<p>We have not reproduced these detections; use GitLab&rsquo;s files for the exact log sources and queries. Practical order of work:</p>
<ol>
<li><strong>Look back before the patch date.</strong> Run the detections over logs as far back as your retention allows, up to the moment you upgraded. Worker crashes in the regex engine are not normal background noise.</li>
<li><strong>Assume the GitLab host is a credential store.</strong> A GitLab server holds CI/CD variables, runner registration tokens, deploy keys, and access tokens. If you find hits, or you cannot rule exploitation out, rotate those secrets and review recent pipeline and runner activity. Our guide to <a href="/posts/how-to-detect-and-rotate-leaked-api-keys-before-attackers-use-them/">detecting and rotating leaked API keys</a> covers the rotation order.</li>
<li><strong>Check for persistence.</strong> Look for new admin users, personal access tokens you did not create, new SSH keys on admin accounts, and unexpected files or processes on the host.</li>
</ol>
<p>KEV marks this entry <code>forensicTriage: Yes</code>. Under CISA&rsquo;s BOD 26-04 that means federal agencies must collect forensic data, not just patch; our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04 explainer</a> covers what that involves and what private teams can borrow.</p>
<h2 id="if-you-run-ruby-but-not-gitlab">If you run Ruby but not GitLab</h2>
<p>Ruby released 4.0.9, 3.4.12 and 3.3.13 with fixes for all the underlying Onigmo bugs. Ruby&rsquo;s advisory is precise about who is affected:</p>
<blockquote>
<p>These bugs are reached only when an application compiles a regular expression from a string that an attacker controls, for example by passing it to <code>Regexp.new</code>, <code>Regexp.union</code> or <code>Regexp.compile</code>, or by interpolating it into a regexp literal such as <code>/#{input}/</code>. Applications that only use regular expressions written by their own developers are not affected.</p>
</blockquote>
<p>Two points in that advisory matter for anyone building search filters, routing rules or admin &ldquo;match by pattern&rdquo; features:</p>
<ul>
<li>&ldquo;<code>Regexp.timeout</code> does not protect against these bugs.&rdquo; Timeouts address ReDoS, not memory corruption during compilation.</li>
<li>&ldquo;Limiting the length of patterns does not reliably protect against them either, because some of them can be triggered by short patterns.&rdquo;</li>
</ul>
<p>Ruby has not requested its own CVE IDs, so scanners may not flag your Ruby runtime. A quick way to find candidates in a codebase (illustrative, not exhaustive):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>grep -rnE <span style="color:#e6db74">&#39;Regexp\.(new|union|compile)\(|/[^/]*#\{&#39;</span> --include<span style="color:#f92672">=</span><span style="color:#e6db74">&#39;*.rb&#39;</span> app lib
</span></span></code></pre></div><p>Review each hit for whether the pattern can come from a user. If it can, upgrade Ruby, and consider whether users need regexes at all: plain substring or glob matching is a much smaller attack surface.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>GitLab has not published the exact preconditions for unauthenticated exploitation, and we have not tested an exploit. The CI Lint connection above is inferred from detection names.</li>
<li>KEV listing tells you exploitation has happened, not how widely. We found no primary source for a count of exposed or compromised instances.</li>
<li>Version data comes from the CVE records as updated on 2026-10-30. If GitLab revises them, the records win over this page.</li>
</ul>
<p>For how to weigh a KEV entry against CVSS and EPSS in your own queue, see <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>, and for reading the SSVC values quoted above, <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">our Vulnrichment guide</a>.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV addition 2026-10-29, due 2026-11-01, CWE-190, forensic triage &ldquo;Yes&rdquo;, ransomware use &ldquo;Unknown&rdquo;</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities.json">CISA KEV catalog (kev-data)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Description, CVSS 10.0 vector, affected ranges, fix versions, reporter credit</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/97xxx/CVE-2026-97229.json">CVE-2026-97229 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC Exploitation &ldquo;active&rdquo; for CVE-2026-97229; &ldquo;none&rdquo; for the other three</td>
          <td>CISA ADP entries in each CVE record (<a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/97xxx/CVE-2026-97229.json">97229</a>, <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/110xxx/CVE-2026-110264.json">110264</a>)</td>
          <td>Verified (as of CISA&rsquo;s timestamps)</td>
      </tr>
      <tr>
          <td>Release date October 12, 2026; GitLab.com and Dedicated already patched; six detections</td>
          <td><a href="https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-4-2-released/">GitLab 19.4.2 patch release</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-110264 range from 7.0; fix versions; release October 21, 2026; two detections</td>
          <td><a href="https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-5-1-released/">GitLab 19.5.1 patch release</a> and <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/110xxx/CVE-2026-110264.json">CVE record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-89078 and CVE-2026-93577 authenticated, 9.9, ranges</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/89xxx/CVE-2026-89078.json">CVE-2026-89078</a>, <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/93xxx/CVE-2026-93577.json">CVE-2026-93577</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Onigmo root cause, Ruby fix versions, who is affected, Regexp.timeout does not help</td>
          <td><a href="https://www.ruby-lang.org/en/news/2026/10/23/regexp-fixes-for-gitlab-cves/">Ruby advisory, 2026-10-23</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Unauthenticated path involves CI Lint</td>
          <td>Inferred from GitLab detection names</td>
          <td>Qualified (our interpretation)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Magento CVE-2026-48356 (PolyShell): Patch and Hunt</title><link>https://firewallsync.com/posts/magento-cve-2026-48356-polyshell-patch-and-hunt/</link><guid>https://firewallsync.com/posts/magento-cve-2026-48356-polyshell-patch-and-hunt/</guid><pubDate>Thu, 29 Oct 2026 04:30:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>CVE-2026-48356 is the PolyShell file-upload flaw in Adobe Commerce and Magento Open Source. CISA added it to the KEV catalog on October 27, 2026. Here are the fixed builds, why a store already on 2.4.9 may still need an update, and the checks that find an existing compromise.</description><enclosure url="https://firewallsync.com/posts/magento-cve-2026-48356-polyshell-patch-and-hunt/cover_hu14701130590101558514.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/magento-cve-2026-48356-polyshell-patch-and-hunt/cover_hu14701130590101558514.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/magento-cve-2026-48356-polyshell-patch-and-hunt/cover_hu14701130590101558514.jpg" alt="Rows of empty grey metal shopping carts with orange handles, nested together against a white wall on a polished concrete floor" width="1200" height="630"></p><p>CVE-2026-48356 is the Magento file-upload flaw that Sansec calls <strong>PolyShell</strong>, and it is now on CISA&rsquo;s <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-48356">Known Exploited Vulnerabilities catalog</a> (added October 27, 2026; federal due date November 10). Adobe shipped the fix in <a href="https://helpx.adobe.com/security/products/magento/apsb26-73.html">APSB26-73</a> on July 14, 2026. If your store is not on a <code>2026-jul</code> build, update now. If it was exposed for any time since March, hunt as well. Attackers have used the flaw to plant webshells and to inject JavaScript loaders into CMS blocks, and an update removes neither.</p>
<h2 id="what-is-affected-and-what-fixes-it">What is affected and what fixes it</h2>
<p>The table comes from Adobe&rsquo;s <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/48xxx/CVE-2026-48356.json">CVE record</a> and the APSB26-73 solution table, which agree on the fixed builds.</p>
<table>
  <thead>
      <tr>
          <th>Product</th>
          <th>Affected (up to and including)</th>
          <th>Fixed build</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Adobe Commerce</td>
          <td>2.4.9, 2.4.8-p5, 2.4.7-p10, 2.4.6-p15, 2.4.5-p17, 2.4.4-p18</td>
          <td>2.4.9-2026-jul, 2.4.8-2026-jul, 2.4.7-2026-jul, 2.4.6-2026-jul, 2.4.5-2026-jul, 2.4.4-2026-jul</td>
      </tr>
      <tr>
          <td>Magento Open Source</td>
          <td>2.4.9, 2.4.8-p5, 2.4.7-p10, 2.4.6-p15</td>
          <td>2.4.9-2026-jul, 2.4.8-2026-jul, 2.4.7-2026-jul, 2.4.6-2026-jul</td>
      </tr>
      <tr>
          <td>Adobe Commerce B2B</td>
          <td>1.5.3, 1.5.2-p5, 1.4.2-p10, 1.3.4-p17, 1.3.3-p18</td>
          <td>1.5.3-2026-jul, 1.5.2-2026-jul, 1.4.2-2026-jul, 1.3.4-2026-jul, 1.3.3-2026-jul</td>
      </tr>
      <tr>
          <td>Adobe Commerce Events</td>
          <td>1.20.0 and earlier</td>
          <td>1.21.0</td>
      </tr>
  </tbody>
</table>
<p>Two details matter here.</p>
<p><strong>2.4.9 is listed as affected.</strong> Sansec&rsquo;s <a href="https://sansec.io/research/magento-polyshell">PolyShell write-up</a> says Adobe released the fix in 2.4.9 on May 12, 2026, and backported it to the supported release lines through APSB26-73 on July 14. Adobe&rsquo;s CVE record, however, lists 2.4.9 among the affected versions and 2.4.9-2026-jul as unaffected. We have not diffed the two releases, so we can&rsquo;t say which is right. The safe reading is Adobe&rsquo;s: move to 2.4.9-2026-jul.</p>
<p><strong>Older Open Source lines are not listed at all.</strong> Adobe lists 2.4.4 and 2.4.5 <code>2026-jul</code> builds only for Adobe Commerce. For Magento Open Source 2.4.5 and older, the record names neither an affected nor a fixed version. Don&rsquo;t read that as safe: Sansec says every version up to 2.4.9-alpha2 is affected. If you run one of those lines, plan an upgrade and treat the server-side mitigations below as essential in the meantime.</p>
<p>Check what you run with <code>bin/magento --version</code> or <code>composer show magento/product-community-edition</code> (<code>magento/product-enterprise-edition</code> for Commerce), and compare the result with the table.</p>
<h2 id="how-polyshell-works">How PolyShell works</h2>
<p>Sansec describes the flaw in the REST API&rsquo;s handling of file uploads for cart item custom options, where a product option has type &ldquo;file&rdquo;. The most exposed routes are the unauthenticated guest-cart ones:</p>
<ul>
<li><code>POST /V1/guest-carts/:cartId/items</code></li>
<li><code>PUT /V1/guest-carts/:cartId/items/:itemId</code></li>
</ul>
<p>Uploaded files are written to <code>pub/media/custom_options/quote/</code>. Sansec says the GraphQL mutations use a different code path and are not vulnerable.</p>
<p>What an upload leads to depends on the server. Sansec lists code execution through an uploaded PHP file for Apache before 2.3.5 without <code>php_flag engine 0</code>, for stock nginx configs on 2.0.0 to 2.2.x, and for any nginx config that passes every <code>.php</code> request to PHP-FPM. Stored XSS is possible on versions before 2.3.5 and on custom web server configurations.</p>
<p>Adobe describes the same CVE differently. The CVE record says exploitation &ldquo;requires user interaction in that a victim must visit a maliciously crafted URL or interact with a compromised web page&rdquo; and scores it CVSS 3.1 <strong>9.3</strong> (<code>AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:N</code>). The regional copy of the bulletin we read shows <strong>9.6</strong> with <code>A:L</code> and labels the impact &ldquo;Privilege escalation&rdquo;. We report both; neither changes the action. A KEV listing outranks the <code>UI:R</code> in the vector. Our <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV guide</a> explains why.</p>
<h2 id="what-attackers-have-done-with-it">What attackers have done with it</h2>
<p>Sansec reported a mass wave on March 30, 2026: <a href="https://sansec.io/research/polyshell-mass-attack-wave">471 stores compromised in one hour</a>, with a backdoor named <code>accesson.php</code> dropped across multiple directories and loaders injected into CMS pages and static blocks. That was months before Adobe&rsquo;s July backport.</p>
<p>On October 26, <a href="https://thehackernews.com/2026/10/hackers-infect-953-online-stores-with.html">The Hacker News reported</a> Sansec&rsquo;s findings on a longer campaign: 953 stores infected since July 2025 with a JavaScript loader stored in Magento&rsquo;s <code>cms_block</code> table. Per that report, Sansec attributes 471 of those infections to CVE-2026-48356, and about 102 more to file uploads through PolyShell or CVE-2025-54236 (SessionReaper). One store on 2.4.7-p10 with no missing core patches was compromised through a rogue administrator account. The report also describes webshells disguised as PNG files in <code>pub/media/custom_options/quote/</code>. The loader&rsquo;s final payload is not known.</p>
<p>These figures are Sansec&rsquo;s, as reported by The Hacker News. CISA&rsquo;s catalog entry does not say who is exploiting the flaw, and lists known ransomware use as &ldquo;Unknown&rdquo;.</p>
<h2 id="step-1-install-the-fixed-build">Step 1: Install the fixed build</h2>
<p>Update to the <code>2026-jul</code> build for your release line, then redeploy static content and clear caches as your normal release process requires. If you can&rsquo;t update this week, do steps 2 and 3 now anyway: they limit what a successful upload can do and tell you whether one has already happened.</p>
<p>Sansec suggests a WAF rule that blocks the vulnerable API calls. That buys time, but it is a mitigation, not a fix.</p>
<h2 id="step-2-make-sure-uploads-cant-be-served-or-executed">Step 2: Make sure uploads can&rsquo;t be served or executed</h2>
<p>Magento ships deny rules for this directory. Check that your live config still has them.</p>
<p><strong>nginx.</strong> The stock <a href="https://github.com/magento/magento2/blob/2.4-develop/nginx.conf.sample"><code>nginx.conf.sample</code></a> in the 2.4-develop branch denies the directory and only hands named entry points to PHP. These are excerpts from that file, not a complete config:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-nginx" data-lang="nginx"><span style="display:flex;"><span><span style="color:#66d9ef">location</span> <span style="color:#e6db74">/media/custom_options/</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">deny</span> <span style="color:#e6db74">all</span>;
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">location</span> ~ <span style="color:#e6db74">^/(index|get|static|errors/report|errors/404|errors/503|health_check)\.php$</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#75715e"># ... fastcgi_pass to PHP-FPM ...
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Banned locations (only reached if the earlier PHP entry point regexes don&#39;t match)
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span><span style="color:#66d9ef">location</span> ~<span style="color:#e6db74">*</span> <span style="color:#e6db74">(\.php</span>$|\.phtml$|\.htaccess$|\.htpasswd$|\.git<span style="color:#e6db74">)</span> {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">deny</span> <span style="color:#e6db74">all</span>;
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>The risky pattern is a hand-written or hosting-panel config with a generic <code>location ~ \.php$ { fastcgi_pass ...; }</code> and no deny for <code>/media/custom_options/</code>. With that, a PHP file uploaded to the directory is executed. Sansec recommends a <code>deny all</code> location for <code>custom_options</code> placed before any <code>\.php$</code> block.</p>
<p><strong>Apache.</strong> The repository ships <code>pub/media/custom_options/.htaccess</code> containing <code>Require all denied</code> (Apache 2.4) or <code>deny from all</code> (older). Confirm it exists on disk and that <code>AllowOverride</code> lets Apache read it. Sansec also recommends <code>php_flag engine 0</code> there.</p>
<p>Neither rule stops the upload. Both stop the uploaded file from being fetched or run through the web server.</p>
<h2 id="step-3-hunt-for-an-existing-compromise">Step 3: Hunt for an existing compromise</h2>
<p>Run these from the Magento root, or adjust the paths. Sansec published the first checks; we added the database queries.</p>
<p><strong>Files in the upload directory.</strong> Customer uploads can be legitimate, but any script file there is a strong sign of compromise:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span><span style="color:#75715e"># Everything uploaded through custom options, newest first</span>
</span></span><span style="display:flex;"><span>find pub/media/custom_options/ -type f ! -name <span style="color:#e6db74">&#39;.htaccess&#39;</span> -printf <span style="color:#e6db74">&#39;%TY-%Tm-%Td %TH:%TM  %s  %p\n&#39;</span> | sort -r | head -50
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Script extensions there are a red flag</span>
</span></span><span style="display:flex;"><span>find pub/media/custom_options/ -type f <span style="color:#ae81ff">\(</span> -iname <span style="color:#e6db74">&#39;*.php*&#39;</span> -o -iname <span style="color:#e6db74">&#39;*.phtml&#39;</span> -o -iname <span style="color:#e6db74">&#39;*.phar&#39;</span> <span style="color:#ae81ff">\)</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># The backdoor name from the March wave, anywhere in the docroot tree</span>
</span></span><span style="display:flex;"><span>find /var/www -name <span style="color:#e6db74">&#39;accesson.php&#39;</span> -type f
</span></span></code></pre></div><p>The October report describes webshells disguised as <code>.png</code>. A PNG that starts with <code>&lt;?php</code> is not an image. Check with <code>file</code> or <code>head -c 64</code> on anything recent in that directory.</p>
<p><strong>Injected scripts in CMS content.</strong> The loader lives in the database, so file scans miss it. Run this against a replica or read-only connection (add your table prefix if you use one):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-sql" data-lang="sql"><span style="display:flex;"><span><span style="color:#66d9ef">SELECT</span> <span style="color:#e6db74">&#39;block&#39;</span> <span style="color:#66d9ef">AS</span> kind, block_id <span style="color:#66d9ef">AS</span> id, identifier, update_time
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">FROM</span> cms_block
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">WHERE</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%&lt;script%&#39;</span>
</span></span><span style="display:flex;"><span>   <span style="color:#66d9ef">OR</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%deepveindrilling%&#39;</span>
</span></span><span style="display:flex;"><span>   <span style="color:#66d9ef">OR</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%lanhd6549tdhse%&#39;</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">UNION</span> <span style="color:#66d9ef">ALL</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">SELECT</span> <span style="color:#e6db74">&#39;page&#39;</span>, page_id, identifier, update_time
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">FROM</span> cms_page
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">WHERE</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%&lt;script%&#39;</span>
</span></span><span style="display:flex;"><span>   <span style="color:#66d9ef">OR</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%deepveindrilling%&#39;</span>
</span></span><span style="display:flex;"><span>   <span style="color:#66d9ef">OR</span> content <span style="color:#66d9ef">LIKE</span> <span style="color:#e6db74">&#39;%lanhd6549tdhse%&#39;</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">ORDER</span> <span style="color:#66d9ef">BY</span> update_time <span style="color:#66d9ef">DESC</span>;
</span></span></code></pre></div><p>Column names come from Magento&rsquo;s <code>Cms</code> module schema; we checked the query&rsquo;s syntax against a small SQLite copy of those tables, not against a live MySQL store. Many stores have legitimate scripts in CMS blocks, so review hits by <code>update_time</code> and by where the script loads from. The October report lists these loader domains: <code>manulfie[.]click</code>, <code>neswnsssn[.]click</code>, <code>lanhd6549tdhse[.]top</code>, <code>additionalver[.]com</code> and <code>deepveindrilling[.]com</code>. Sansec&rsquo;s March write-up adds <code>jslibrary[.]net</code> and <code>canevaslab[.]com</code>. Add any you want to the <code>LIKE</code> list.</p>
<p><strong>Admin accounts you didn&rsquo;t create.</strong> One store in the October report was compromised through a rogue admin:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-sql" data-lang="sql"><span style="display:flex;"><span><span style="color:#66d9ef">SELECT</span> user_id, username, email, created, logdate, is_active
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">FROM</span> admin_user
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">WHERE</span> created <span style="color:#f92672">&gt;=</span> <span style="color:#e6db74">&#39;2026-03-01&#39;</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">ORDER</span> <span style="color:#66d9ef">BY</span> created <span style="color:#66d9ef">DESC</span>;
</span></span></code></pre></div><p><strong>Web logs.</strong> Look for requests to the guest-cart item routes, which usually sit under <code>/rest/V1/</code> or <code>/rest/&lt;store_code&gt;/V1/</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>grep -E <span style="color:#e6db74">&#39;&#34;(POST|PUT) /rest/([a-z0-9_]+/)?V1/guest-carts/[^/ ]+/items&#39;</span> /var/log/nginx/access.log* | head
</span></span></code></pre></div><p>Headless storefronts call these routes legitimately, so a hit is not proof. Line up timestamps with new files in <code>pub/media/custom_options/quote/</code> and with <code>cms_block</code> updates.</p>
<p>If you find a webshell or an injected loader, treat the store as compromised. Remove the artifacts, rotate admin, database and integration credentials, check for other persistence (cron, other PHP files, modified <code>app/etc/env.php</code>), and consider your payment-card obligations, since these loaders run in shoppers&rsquo; browsers. Our <a href="/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/">PHP2Shell response guide</a> covers the server-side hunt for PHP webshells in more detail.</p>
<h2 id="hardening-after-the-patch">Hardening after the patch</h2>
<ul>
<li>Keep <code>pub/</code> as the document root and let PHP run only for Magento&rsquo;s named entry points. Sansec&rsquo;s RCE conditions are old releases or web server configs that execute PHP files in upload directories.</li>
<li>Turn off PHP execution in every writable media directory, not just this one. Our <a href="/posts/php-ini-security-hardening-production-settings/">php.ini hardening guide</a> covers related settings.</li>
<li>If you don&rsquo;t use file-type custom options, Sansec&rsquo;s WAF suggestion is cheap: block file uploads to the guest-cart item routes.</li>
<li>Monitor <code>cms_block</code> and <code>cms_page</code> for changes. A loader injected through the database survives code deploys.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We did not reproduce the exploit or diff the patched releases. Affected and fixed versions come from Adobe&rsquo;s CVE record and bulletin. The upload mechanics, server conditions and campaign figures come from Sansec, the latter through The Hacker News. The <code>adobe.com/trust</code> copy of APSB26-73 could not be retrieved during research; we used Adobe&rsquo;s regional <code>helpx</code> copy, which shows &ldquo;Last updated on 14 July 2026&rdquo; and may not reflect later revisions. We ran the <code>find</code> and <code>grep</code> commands against a mock directory tree and sample log lines, and the SQL against SQLite, not against a production store.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>KEV addition 2026-10-27, due 2026-11-10, CWE-434, ransomware use &ldquo;Unknown&rdquo;</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities.json">CISA KEV catalog (kev-data)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Affected and unaffected versions; CVSS 3.1 9.3 vector; &ldquo;requires user interaction&rdquo;</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/48xxx/CVE-2026-48356.json">CVE-2026-48356 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA SSVC Exploitation &ldquo;active&rdquo;</td>
          <td>CISA ADP in the same CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Published July 14, 2026; priority 2; fixed builds; 9.6 score; &ldquo;not aware of any exploits&rdquo; at publication</td>
          <td><a href="https://helpx.adobe.com/security/products/magento/apsb26-73.html">Adobe APSB26-73</a> (regional helpx copy)</td>
          <td>Qualified: regional copy</td>
      </tr>
      <tr>
          <td>Guest-cart REST routes, upload path, RCE conditions, 2.4.9 fix and July backport</td>
          <td><a href="https://sansec.io/research/magento-polyshell">Sansec: PolyShell</a></td>
          <td>Qualified: researcher&rsquo;s account</td>
      </tr>
      <tr>
          <td>471 stores in one hour on March 30, 2026; accesson.php</td>
          <td><a href="https://sansec.io/research/polyshell-mass-attack-wave">Sansec: mass PolyShell wave</a></td>
          <td>Qualified: researcher&rsquo;s account</td>
      </tr>
      <tr>
          <td>953-store campaign; 471 via CVE-2026-48356; loader domains</td>
          <td><a href="https://thehackernews.com/2026/10/hackers-infect-953-online-stores-with.html">The Hacker News, Oct 26, 2026</a> citing Sansec</td>
          <td>Qualified: press report</td>
      </tr>
      <tr>
          <td>Stock nginx deny rules; custom_options .htaccess; cms_block / admin_user columns</td>
          <td><a href="https://github.com/magento/magento2/tree/2.4-develop">magento/magento2 2.4-develop</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Brotli CVE-2026-111497: Find and Patch Every Copy</title><link>https://firewallsync.com/posts/brotli-cve-2026-111497-find-and-patch-every-copy/</link><guid>https://firewallsync.com/posts/brotli-cve-2026-111497-find-and-patch-every-copy/</guid><pubDate>Wed, 28 Oct 2026 04:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CVE-2026-111497 is an out-of-bounds write in Google's Brotli decoder, now on CISA's KEV list. It ships inside Node.js, Python packages, Go modules, Envoy and more, and patched builds can still report the old Brotli version. Here is how to find every copy and check it for the fix.</description><enclosure url="https://firewallsync.com/posts/brotli-cve-2026-111497-find-and-patch-every-copy/cover_hu14059145143623990521.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/brotli-cve-2026-111497-find-and-patch-every-copy/cover_hu14059145143623990521.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/brotli-cve-2026-111497-find-and-patch-every-copy/cover_hu14059145143623990521.jpg" alt="Round straw bales stacked in tight columns, each bound with thin blue netting" width="1200" height="630"></p><p>CVE-2026-111497 is a memory-corruption bug in Google&rsquo;s Brotli decompressor. Upstream versions 1.0.3 through 1.2.0 are affected, and 1.0.10, 1.1.1 and 1.2.1 fix it. CISA added it to the <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">Known Exploited Vulnerabilities catalog</a> on October 27, 2026, with a federal remediation date of November 10. The hard part is not the patch. Brotli is compiled into Node.js, Python wheels, Go modules, proxies and service meshes, and several patched builds still report the old version number. Here is where it hides, what &ldquo;fixed&rdquo; looks like for each, and a tested way to check code you vendor yourself.</p>
<h2 id="are-you-affected">Are you affected?</h2>
<p>You are exposed if a vulnerable Brotli decoder processes data an attacker can supply. The <a href="https://www.cve.org/CVERecord?id=CVE-2026-111497">CVE record</a> says an attacker with &ldquo;a crafted Brotli-compressed stream&rdquo; can &ldquo;corrupt adjacent decoder state memory, disclose sensitive memory contents, cause a denial of service, or potentially execute arbitrary code&rdquo;. Google&rsquo;s CNA scored it CVSS 4.0 9.2 (Critical). The Node.js project rates its copy High.</p>
<table>
  <thead>
      <tr>
          <th>Where Brotli lives</th>
          <th>Affected</th>
          <th>Fixed in</th>
          <th>How to check</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Google Brotli / libbrotli</td>
          <td>1.0.3 to 1.2.0</td>
          <td>1.0.10, 1.1.1, 1.2.1</td>
          <td>Package version, or the code check below for vendored copies</td>
          <td>CVE record, upstream CHANGELOG</td>
      </tr>
      <tr>
          <td>Node.js 22, 24, 26</td>
          <td>Earlier builds on those lines</td>
          <td>22.23.4, 24.22.1, 26.12.1</td>
          <td><code>node -v</code> (not <code>process.versions.brotli</code>)</td>
          <td>Node.js advisory</td>
      </tr>
      <tr>
          <td>End-of-life Node.js</td>
          <td>Assume affected</td>
          <td>None: upgrade</td>
          <td><code>node -v</code></td>
          <td>Node.js advisory</td>
      </tr>
      <tr>
          <td>Python <code>Brotli</code></td>
          <td>1.0.3 to 1.2.0</td>
          <td>1.2.1</td>
          <td><code>pip show Brotli</code></td>
          <td>Aikido; PyPI release date</td>
      </tr>
      <tr>
          <td>Python <code>brotlicffi</code></td>
          <td>1.2.0.x</td>
          <td>1.2.1.0</td>
          <td><code>pip show brotlicffi</code></td>
          <td>Aikido; PyPI release date</td>
      </tr>
      <tr>
          <td>Go <code>github.com/andybalholm/brotli</code></td>
          <td>Before v1.2.7</td>
          <td>v1.2.7</td>
          <td><code>go list -m github.com/andybalholm/brotli</code></td>
          <td>CVE record reference, source diff</td>
      </tr>
      <tr>
          <td>Envoy</td>
          <td>1.36.11, 1.37.7, 1.38.5, 1.39.2</td>
          <td>1.36.12, 1.37.8, 1.38.6, 1.39.3, 1.40.0</td>
          <td><code>envoy --version</code></td>
          <td>Wiz (qualified); we checked 1.39.2 against 1.39.3</td>
      </tr>
      <tr>
          <td>Istio</td>
          <td>1.29.0 to 1.29.8, 1.30.0 to 1.30.5, 1.31.0 to 1.31.1</td>
          <td>1.29.9, 1.30.6, 1.31.2</td>
          <td><code>istioctl version</code></td>
          <td>Wiz (qualified)</td>
      </tr>
      <tr>
          <td>Bun</td>
          <td>1.0.21 to 1.4.3</td>
          <td>1.4.4</td>
          <td><code>bun --version</code></td>
          <td>Wiz (qualified)</td>
      </tr>
      <tr>
          <td>Distro packages (libbrotli1, brotli)</td>
          <td>Depends on backport</td>
          <td>Distro security update</td>
          <td>Your distro&rsquo;s security tracker</td>
          <td>—</td>
      </tr>
  </tbody>
</table>
<p>The Envoy, Istio and Bun ranges come from <a href="https://www.wiz.io/blog/cve-2026-111497-brotli-rce-node-js">Wiz&rsquo;s write-up</a>, not from those projects&rsquo; advisories. We did not read those advisories directly. For Envoy we checked the source instead: <code>bazel/repository_locations.bzl</code> pins Brotli <code>1.2.0</code> at tag v1.39.2, while v1.39.3 pins a Brotli commit whose <code>decode.c</code> contains the fix.</p>
<p>Linux distributions usually backport fixes and keep the upstream version. Our Ubuntu 24.04 sandbox, for example, has <code>libbrotli1 1.1.0-2ubuntu0.1</code>. That string alone does not tell you whether the fix is in. Use your distribution&rsquo;s security tracker for the CVE.</p>
<h3 id="why-the-version-string-can-mislead-you">Why the version string can mislead you</h3>
<p>Node.js patched its bundled Brotli in place. On the fixed Node.js 22.23.4 the reported library version is still the old one:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-console" data-lang="console"><span style="display:flex;"><span>$ node -v
</span></span><span style="display:flex;"><span>v22.23.4
</span></span><span style="display:flex;"><span>$ node -p process.versions.brotli
</span></span><span style="display:flex;"><span>1.1.0
</span></span></code></pre></div><p>We ran that in our sandbox. The source tree agrees: <code>deps/brotli/c/common/version.h</code> declares 1.1.0 at both v22.23.3 and v22.23.4, and 1.2.0 at v24.22.1 and v26.12.1, but only the fixed tags contain the patched decoder (v22.23.3 and v24.22.0 do not). A scanner that flags <code>process.versions.brotli &lt; 1.2.1</code> will raise false positives on fixed Node.js. A scanner that trusts a vendored <code>version.h</code> can miss real exposure the other way round. For Node.js, use the Node.js version. For anything else you vendor, check the code.</p>
<h2 id="what-the-bug-is">What the bug is</h2>
<p>The <a href="https://nodejs.org/en/blog/vulnerability/october-2026-security-releases">Node.js advisory</a> gives the clearest description: &ldquo;An invalid block switch was reported as a temporary need for more input instead of a permanent format error, so the streaming decoder kept appending input bytes to an eight-byte internal buffer.&rdquo;</p>
<p>The upstream diff between v1.2.0 and v1.2.1 of <code>c/dec/decode.c</code> shows exactly that. <code>DecodeBlockTypeAndLength()</code> used to return a plain boolean, so &ldquo;invalid stream&rdquo; and &ldquo;need more input&rdquo; looked the same, and the non-safe callers ignored the result altogether (<code>static void DecodeLiteralBlockSwitch(...)</code>). After the fix the function returns a <code>BrotliDecoderErrorCode</code>, emits <code>BROTLI_DECODER_ERROR_FORMAT_BLOCK_SWITCH</code> for the invalid case, and every caller checks it through a new <code>BROTLI_SAFE_WITH_STATUS</code> macro. That matches the two weaknesses in the CVE record: CWE-252 (unchecked return value) and CWE-787 (out-of-bounds write).</p>
<p>The flaw was found by Khai Tran and Sylvie Mayer of OpenAI Codex Security, who are credited in both the CVE record and the Node.js advisory.</p>
<h2 id="where-untrusted-brotli-reaches-your-code">Where untrusted Brotli reaches your code</h2>
<p>The flaw is in decompression, so look for places that decode data you did not produce:</p>
<ul>
<li><strong>Request bodies.</strong> Any endpoint or middleware that inflates <code>Content-Encoding: br</code> uploads.</li>
<li><strong>Outbound HTTP.</strong> HTTP clients decode Brotli responses for you. If an attacker controls or can redirect the URL (webhooks, link previews, feed fetchers, scrapers), they control the compressed bytes.</li>
<li><strong>Files and archives</strong> that contain Brotli streams.</li>
<li><strong>Proxies and meshes</strong> that decompress traffic to inspect or transform it.</li>
</ul>
<p>Two behaviours we checked, because they mean code that never mentions Brotli can still run it:</p>
<ul>
<li><strong>Node.js <code>fetch()</code> decodes Brotli it never asked for.</strong> On Node.js 22.23.4, <code>fetch()</code> sent <code>accept-encoding: gzip, deflate</code>. When our local test server replied with <code>content-encoding: br</code>, <code>fetch()</code> decompressed the body anyway. A hostile server can choose Brotli whatever the client advertises.</li>
<li><strong>Python&rsquo;s requests turns Brotli on when it is installed.</strong> urllib3&rsquo;s <code>util/request.py</code> appends <code>br</code> to its default <code>Accept-Encoding</code> when <code>brotlicffi</code> or <code>brotli</code> can be imported. With Brotli 1.2.1 installed, <code>requests.utils.default_headers()</code> returned <code>gzip, deflate, br</code> in our test. The decoding then runs in whichever Brotli package is installed.</li>
</ul>
<h2 id="what-to-do-now">What to do now</h2>
<ol>
<li><strong>Update the runtimes.</strong> Node.js to 22.23.4, 24.22.1 or 26.12.1 or later. Move off end-of-life Node.js lines: the advisory notes that &ldquo;End-of-Life versions are always affected when a security release occurs.&rdquo;</li>
<li><strong>Update language packages</strong> in every lockfile and image: <code>Brotli&gt;=1.2.1</code>, <code>brotlicffi&gt;=1.2.1.0</code>, <code>github.com/andybalholm/brotli v1.2.7</code>. Both Python fixes were published to PyPI on October 22, 2026.</li>
<li><strong>Update proxies and meshes</strong> (Envoy, Istio) and Bun to the fixed releases above. Confirm the ranges against each project&rsquo;s own advisory.</li>
<li><strong>Apply distro updates</strong> for <code>libbrotli</code>/<code>brotli</code>, then restart the long-running processes that loaded the old library.</li>
<li><strong>Rebuild container images.</strong> A fixed base image does not help a running container that was built from the old one.</li>
<li><strong>Hunt for vendored copies</strong> with the check below.</li>
</ol>
<p>If you cannot patch a service today, reduce what it decodes. As a stopgap, reject <code>Content-Encoding: br</code> request bodies at the edge for services that do not need them, and stop server-side fetchers from decoding responses from untrusted URLs. This is our editorial suggestion, not vendor guidance, and it does not replace the update.</p>
<h3 id="inventory-commands">Inventory commands</h3>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-sh" data-lang="sh"><span style="display:flex;"><span><span style="color:#75715e"># Node.js: the version that matters</span>
</span></span><span style="display:flex;"><span>node -v
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Python: every environment and image</span>
</span></span><span style="display:flex;"><span>pip show Brotli brotlicffi 2&gt;/dev/null | grep -E <span style="color:#e6db74">&#39;^(Name|Version)&#39;</span>
</span></span><span style="display:flex;"><span>pipdeptree -r -p Brotli          <span style="color:#75715e"># who pulls it in</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Go modules</span>
</span></span><span style="display:flex;"><span>go list -m all | grep andybalholm/brotli
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Debian/Ubuntu and RHEL-family system libraries</span>
</span></span><span style="display:flex;"><span>dpkg -l <span style="color:#e6db74">&#39;libbrotli*&#39;</span> 2&gt;/dev/null | awk <span style="color:#e6db74">&#39;/^ii/{print $2, $3}&#39;</span>
</span></span><span style="display:flex;"><span>rpm -q brotli libbrotli 2&gt;/dev/null
</span></span></code></pre></div><p>The <code>pip</code>, <code>node</code> and <code>dpkg</code> lines were run in our sandbox. The <code>go</code>, <code>pipdeptree</code> and <code>rpm</code> lines are standard tool usage that we did not run here.</p>
<h3 id="checking-vendored-source-for-the-fix">Checking vendored source for the fix</h3>
<p>C and C++ projects often copy <code>c/dec/decode.c</code> into their tree, and Go programs carry <code>decode.go</code> in the module cache or a <code>vendor/</code> directory. This script looks for the patched function signature instead of trusting any version string:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-sh" data-lang="sh"><span style="display:flex;"><span><span style="color:#75715e">#!/bin/sh
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span><span style="color:#75715e"># brotli-fixcheck.sh DIR...: find vendored Brotli decoder sources and report</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># whether they carry the CVE-2026-111497 fix. Looks at the code, not the</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># version string, because patched Node.js still reports the old version.</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> root in <span style="color:#e6db74">&#34;</span>$@<span style="color:#e6db74">&#34;</span>; <span style="color:#66d9ef">do</span>
</span></span><span style="display:flex;"><span>  find <span style="color:#e6db74">&#34;</span>$root<span style="color:#e6db74">&#34;</span> <span style="color:#ae81ff">\(</span> -name decode.c -path <span style="color:#e6db74">&#39;*dec*&#39;</span> <span style="color:#ae81ff">\)</span> -o <span style="color:#ae81ff">\(</span> -name decode.go -path <span style="color:#e6db74">&#39;*brotli*&#39;</span> <span style="color:#ae81ff">\)</span> 2&gt;/dev/null |
</span></span><span style="display:flex;"><span>  <span style="color:#66d9ef">while</span> read -r f; <span style="color:#66d9ef">do</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> grep -q <span style="color:#e6db74">&#39;BrotliDecoderErrorCode DecodeBlockTypeAndLength&#39;</span> <span style="color:#e6db74">&#34;</span>$f<span style="color:#e6db74">&#34;</span> <span style="color:#f92672">||</span>
</span></span><span style="display:flex;"><span>       grep -q <span style="color:#e6db74">&#39;func decodeBlockTypeAndLength(.*) int {&#39;</span> <span style="color:#e6db74">&#34;</span>$f<span style="color:#e6db74">&#34;</span>; <span style="color:#66d9ef">then</span>
</span></span><span style="display:flex;"><span>      echo <span style="color:#e6db74">&#34;PATCHED    </span>$f<span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">elif</span> grep -q <span style="color:#e6db74">&#39;DecodeBlockTypeAndLength\|decodeBlockTypeAndLength&#39;</span> <span style="color:#e6db74">&#34;</span>$f<span style="color:#e6db74">&#34;</span>; <span style="color:#66d9ef">then</span>
</span></span><span style="display:flex;"><span>      echo <span style="color:#e6db74">&#34;VULNERABLE </span>$f<span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">fi</span>
</span></span><span style="display:flex;"><span>  <span style="color:#66d9ef">done</span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">done</span>
</span></span></code></pre></div><p>We tested it against 12 sources pulled from GitHub: upstream Brotli v1.0.9, v1.0.10, v1.1.0, v1.1.1, v1.2.0 and v1.2.1; Node.js <code>deps/brotli</code> at v22.23.3, v22.23.4, v24.22.0 and v24.22.1; and <code>andybalholm/brotli</code> v1.2.6 and v1.2.7. It reported every fixed release as <code>PATCHED</code> and every earlier one as <code>VULNERABLE</code>. Run it over source trees, <code>vendor/</code> directories and <code>$(go env GOMODCACHE)</code>.</p>
<p>It cannot inspect compiled binaries, and a heavily modified fork may not match either pattern. Treat &ldquo;no output&rdquo; as &ldquo;no recognisable copy found&rdquo;, not as &ldquo;safe&rdquo;.</p>
<h2 id="what-to-hunt-for">What to hunt for</h2>
<p>No vendor or agency has published indicators of compromise for this CVE, and CISA&rsquo;s KEV entry does not require forensic triage (<code>forensicTriage: No</code>). Practical checks:</p>
<ul>
<li><strong>Crashes.</strong> Segfaults or aborts in Node.js, Python or proxy processes since late October, especially in services that fetch remote content or accept compressed uploads.</li>
<li><strong>Unexpected Brotli.</strong> Requests with <code>Content-Encoding: br</code> to endpoints that never receive compressed bodies. Wiz recommends reviewing WAF, application and network logs for suspicious Brotli-compressed traffic.</li>
<li><strong>Outbound fetches</strong> from services that took user-supplied URLs, close in time to a crash.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We did not reproduce the memory corruption or test any exploit. Our checks cover version strings, source code and HTTP client behaviour in a sandbox: Node.js 22.23.4, Python with Brotli 1.2.1, Ubuntu 24.04. The Envoy, Istio and Bun ranges come from a security vendor&rsquo;s summary. The list of affected software is not complete. Wiz notes that other products that bundle or link vulnerable versions of the Brotli C decoder may also be affected.</p>
<p>For the wider process of keeping dependency updates moving, see <a href="/posts/dependency-audits-without-slowing-down-releases/">dependency audits without slowing down releases</a>. For publishing or reading &ldquo;not affected&rdquo; statements about a library like this, see <a href="/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/">VEX explained</a>. Last week&rsquo;s <a href="/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/">PHP2Shell patch-and-hunt guide</a> covers the two PHP runtime flaws CISA added to KEV the day before.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Affected 1.0.3–1.2.0; fixed 1.0.10, 1.1.1, 1.2.1; CVSS 4.0 9.2; CWE-252/787; credit</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/111xxx/CVE-2026-111497.json">CVE-2026-111497 record (cvelistV5)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV added 2026-10-27, due 2026-11-10, forensicTriage No; SSVC exploitation active</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data</a>; CISA ADP in the CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>1.2.1 released 2026-10-21 with the CVE fix; code change</td>
          <td><a href="https://github.com/google/brotli/blob/v1.2.1/CHANGELOG.md">google/brotli CHANGELOG</a>; <code>c/dec/decode.c</code> v1.2.0 vs v1.2.1</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Node.js fixed releases, root cause, EOL statement</td>
          <td><a href="https://nodejs.org/en/blog/vulnerability/october-2026-security-releases">Node.js October 2026 security releases</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Patched Node.js reports the old Brotli version</td>
          <td><code>node -p process.versions.brotli</code> on v22.23.4; <code>deps/brotli</code> at the listed tags</td>
          <td>Verified (our test)</td>
      </tr>
      <tr>
          <td>Go fix in andybalholm/brotli v1.2.7</td>
          <td><a href="https://github.com/andybalholm/brotli/releases/tag/v1.2.7">Release v1.2.7</a>; <code>decode.go</code> v1.2.6 vs v1.2.7</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Python Brotli 1.2.1, brotlicffi 1.2.1.0</td>
          <td><a href="https://www.aikido.dev/blog/python-packages-vulnerable-node2shell-brotli-cve-2026-111497">Aikido</a>; PyPI upload dates 2026-10-22</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>urllib3/requests advertise <code>br</code> when Brotli is installed</td>
          <td><a href="https://github.com/urllib3/urllib3/blob/main/src/urllib3/util/request.py">urllib3 <code>util/request.py</code></a>; our test</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Envoy, Istio, Bun ranges</td>
          <td><a href="https://www.wiz.io/blog/cve-2026-111497-brotli-rce-node-js">Wiz</a>; Envoy <code>repository_locations.bzl</code> at v1.39.2/v1.39.3</td>
          <td>Qualified</td>
      </tr>
      <tr>
          <td>Envoy and other consumers affected</td>
          <td><a href="https://www.openwall.com/lists/oss-security/2026/10/23/25">oss-security, 2026-10-23</a>, <a href="https://www.openwall.com/lists/oss-security/2026/10/24/6">2026-10-24</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>PHP2Shell CVE-2026-109618 and -109619: Patch, Then Hunt</title><link>https://firewallsync.com/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/</link><guid>https://firewallsync.com/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/</guid><pubDate>Tue, 27 Oct 2026 04:50:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>Two PHP core flaws, CVE-2026-109618 and CVE-2026-109619, allow unauthenticated remote code execution against PHP that serves web requests. CISA added both to the KEV catalog on October 26 with a three-day deadline. Here are the fixed versions, why php -v can mislead you on distribution builds, the stopgaps PHP itself recommends, and what to hunt for.</description><enclosure url="https://firewallsync.com/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/cover_hu12286854602341724834.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/cover_hu12286854602341724834.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/php2shell-cve-2026-109618-cve-2026-109619-patch-and-hunt/cover_hu12286854602341724834.jpg" alt="Close-up of yellow barrier tape printed with the word ATTENTION, stretched in front of a blurred orange road sign and greenery" width="1200" height="630"></p><p>CVE-2026-109618 and CVE-2026-109619, which the security community has nicknamed &ldquo;PHP2Shell&rdquo;, are two memory-safety flaws in PHP&rsquo;s core that can give an unauthenticated attacker remote code execution on a server running PHP for the web. The fixes are PHP <strong>8.2.35, 8.3.36, 8.4.27 and 8.5.12</strong>. CISA added both CVEs to its Known Exploited Vulnerabilities (KEV) catalog on October 26, 2026, and gave federal agencies until <strong>October 29</strong> to act.</p>
<p>These are bugs in the PHP interpreter, so they apply regardless of your framework or CMS. If you run PHP behind a web server, update the PHP package today, restart every PHP worker, then look for signs that someone got in first. Everything below comes from the PHP project&rsquo;s advisory and CVE records and CISA&rsquo;s KEV entry unless we say otherwise.</p>
<h2 id="are-you-affected">Are you affected?</h2>
<p>The CVE records list PHP 8.2 before 8.2.35, 8.3 before 8.3.36, 8.4 before 8.4.27 and 8.5 before 8.5.12 as affected. PHP&rsquo;s advisory adds who is at risk: &ldquo;all PHP installations serving web requests, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP and LiteSpeed.&rdquo; It rates mod_php highest, &ldquo;with a demonstrated exploit where an unauthenticated request can trigger the attack before any application code runs.&rdquo;</p>
<table>
  <thead>
      <tr>
          <th>Branch</th>
          <th>Vulnerable upstream</th>
          <th>Fixed upstream</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>8.5</td>
          <td>below 8.5.12</td>
          <td>8.5.12</td>
      </tr>
      <tr>
          <td>8.4</td>
          <td>below 8.4.27</td>
          <td>8.4.27</td>
      </tr>
      <tr>
          <td>8.3</td>
          <td>below 8.3.36</td>
          <td>8.3.36</td>
      </tr>
      <tr>
          <td>8.2</td>
          <td>below 8.2.35</td>
          <td>8.2.35</td>
      </tr>
      <tr>
          <td>8.1 and older</td>
          <td>not listed in the CVE records; no upstream fix</td>
          <td>vendor backport, PHP&rsquo;s source backport branches for 7.4/8.0/8.1, or upgrade</td>
      </tr>
  </tbody>
</table>
<p>The CVE records only list 8.2 to 8.5. Do not read that as &ldquo;8.1 is safe&rdquo;: PHP says 8.1 and older &ldquo;no longer receive updates&rdquo; and published source backports for 7.4, 8.0 and 8.1 because of the severity.</p>
<h3 id="why-php--v-can-mislead-you">Why <code>php -v</code> can mislead you</h3>
<p>Most production PHP comes from a distribution or a container base image, and those vendors backport fixes without changing the upstream version number. PHP says Linux distributions and LTS vendors &ldquo;received the patches in advance&rdquo;.</p>
<p>We checked an Ubuntu 24.04 host. <code>php -v</code> reports 8.3.6, which is far below 8.3.36, yet the package changelog shows the fix:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-console" data-lang="console"><span style="display:flex;"><span>$ dpkg -l php8.3-cli | tail -1 | awk <span style="color:#e6db74">&#39;{print $2, $3}&#39;</span>
</span></span><span style="display:flex;"><span>php8.3-cli 8.3.6-0ubuntu0.24.04.13
</span></span><span style="display:flex;"><span>$ zcat /usr/share/doc/php8.3-cli/changelog.Debian.gz | head -4
</span></span><span style="display:flex;"><span>php8.3 (8.3.6-0ubuntu0.24.04.13) noble-security; urgency=medium
</span></span><span style="display:flex;"><span><span style="color:#960050;background-color:#1e0010">
</span></span></span><span style="display:flex;"><span><span style="color:#960050;background-color:#1e0010"></span>  * SECURITY UPDATE: Remote RCE via state corruption issue
</span></span><span style="display:flex;"><span>    - debian/patches/CVE-2026-109618_109619.patch: Avoid state corruption via
</span></span></code></pre></div><p>So for a distribution build, search the package changelog for the CVE IDs. On Debian or Ubuntu, <code>zcat /usr/share/doc/php*-common/changelog.Debian.gz | grep -m1 CVE-2026-109618</code> works; on RPM-based systems the equivalent is <code>rpm -q --changelog &lt;php package&gt; | grep CVE-2026-109618</code> (illustrative; we did not run the RPM form). For official <code>php</code> Docker images and source builds, the upstream version is the one that counts.</p>
<p>Also check the binary that actually serves traffic. The CLI, <code>php-fpm</code> and <code>libapache2-mod-php</code> can come from different packages and be updated at different times.</p>
<h3 id="a-version-check-for-upstream-builds">A version check for upstream builds</h3>
<p>This script compares upstream version strings against the fixed releases. We ran it with PHP 8.3 CLI (<code>php php2shell-check.php 8.2.34 8.4.27 8.1.31</code>). It cannot see distribution backports, so a &ldquo;VULNERABLE&rdquo; result on a distro build means &ldquo;check the changelog&rdquo;, not &ldquo;definitely exposed&rdquo;.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-php" data-lang="php"><span style="display:flex;"><span><span style="color:#f92672">&lt;?</span><span style="color:#a6e22e">php</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">// Compares upstream PHP versions with the fixed releases for
</span></span></span><span style="display:flex;"><span><span style="color:#75715e">// CVE-2026-109618 / CVE-2026-109619. Does not detect vendor backports.
</span></span></span><span style="display:flex;"><span><span style="color:#75715e"></span>$fixed <span style="color:#f92672">=</span> [<span style="color:#e6db74">&#39;8.2&#39;</span> <span style="color:#f92672">=&gt;</span> <span style="color:#e6db74">&#39;8.2.35&#39;</span>, <span style="color:#e6db74">&#39;8.3&#39;</span> <span style="color:#f92672">=&gt;</span> <span style="color:#e6db74">&#39;8.3.36&#39;</span>, <span style="color:#e6db74">&#39;8.4&#39;</span> <span style="color:#f92672">=&gt;</span> <span style="color:#e6db74">&#39;8.4.27&#39;</span>, <span style="color:#e6db74">&#39;8.5&#39;</span> <span style="color:#f92672">=&gt;</span> <span style="color:#e6db74">&#39;8.5.12&#39;</span>];
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">function</span> <span style="color:#a6e22e">status</span>(<span style="color:#a6e22e">string</span> $v, <span style="color:#66d9ef">array</span> $fixed)<span style="color:#f92672">:</span> <span style="color:#a6e22e">string</span> {
</span></span><span style="display:flex;"><span>    $branch <span style="color:#f92672">=</span> <span style="color:#a6e22e">implode</span>(<span style="color:#e6db74">&#39;.&#39;</span>, <span style="color:#a6e22e">array_slice</span>(<span style="color:#a6e22e">explode</span>(<span style="color:#e6db74">&#39;.&#39;</span>, $v), <span style="color:#ae81ff">0</span>, <span style="color:#ae81ff">2</span>));
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> (<span style="color:#f92672">!</span><span style="color:#a6e22e">isset</span>($fixed[$branch])) {
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#a6e22e">version_compare</span>($v, <span style="color:#e6db74">&#39;8.2&#39;</span>, <span style="color:#e6db74">&#39;&lt;&#39;</span>)
</span></span><span style="display:flex;"><span>            <span style="color:#f92672">?</span> <span style="color:#e6db74">&#39;END OF LIFE: no upstream fix; needs vendor backport or upgrade&#39;</span>
</span></span><span style="display:flex;"><span>            <span style="color:#f92672">:</span> <span style="color:#e6db74">&#39;UNKNOWN branch: check php.net&#39;</span>;
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#a6e22e">version_compare</span>($v, $fixed[$branch], <span style="color:#e6db74">&#39;&gt;=&#39;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#f92672">?</span> <span style="color:#e6db74">&#34;fixed upstream (&gt;= </span><span style="color:#e6db74">{</span>$fixed[$branch]<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#f92672">:</span> <span style="color:#e6db74">&#34;VULNERABLE upstream (&lt; </span><span style="color:#e6db74">{</span>$fixed[$branch]<span style="color:#e6db74">}</span><span style="color:#e6db74">) unless your vendor backported the fix&#34;</span>;
</span></span><span style="display:flex;"><span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">foreach</span> (<span style="color:#a6e22e">array_slice</span>($argv, <span style="color:#ae81ff">1</span>) <span style="color:#f92672">?:</span> [<span style="color:#66d9ef">PHP_VERSION</span>] <span style="color:#66d9ef">as</span> $v) {
</span></span><span style="display:flex;"><span>    <span style="color:#a6e22e">printf</span>(<span style="color:#e6db74">&#34;%-10s %s</span><span style="color:#ae81ff">\n</span><span style="color:#e6db74">&#34;</span>, $v, <span style="color:#a6e22e">status</span>($v, $fixed));
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-console" data-lang="console"><span style="display:flex;"><span>$ php php2shell-check.php 8.2.34 8.4.27 8.1.31
</span></span><span style="display:flex;"><span>8.2.34     VULNERABLE upstream (&lt; 8.2.35) unless your vendor backported the fix
</span></span><span style="display:flex;"><span>8.4.27     fixed upstream (&gt;= 8.4.27)
</span></span><span style="display:flex;"><span>8.1.31     END OF LIFE: no upstream fix; needs vendor backport or upgrade
</span></span></code></pre></div><h2 id="what-the-two-flaws-do">What the two flaws do</h2>
<p>Both bugs leave PHP&rsquo;s internal request state inconsistent and let later code trust it.</p>
<ul>
<li><strong>CVE-2026-109618, &ldquo;RCE via OOM corruption&rdquo;.</strong> The CVE record says PHP &ldquo;uses longjmp() for out-of-memory (OOM) errors to exit the current context&rdquo;. Some code paths &ldquo;leave request memory in an uninitialized state before performing an allocation&rdquo;, and &ldquo;the subsequent request cleanup procedure may read or branch on this memory.&rdquo; An attacker who can push a request into an out-of-memory failure at the right moment can trigger it.</li>
<li><strong>CVE-2026-109619, &ldquo;RCE via stale global state&rdquo;.</strong> Global state that &ldquo;temporarily points into request memory&rdquo; may not be reset, so it &ldquo;may refer to request memory even after it has been freed.&rdquo; PHP&rsquo;s advisory explains the consequence: &ldquo;A later request handled by the same worker may use that stale state before normal initialization resets it.&rdquo;</li>
</ul>
<p>Both carry a CVSS v4.0 base score of 9.2 (<code>AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H</code>): network-reachable, no privileges, no user interaction, but high attack complexity with attack requirements present. If you want to decode the vector, see our guide to <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">reading a CVSS v4.0 vector string</a>. The fix shipped in all four branches as one change, described in PHP&rsquo;s NEWS file as &ldquo;Avoid state corruption via bailout&rdquo;.</p>
<p>Credits in the CVE records name Wiz, Ryan at Calif.io and OpenAI as reporters or finders.</p>
<h2 id="exploitation-status">Exploitation status</h2>
<p>The public picture is mixed, so here is what each source says:</p>
<table>
  <thead>
      <tr>
          <th>Date</th>
          <th>Source</th>
          <th>What it says</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Oct 17</td>
          <td>Amazon Threat Intelligence, as reported by Infosecurity Magazine</td>
          <td>An exploitation attempt against its MadPot honeypot &ldquo;dating October 17&rdquo;, four days before disclosure</td>
      </tr>
      <tr>
          <td>Oct 21</td>
          <td>PHP advisory</td>
          <td>A demonstrated exploit against mod_php; patches released</td>
      </tr>
      <tr>
          <td>Oct 21</td>
          <td>oss-security disclosure</td>
          <td>&ldquo;a complete FPM exploit has not been demonstrated&rdquo;, though source analysis suggests FPM is susceptible</td>
      </tr>
      <tr>
          <td>Oct 21</td>
          <td>VulnCheck</td>
          <td>Scanning with a <code>php2shell-scan/109618</code> user agent, and one attempt that &ldquo;tried to (unsuccessfully) drop a webshell via a server-side session file&rdquo;</td>
      </tr>
      <tr>
          <td>Oct 23</td>
          <td>Security Affairs, citing honeypot operators</td>
          <td>&ldquo;no successful remote code execution has been confirmed in real-world attacks&rdquo;</td>
      </tr>
      <tr>
          <td>Oct 23</td>
          <td>FBI IC3 public service announcement</td>
          <td>Updates &ldquo;should be applied immediately&rdquo;</td>
      </tr>
      <tr>
          <td>Oct 26</td>
          <td>CISA KEV</td>
          <td>Both CVEs added; due October 29; forensic triage required for federal agencies</td>
      </tr>
  </tbody>
</table>
<p>CISA&rsquo;s Vulnrichment SSVC entry in the CVE records, dated October 21, still says <code>Exploitation: none</code>. That predates the KEV listing. For how these signals usually relate, see <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>.</p>
<p>Our reading: mass scanning started within hours, working exploits exist, and CISA has listed both as exploited. Whether specific attacks succeeded against production servers is not yet established in public sources. That uncertainty is a reason to hunt after patching, not a reason to wait.</p>
<h2 id="what-to-do-now">What to do now</h2>
<ol>
<li><strong>Inventory.</strong> Find every PHP runtime that serves HTTP: VMs, containers, base images, appliances with embedded PHP, and managed hosting. Container images need a rebuild and redeploy; patching the host does nothing for them.</li>
<li><strong>Update.</strong> Move to 8.2.35, 8.3.36, 8.4.27, 8.5.12 or your vendor&rsquo;s patched package. For 8.1 and older, use a vendor backport or upgrade. PHP 8.2 itself reaches end of life on December 31, 2026, according to PHP&rsquo;s release notes, so plan that upgrade too.</li>
<li><strong>Restart every worker.</strong> CVE-2026-109619 is about state that persists across requests in the same worker. Restart PHP-FPM pools, Apache (for mod_php) and any long-running PHP servers after the package update.</li>
<li><strong>Prioritise internet-facing mod_php</strong>, then FPM behind public web servers, then internal applications.</li>
</ol>
<h3 id="if-you-cannot-update-today">If you cannot update today</h3>
<p>PHP lists stopgaps and is explicit that &ldquo;only an update is a complete solution&rdquo; and these &ldquo;do not make an installation safe&rdquo;:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-ini" data-lang="ini"><span style="display:flex;"><span><span style="color:#75715e">; php.ini or FPM pool config, applied before request parsing, then restart workers.</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; Illustrative: check each value against your application&#39;s needs.</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; CVE-2026-109618 (OOM corruption)</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; In an FPM pool use php_admin_value[memory_limit] = -1 so applications can&#39;t override it.</span>
</span></span><span style="display:flex;"><span><span style="color:#a6e22e">memory_limit</span> <span style="color:#f92672">=</span> <span style="color:#e6db74">-1</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; 8M if the app takes no uploads, 32M if it needs them; keep upload_max_filesize below it</span>
</span></span><span style="display:flex;"><span><span style="color:#a6e22e">post_max_size</span> <span style="color:#f92672">=</span> <span style="color:#e6db74">8M</span>
</span></span><span style="display:flex;"><span><span style="color:#a6e22e">upload_max_filesize</span> <span style="color:#f92672">=</span> <span style="color:#e6db74">2M</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; CVE-2026-109619 (stale global state)</span>
</span></span><span style="display:flex;"><span><span style="color:#a6e22e">session.upload_progress.enabled</span> <span style="color:#f92672">=</span> <span style="color:#e6db74">0</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; PHP 8.5 only:</span>
</span></span><span style="display:flex;"><span><span style="color:#75715e">; fatal_error_backtraces = 0</span>
</span></span></code></pre></div><p>The reasoning behind each, from the advisory:</p>
<ul>
<li><code>memory_limit=-1</code> &ldquo;removes the controllable PHP memory limit failure, but an underlying allocation failure can still reach the vulnerable path.&rdquo; With no PHP memory limit, a container or cgroup memory limit becomes your availability boundary. PHP warns: &ldquo;Do not substitute <code>RLIMIT_AS</code> or <code>RLIMIT_DATA</code>, as these can cause the vulnerable allocation failure.&rdquo;</li>
<li>Low <code>post_max_size</code>, plus limits at the web server or proxy on decoded request size, field count, multipart part count, header count and nesting depth. These &ldquo;must also cover requests without a usable <code>Content-Length</code>.&rdquo;</li>
<li><code>session.upload_progress.enabled=0</code> &ldquo;blocks the confirmed session paths but not other variants&rdquo;. Set it in <code>php.ini</code>, the FPM pool or server config; PHP says &ldquo;Do not rely on <code>ini_set()</code> or FPM <code>.user.ini</code>.&rdquo;</li>
<li>Hiding errors does not help: &ldquo;Hiding diagnostics with <code>error_reporting</code>, <code>display_errors</code> or <code>@</code> does not help.&rdquo;</li>
</ul>
<p><code>php_admin_value</code> matters here because, per the PHP-FPM documentation, &ldquo;Settings defined with php_admin_value and php_admin_flag cannot be overridden with ini_set.&rdquo;</p>
<h2 id="what-to-hunt-for-after-patching">What to hunt for after patching</h2>
<p>Patching removes the bug, not an attacker who already used it. Preserve web server and PHP logs before you rotate or rebuild anything. The indicators below come from VulnCheck&rsquo;s published honeypot observations. They describe scanners and one attempt; real attackers can change every one of them.</p>
<ul>
<li><strong>User agents:</strong> <code>php2shell-scan/109618</code>, <code>php2shell-verify/1.0</code>, <code>Mozilla/5.0 (php2shell-poc)</code>.</li>
<li><strong>Requests:</strong> multipart POSTs to <code>/index.php</code> that upload two files named <code>a</code> and <code>b</code> through the upload-progress path; follow-up <code>GET</code> requests using <code>php://filter</code> to read a session file.</li>
<li><strong>Session files containing PHP code.</strong> The attempt VulnCheck saw tried to plant a webshell through a session file, with a payload that prints <code>PWNED::</code>. A legitimate PHP session file contains serialized data, never <code>&lt;?php</code>.</li>
</ul>
<p>Two checks you can run on a Linux host. We tested both against sample files; adjust paths to your log format and to the <code>session.save_path</code> of the pool that serves traffic (<code>php -i</code> shows the CLI&rsquo;s value, not necessarily FPM&rsquo;s):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span><span style="color:#75715e"># Web logs: known scanner user agents and php://filter reads</span>
</span></span><span style="display:flex;"><span>grep -hE <span style="color:#e6db74">&#39;php2shell|php://filter&#39;</span> /var/log/nginx/access.log* /var/log/apache2/access.log* 2&gt;/dev/null
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Session files that contain PHP code or the test marker</span>
</span></span><span style="display:flex;"><span>grep -lE <span style="color:#e6db74">&#39;&lt;\?php|PWNED::&#39;</span> /var/lib/php/sessions/sess_* 2&gt;/dev/null
</span></span></code></pre></div><p>Then do the usual post-compromise checks for a web server: new or changed <code>.php</code> files under the document root since October 17, unexpected child processes of the PHP worker or web server user (shells, <code>curl</code>, <code>wget</code>), and outbound connections from web hosts. Our guide to <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents</a> covers which of these you will wish you had recorded.</p>
<p>Federal agencies have to follow CISA&rsquo;s forensic triage requirements for this KEV entry. If you want to borrow that process, our explainer on <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">CISA BOD 26-04</a> covers it.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We have not read the full GitHub security advisories, which PHP says contain the technical details; they were not reachable from our research environment. Exploitation reporting comes from vendors and press summaries we cite, not from our own telemetry. The hunting indicators come from one vendor&rsquo;s honeypots and will not catch a modified attack. Exposure counts published by vendors differ widely, so we have not repeated them.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Affected ranges 8.2 &lt; 8.2.35, 8.3 &lt; 8.3.36, 8.4 &lt; 8.4.27, 8.5 &lt; 8.5.12; mechanisms; CVSS v4.0 9.2 vector; CWEs; credits</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/109xxx/CVE-2026-109618.json">CVE-2026-109618</a> and <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/109xxx/CVE-2026-109619.json">CVE-2026-109619</a> records (PHP CNA)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SAPIs at risk, mod_php highest, fixed releases, 8.1-and-older guidance, source backports, all stopgaps and their caveats</td>
          <td><a href="https://www.php.net/archive/2026.php">PHP security advisory, Oct 21, 2026</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Fix shipped in 8.2.35, 8.3.36, 8.4.27, 8.5.12 as &ldquo;Avoid state corruption via bailout&rdquo;</td>
          <td>PHP NEWS files for <a href="https://github.com/php/php-src/blob/php-8.4.27/NEWS">8.4.27</a> and the other tags</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Full FPM exploit not demonstrated at disclosure</td>
          <td><a href="https://openwall.com/lists/oss-security/2026/10/21/6">oss-security, Oct 21, 2026</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV addition Oct 26, due Oct 29, forensic triage flag</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV catalog data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Ubuntu 24.04 package 8.3.6-0ubuntu0.24.04.13 includes the fix</td>
          <td>Package changelog on an Ubuntu 24.04 host, inspected October 27, 2026</td>
          <td>Verified (one package)</td>
      </tr>
      <tr>
          <td>php_admin_value cannot be overridden with ini_set</td>
          <td><a href="https://www.php.net/manual/en/install.fpm.configuration.php">PHP-FPM configuration docs</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Exploitation attempt from Oct 17</td>
          <td>Amazon, as reported by <a href="https://infosecurity-magazine.com/news/php2shell-exploitation-attempt">Infosecurity Magazine</a></td>
          <td>Qualified (secondary report)</td>
      </tr>
      <tr>
          <td>Scanner user agents, request pattern, session-file webshell attempt</td>
          <td><a href="https://www.vulncheck.com/blog/php2shell-vulnerabilities">VulnCheck, Oct 21, 2026</a></td>
          <td>Qualified (one vendor&rsquo;s honeypots)</td>
      </tr>
      <tr>
          <td>No confirmed successful RCE as of Oct 23</td>
          <td><a href="https://securityaffairs.com/?p=201538">Security Affairs</a></td>
          <td>Qualified (secondary report)</td>
      </tr>
      <tr>
          <td>FBI: apply updates immediately</td>
          <td><a href="https://www.ic3.gov/PSA/2026/PSA261023">FBI IC3 PSA I-102326-PSA</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>PAN-OS CVE-2026-0322: Patch the Management Interface RCE</title><link>https://firewallsync.com/posts/pan-os-cve-2026-0322-management-interface-rce-patch-and-hunt/</link><guid>https://firewallsync.com/posts/pan-os-cve-2026-0322-management-interface-rce-patch-and-hunt/</guid><pubDate>Mon, 26 Oct 2026 04:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CVE-2026-0322 lets an authenticated attacker who can reach the PAN-OS management interface run code as root on PA-Series and VM-Series firewalls and Panorama. Palo Alto Networks has seen very limited exploitation, there is no workaround, and CISA gave federal agencies three days. Here are the fixed releases, a version checker built from the CVE record, and what to review after you patch.</description><enclosure url="https://firewallsync.com/posts/pan-os-cve-2026-0322-management-interface-rce-patch-and-hunt/cover_hu12832646850512031119.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/pan-os-cve-2026-0322-management-interface-rce-patch-and-hunt/cover_hu12832646850512031119.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/pan-os-cve-2026-0322-management-interface-rce-patch-and-hunt/cover_hu12832646850512031119.jpg" alt="A brass padlock hanging from the latch of a galvanised chain-link gate, with gravel on the ground behind it" width="1200" height="630"></p><p>CVE-2026-0322 is a code injection vulnerability in the management interface of Palo Alto Networks PAN-OS. An attacker who is authenticated and can reach that interface over the network can execute arbitrary code with root privileges on the firewall. It affects PA-Series and VM-Series firewalls and Panorama, there is no workaround, and Palo Alto Networks says it has observed &ldquo;very limited exploitation&rdquo;. CISA added it to the Known Exploited Vulnerabilities (KEV) catalog on October 14, 2026, and gave federal agencies until October 17 to act.</p>
<p>If you run PAN-OS, find your exact version, upgrade to the matching fixed release below, and use the upgrade window to check who can reach the management interface at all. Everything in this article comes from Palo Alto Networks&rsquo; CVE record and CISA&rsquo;s KEV entry unless we say otherwise.</p>
<h2 id="are-you-affected">Are you affected?</h2>
<p>You are affected if you run PAN-OS on a PA-Series or VM-Series firewall, or Panorama (virtual or M-Series), and your version is below the fixed release for its maintenance range. Palo Alto Networks says &ldquo;no special configuration is required to be affected&rdquo;. Cloud NGFW and Prisma Access are not impacted.</p>
<table>
  <thead>
      <tr>
          <th>Branch</th>
          <th>If you run</th>
          <th>Upgrade to</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>12.2</td>
          <td>12.2.0 through 12.2.3</td>
          <td>12.2.4 or later</td>
      </tr>
      <tr>
          <td>12.1</td>
          <td>12.1.8 through 12.1.10-h*</td>
          <td>12.1.10-h2 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>12.1.5 through 12.1.7-h*</td>
          <td>12.1.7-h8, or 12.1.10-h2 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>12.1.2 through 12.1.4-h*</td>
          <td>12.1.4-h12, or 12.1.10-h2 or later</td>
      </tr>
      <tr>
          <td>11.2</td>
          <td>11.2.14 through 11.2.15</td>
          <td>11.2.16 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.2.11 through 11.2.13-h*</td>
          <td>11.2.13-h3, or 11.2.16 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.2.8 through 11.2.10-h*</td>
          <td>11.2.10-h15, or 11.2.16 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.2.5 through 11.2.7-h*</td>
          <td>11.2.7-h21, or 11.2.16 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.2.0 through 11.2.4-h*</td>
          <td>11.2.4-h22, or 11.2.16 or later</td>
      </tr>
      <tr>
          <td>11.1</td>
          <td>11.1.14 through 11.1.16-h*</td>
          <td>11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.1.11 through 11.1.13-h*</td>
          <td>11.1.13-h15, or 11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.1.8 through 11.1.10-h*</td>
          <td>11.1.10-h35, or 11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.1.7 through 11.1.7-h*</td>
          <td>11.1.7-h11, or 11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.1.5 through 11.1.6-h*</td>
          <td>11.1.6-h39, or 11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>11.1.0 through 11.1.4-h*</td>
          <td>11.1.4-h37, or 11.1.16-h3 or later</td>
      </tr>
      <tr>
          <td>10.2</td>
          <td>10.2.17 through 10.2.18-h*</td>
          <td>10.2.18-h12 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>10.2.14 through 10.2.16-h*</td>
          <td>10.2.16-h11, or 10.2.18-h12 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>10.2.11 through 10.2.13-h*</td>
          <td>10.2.13-h26, or 10.2.18-h12 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>10.2.8 through 10.2.10-h*</td>
          <td>10.2.10-h41, or 10.2.18-h12 or later</td>
      </tr>
      <tr>
          <td></td>
          <td>10.2.0 through 10.2.7-h*</td>
          <td>10.2.7-h38, or 10.2.18-h12 or later</td>
      </tr>
  </tbody>
</table>
<p>This is the solutions table from the CVE record as published on October 14, 2026. Two details are easy to miss. First, a hotfix only fixes its own maintenance release: 11.1.4-h37 is fixed, but 11.1.5 is not, because 11.1.5 sits in the next row with its own fix (11.1.6-h39). Second, the record&rsquo;s affected range for 12.1 starts at 12.1.0, while the solutions table starts at 12.1.2; if you are on 12.1.0 or 12.1.1, move to 12.1.10-h2. Branches older than 10.2 are not listed; check the <a href="https://security.paloaltonetworks.com/CVE-2026-0322">Palo Alto Networks advisory</a> for their status.</p>
<h3 id="check-a-list-of-versions">Check a list of versions</h3>
<p>With a fleet, a lookup beats reading the table by eye. This Python 3 script encodes the table above. We checked it against the record: its 20 in-line fixes are the same set as the 20 <code>changes</code> entries in the record&rsquo;s <code>affected</code> data, and its per-branch upper bounds match the record&rsquo;s <code>lessThan</code> values. We ran it on Python 3 with the versions shown.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Check PAN-OS versions against the CVE-2026-0322 fix table (CVE record, 2026-10-14).
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 check_0322.py 11.1.6-h20 10.2.18-h12 ...&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> re<span style="color:#f92672">,</span> sys
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># (first version in range, last minor in range, in-line fix, newest fix on the branch)</span>
</span></span><span style="display:flex;"><span>ROWS <span style="color:#f92672">=</span> [
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;12.2.0&#34;</span>, <span style="color:#e6db74">&#34;12.2.3&#34;</span>, <span style="color:#e6db74">&#34;12.2.4&#34;</span>, <span style="color:#e6db74">&#34;12.2.4&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;12.1.8&#34;</span>, <span style="color:#e6db74">&#34;12.1.10&#34;</span>, <span style="color:#e6db74">&#34;12.1.10-h2&#34;</span>, <span style="color:#e6db74">&#34;12.1.10-h2&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;12.1.5&#34;</span>, <span style="color:#e6db74">&#34;12.1.7&#34;</span>, <span style="color:#e6db74">&#34;12.1.7-h8&#34;</span>, <span style="color:#e6db74">&#34;12.1.10-h2&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;12.1.2&#34;</span>, <span style="color:#e6db74">&#34;12.1.4&#34;</span>, <span style="color:#e6db74">&#34;12.1.4-h12&#34;</span>, <span style="color:#e6db74">&#34;12.1.10-h2&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.2.14&#34;</span>, <span style="color:#e6db74">&#34;11.2.15&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.2.11&#34;</span>, <span style="color:#e6db74">&#34;11.2.13&#34;</span>, <span style="color:#e6db74">&#34;11.2.13-h3&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.2.8&#34;</span>, <span style="color:#e6db74">&#34;11.2.10&#34;</span>, <span style="color:#e6db74">&#34;11.2.10-h15&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.2.5&#34;</span>, <span style="color:#e6db74">&#34;11.2.7&#34;</span>, <span style="color:#e6db74">&#34;11.2.7-h21&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.2.0&#34;</span>, <span style="color:#e6db74">&#34;11.2.4&#34;</span>, <span style="color:#e6db74">&#34;11.2.4-h22&#34;</span>, <span style="color:#e6db74">&#34;11.2.16&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.14&#34;</span>, <span style="color:#e6db74">&#34;11.1.16&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.11&#34;</span>, <span style="color:#e6db74">&#34;11.1.13&#34;</span>, <span style="color:#e6db74">&#34;11.1.13-h15&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.8&#34;</span>, <span style="color:#e6db74">&#34;11.1.10&#34;</span>, <span style="color:#e6db74">&#34;11.1.10-h35&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.7&#34;</span>, <span style="color:#e6db74">&#34;11.1.7&#34;</span>, <span style="color:#e6db74">&#34;11.1.7-h11&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.5&#34;</span>, <span style="color:#e6db74">&#34;11.1.6&#34;</span>, <span style="color:#e6db74">&#34;11.1.6-h39&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;11.1.0&#34;</span>, <span style="color:#e6db74">&#34;11.1.4&#34;</span>, <span style="color:#e6db74">&#34;11.1.4-h37&#34;</span>, <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;10.2.17&#34;</span>, <span style="color:#e6db74">&#34;10.2.18&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;10.2.14&#34;</span>, <span style="color:#e6db74">&#34;10.2.16&#34;</span>, <span style="color:#e6db74">&#34;10.2.16-h11&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;10.2.11&#34;</span>, <span style="color:#e6db74">&#34;10.2.13&#34;</span>, <span style="color:#e6db74">&#34;10.2.13-h26&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;10.2.8&#34;</span>, <span style="color:#e6db74">&#34;10.2.10&#34;</span>, <span style="color:#e6db74">&#34;10.2.10-h41&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>),
</span></span><span style="display:flex;"><span>    (<span style="color:#e6db74">&#34;10.2.0&#34;</span>, <span style="color:#e6db74">&#34;10.2.7&#34;</span>, <span style="color:#e6db74">&#34;10.2.7-h38&#34;</span>, <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>),
</span></span><span style="display:flex;"><span>]
</span></span><span style="display:flex;"><span><span style="color:#75715e"># &#34;lessThan&#34; bound of each affected range in the record&#39;s affected[] data</span>
</span></span><span style="display:flex;"><span>LESS_THAN <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;12.2&#34;</span>: <span style="color:#e6db74">&#34;12.2.4&#34;</span>, <span style="color:#e6db74">&#34;12.1&#34;</span>: <span style="color:#e6db74">&#34;12.1.10-h2&#34;</span>, <span style="color:#e6db74">&#34;11.2&#34;</span>: <span style="color:#e6db74">&#34;11.2.16&#34;</span>,
</span></span><span style="display:flex;"><span>             <span style="color:#e6db74">&#34;11.1&#34;</span>: <span style="color:#e6db74">&#34;11.1.16-h3&#34;</span>, <span style="color:#e6db74">&#34;10.2&#34;</span>: <span style="color:#e6db74">&#34;10.2.18-h12&#34;</span>}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">key</span>(v):
</span></span><span style="display:flex;"><span>    m <span style="color:#f92672">=</span> re<span style="color:#f92672">.</span>fullmatch(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;(\d+)\.(\d+)\.(\d+)(?:-h(\d+))?&#34;</span>, v<span style="color:#f92672">.</span>strip())
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> m:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">raise</span> <span style="color:#a6e22e">ValueError</span>(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;unrecognised version: </span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    a, b, c, h <span style="color:#f92672">=</span> m<span style="color:#f92672">.</span>groups()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> (int(a), int(b), int(c), int(h <span style="color:#f92672">or</span> <span style="color:#ae81ff">0</span>))
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">check</span>(v):
</span></span><span style="display:flex;"><span>    k <span style="color:#f92672">=</span> key(v)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> start, last, fix, newest <span style="color:#f92672">in</span> ROWS:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> key(start) <span style="color:#f92672">&lt;=</span> k <span style="color:#f92672">and</span> k[:<span style="color:#ae81ff">3</span>] <span style="color:#f92672">&lt;=</span> key(last)[:<span style="color:#ae81ff">3</span>]:
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">if</span> k <span style="color:#f92672">&gt;=</span> key(fix):
</span></span><span style="display:flex;"><span>                <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: fixed (in-line fix </span><span style="color:#e6db74">{</span>fix<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: AFFECTED -&gt; upgrade to </span><span style="color:#e6db74">{</span>fix<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span> <span style="color:#f92672">+</span> (<span style="color:#e6db74">&#34;&#34;</span> <span style="color:#66d9ef">if</span> fix <span style="color:#f92672">==</span> newest <span style="color:#66d9ef">else</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34; or </span><span style="color:#e6db74">{</span>newest<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>    branch <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;.&#34;</span><span style="color:#f92672">.</span>join(map(str, k[:<span style="color:#ae81ff">2</span>]))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> branch <span style="color:#f92672">in</span> LESS_THAN:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> k <span style="color:#f92672">&gt;=</span> key(LESS_THAN[branch]):
</span></span><span style="display:flex;"><span>            <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: not affected (at or above </span><span style="color:#e6db74">{</span>LESS_THAN[branch]<span style="color:#e6db74">}</span><span style="color:#e6db74">)&#34;</span>
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: AFFECTED per the record, no row in the fix table -&gt; </span><span style="color:#e6db74">{</span>LESS_THAN[branch]<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>v<span style="color:#e6db74">}</span><span style="color:#e6db74">: branch </span><span style="color:#e6db74">{</span>branch<span style="color:#e6db74">}</span><span style="color:#e6db74"> not listed in the CVE record (check the advisory)&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]:
</span></span><span style="display:flex;"><span>    print(check(v))
</span></span></code></pre></div><p>Sample output from our run:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>$ python3 check_0322.py 11.1.6-h20 11.2.15 10.2.18-h12 12.1.1 10.1.14
</span></span><span style="display:flex;"><span>11.1.6-h20: AFFECTED -&gt; upgrade to 11.1.6-h39 or 11.1.16-h3
</span></span><span style="display:flex;"><span>11.2.15: AFFECTED -&gt; upgrade to 11.2.16
</span></span><span style="display:flex;"><span>10.2.18-h12: fixed (in-line fix 10.2.18-h12)
</span></span><span style="display:flex;"><span>12.1.1: AFFECTED per the record, no row in the fix table -&gt; 12.1.10-h2
</span></span><span style="display:flex;"><span>10.1.14: branch 10.1 not listed in the CVE record (check the advisory)
</span></span></code></pre></div><p>To get the input, read the running software version from each firewall and Panorama: on the CLI, <code>show system info</code> is the usual place to look (illustrative; we did not run it against a device). Feed those values to the script, and remember to include Panorama itself.</p>
<h2 id="what-the-vulnerability-does">What the vulnerability does</h2>
<p>The record classifies the flaw as CWE-94, improper control of generation of code (&ldquo;code injection&rdquo;). Its title is &ldquo;Authenticated Remote Code Execution (RCE) in the Management Interface&rdquo;, and the impact is code execution &ldquo;with root privileges on the firewall&rdquo;. Palo Alto Networks found it internally; the record credits an internal reporter and gives no technical detail of the vulnerable code path.</p>
<p>Palo Alto Networks publishes two CVSS v4.0 scores for two deployment scenarios:</p>
<table>
  <thead>
      <tr>
          <th>Scenario</th>
          <th>Score</th>
          <th>Attack vector</th>
          <th>Privileges required</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Management interface reachable from external IP addresses on the internet</td>
          <td>8.6 High</td>
          <td>Network</td>
          <td>High</td>
      </tr>
      <tr>
          <td>Access restricted to a jump box that is the only system allowed to reach the interface</td>
          <td>8.5 High</td>
          <td>Adjacent</td>
          <td>High</td>
      </tr>
  </tbody>
</table>
<p>Both vectors carry <code>E:A</code> (attacked), <code>AU:N</code> (not automatable) and <code>U:Red</code>. If you want to read the vector strings metric by metric, our <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">guide to CVSS v4.0 vectors</a> walks through each field.</p>
<p>The small gap between the two scores understates the operational difference. Restricting the interface to a jump box does not fix the bug, but it shrinks the set of places an attacker can attempt it from to one host you control and monitor.</p>
<h3 id="authenticated-does-not-mean-safe-to-wait">&ldquo;Authenticated&rdquo; does not mean &ldquo;safe to wait&rdquo;</h3>
<p>The CVSS vectors rate privileges required as High, so an attacker needs a valid management login. That is a real barrier, but not a high one when the interface is reachable. Speaking to CSO Online, IDC analyst Yih Khai Wong pointed to &ldquo;weak management-access hygiene&rdquo;: administrator credentials stolen through phishing or reuse, internet-reachable management interfaces, or an earlier compromise that left the attacker with valid logins.</p>
<p>PAN-OS management and edge services have a history here. CISA&rsquo;s KEV catalog lists 16 Palo Alto Networks entries, including two management-interface flaws added on the same day in November 2024: CVE-2024-0012, an authentication bypass, and CVE-2024-9474, an OS command injection. In 2026 alone, CVE-2026-0322 is the third PAN-OS entry, after CVE-2026-0300 (an unauthenticated buffer overflow in the User-ID Authentication Portal, added May 6) and CVE-2026-0257 (a GlobalProtect authentication bypass, added May 29, which CISA marks as known to be used in ransomware campaigns). None of those records links them to CVE-2026-0322. The point is simpler: if your management access depends on nobody having a valid password, it is weaker than it looks.</p>
<h2 id="exploitation-status">Exploitation status</h2>
<p>What the primary records say:</p>
<ul>
<li><strong>Palo Alto Networks:</strong> &ldquo;Very limited exploitation has been observed targeting the Palo Alto Networks management interface.&rdquo; The record&rsquo;s initial publication time is October 14, 2026, 16:00 UTC.</li>
<li><strong>CISA KEV:</strong> added October 14, 2026; due date October 17; known ransomware campaign use &ldquo;Unknown&rdquo;; forensic triage flag &ldquo;Yes&rdquo;. The required action points to <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04</a> and its forensic triage requirements.</li>
<li><strong>CISA SSVC (in the CVE record):</strong> Exploitation &ldquo;active&rdquo;, Automatable &ldquo;no&rdquo;, Technical Impact &ldquo;total&rdquo;. Our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">explainer on Vulnrichment SSVC values</a> covers how to turn those into a patch decision.</li>
</ul>
<p>What is reported but we could not confirm at the source: BleepingComputer reported that Palo Alto Networks shared three IP addresses linked to the activity, <code>66.42.72[.]14</code>, <code>208.83.239[.]114</code> and <code>149.248.47[.]234</code>. The Hacker News lists the same three as &ldquo;assessed to be linked to the exploit attempts&rdquo;. We did not read them on the advisory page itself, so treat them as reported indicators, and check the advisory for the current list. Neither outlet describes what attackers did after exploitation, and Palo Alto Networks has not said who is behind it or when it began.</p>
<h2 id="what-to-do-now">What to do now</h2>
<ol>
<li><strong>Upgrade</strong> to the fixed release for your maintenance range, firewalls and Panorama both. There is no workaround.</li>
<li><strong>Preserve evidence before you upgrade</strong> if the management interface was reachable from anywhere you do not fully control. Export system, configuration and authentication logs, and take a tech support file, so the upgrade does not become the only record of the device&rsquo;s state.</li>
<li><strong>Restrict who can reach the management interface.</strong> Palo Alto Networks&rsquo; administrative access guidance says: &ldquo;Do not enable access to your management interface from the internet or from other untrusted zones&rdquo;, &ldquo;Isolate the management interface on a dedicated management VLAN&rdquo; and &ldquo;Restrict the IP addresses that are permitted to access the management interface&rdquo;. On the firewall this is the Permitted IP Addresses list on the management interface (Device &gt; Setup &gt; Interfaces &gt; Management in the PAN-OS admin guide). An empty list allows access from any IP address.</li>
<li><strong>Check dataplane interfaces too.</strong> If you manage the firewall through an interface management profile on a data port, restrict it the same way. The guidance is explicit that a profile allowing HTTP, HTTPS, Telnet or SSH must not be attached to an interface where a GlobalProtect portal or gateway is configured, because that exposes management to the internet.</li>
<li><strong>Tighten administrator accounts.</strong> One account per person, the default <code>admin</code> password changed, and MFA through external authentication for administrators with change privileges. Rotate credentials for any administrator who logged in from outside the management network recently.</li>
<li><strong>Use the Threat IDs only as a stopgap.</strong> Threat IDs 510046 and 510047 (Applications and Threats content 9159-10287 and later) offer what Palo Alto Networks calls limited coverage. They only work if management traffic is routed through a dataplane port with a management profile, its certificate is replaced, inbound management traffic is decrypted, and threat prevention is applied to it.</li>
</ol>
<h2 id="what-to-review-after-patching">What to review after patching</h2>
<p>Upgrading closes the hole; it does not tell you whether someone used it. Root code execution on a firewall means an attacker could have changed configuration, added accounts, or used the device as a foothold. Scope the review from the earliest date you think the interface was exposed, and at least from early October 2026.</p>
<table>
  <thead>
      <tr>
          <th>Question</th>
          <th>Where to look</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Who logged in to management, from where?</td>
          <td>System and authentication logs: administrator logins from source addresses outside your management network or jump hosts, and logins at unusual times</td>
      </tr>
      <tr>
          <td>Did anyone connect from the reported addresses?</td>
          <td>Firewall traffic logs and any upstream logs for <code>66.42.72.14</code>, <code>208.83.239.114</code> and <code>149.248.47.234</code></td>
      </tr>
      <tr>
          <td>Were admin accounts added or changed?</td>
          <td>Configuration logs and the current administrator list against your expected set</td>
      </tr>
      <tr>
          <td>What changed in the configuration?</td>
          <td>Configuration logs and commit history: new NAT or security rules, new management profiles, changed authentication profiles, new certificates</td>
      </tr>
      <tr>
          <td>Is the device talking somewhere new?</td>
          <td>Outbound connections from the management address and the firewall itself to destinations it does not normally reach</td>
      </tr>
  </tbody>
</table>
<p>As illustrative starting points in the PAN-OS CLI, <code>show admins all</code> lists administrators and their sessions, and <code>show log system</code> and <code>show log config</code> read the system and configuration logs. Check the CLI reference for your release for filters and exact syntax; we did not run these against a device.</p>
<p>If you find activity you cannot explain, treat the device as compromised rather than just patched: rebuild from a known-good configuration, rotate every credential and key the firewall holds (administrator passwords, API keys, certificates, VPN and directory service secrets), and involve Palo Alto Networks support. Our <a href="/posts/logging-for-incidents-not-for-dashboards/">incident logging guide</a> covers what to keep while you do.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We did not test the vulnerability or run the CLI commands on a PAN-OS device. The fixed-release table and checker come from the CVE record as published on October 14, 2026; Palo Alto Networks can update the advisory, so check it before you finalise a change. The three IP addresses come from press reports, not from our reading of the advisory. Palo Alto Networks has not published technical details of the flaw or of the exploitation it observed.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Authenticated RCE as root via the management interface; PA-Series, VM-Series and Panorama affected; Cloud NGFW and Prisma Access not</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/0xxx/CVE-2026-0322.json">CVE-2026-0322 record (Palo Alto Networks CNA)</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Fixed-release table, no workaround, Threat IDs 510046/510047 and their conditions, no special configuration needed</td>
          <td>Same record, solutions, workarounds and configurations</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CWE-94; CVSS v4.0 8.6 and 8.5 with the vectors shown; internal discovery</td>
          <td>Same record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>&ldquo;Very limited exploitation has been observed&rdquo;</td>
          <td>Same record, exploits</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV added 2026-10-14, due 2026-10-17, ransomware Unknown, forensic triage Yes</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV catalog data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC active / no / total</td>
          <td>CISA ADP container in the CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>16 Palo Alto Networks KEV entries; CVE-2024-0012, CVE-2024-9474, CVE-2026-0300, CVE-2026-0257 details</td>
          <td>CISA KEV catalog data; CVE records for CVE-2026-0300 and CVE-2026-0257</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Three IP addresses linked to the activity</td>
          <td><a href="https://bleepingcomputer.com/news/security/palo-alto-networks-warns-of-pan-os-rce-flaw-exploited-in-attacks">BleepingComputer</a>, <a href="https://thehackernews.com/2026/10/palo-alto-networks-patches-pan-os.html">The Hacker News</a></td>
          <td>Qualified: press reports</td>
      </tr>
      <tr>
          <td>Analyst comment on credential hygiene</td>
          <td><a href="https://www.csoonline.com/article/4235787/attackers-exploit-palo-alto-pan-os-flaw-for-root-access-as-cisa-orders-three-day-fix.html">CSO Online</a></td>
          <td>Verified (attributed)</td>
      </tr>
      <tr>
          <td>Management access guidance quoted</td>
          <td><a href="https://docs.paloaltonetworks.com/pan-os/9-1/pan-os-admin/getting-started/best-practices-for-securing-administrative-access">PAN-OS admin guide: securing administrative access</a>; <a href="https://docs.paloaltonetworks.com/ngfw/administration/set-up-firewalls/device-setup/configure-the-management-interface-settings">management interface settings</a></td>
          <td>Verified; the first page is from the 9.1 guide</td>
      </tr>
      <tr>
          <td>Version checker output</td>
          <td>Our run of the script above</td>
          <td>Verified (computed)</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>VeloCloud Orchestrator CVE-2026-114625: Patch and Hunt</title><link>https://firewallsync.com/posts/velocloud-orchestrator-cve-2026-114625-patch-and-hunt/</link><guid>https://firewallsync.com/posts/velocloud-orchestrator-cve-2026-114625-patch-and-hunt/</guid><pubDate>Sun, 25 Oct 2026 04:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CVE-2026-114625 is a SQL injection in on-premises Arista VeloCloud Orchestrator that lets any authenticated user, down to the Enterprise Read Only role, act as the built-in superuser operator. CISA added it to KEV on October 23, 2026. It is the third exploited VCO flaw since July, and builds that fixed the first two are still vulnerable.</description><enclosure url="https://firewallsync.com/posts/velocloud-orchestrator-cve-2026-114625-patch-and-hunt/cover_hu1969827314249195703.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/velocloud-orchestrator-cve-2026-114625-patch-and-hunt/cover_hu1969827314249195703.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/velocloud-orchestrator-cve-2026-114625-patch-and-hunt/cover_hu1969827314249195703.jpg" alt="Dozens of labelled blue network cables fanning out from rows of switch ports lit by small yellow-green status lights in a dark rack" width="1200" height="630"></p><p>CVE-2026-114625 is a SQL injection in on-premises Arista VeloCloud Orchestrator (VCO) that lets any authenticated portal user gain operator privileges and act as the local &ldquo;superuser operator&rdquo;. The lowest built-in tenant role, Enterprise Read Only, is enough. CISA added it to the Known Exploited Vulnerabilities (KEV) catalog on October 23, 2026. Upgrade to VCO 5.2.3.18, 6.1.3.10 or 6.4.2.12 or later, restrict the web interface to administrative networks until you can, and then check the orchestrator for signs that someone used it before you did.</p>
<p>It is also the third exploited VCO vulnerability in three months. If you patched in July or September and stopped there, you are still exposed.</p>
<h2 id="are-you-affected">Are you affected?</h2>
<p>You are affected if you run VCO on-premises on one of these versions, as listed in Arista&rsquo;s CVE record:</p>
<table>
  <thead>
      <tr>
          <th>Release train</th>
          <th>Affected</th>
          <th>Fixed in</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>5.2</td>
          <td>5.2.0 up to, not including, 5.2.3.18</td>
          <td>5.2.3.18 and later</td>
      </tr>
      <tr>
          <td>6.1</td>
          <td>6.1.0 up to, not including, 6.1.3.10</td>
          <td>6.1.3.10 and later</td>
      </tr>
      <tr>
          <td>6.4</td>
          <td>6.4.0 up to, not including, 6.4.2.12</td>
          <td>6.4.2.12 and later</td>
      </tr>
  </tbody>
</table>
<p>Arista says hosted VCO, including Dedicated, was affected and has already been patched. The record lists no 7.0 range for this CVE and marks unlisted versions as unaffected, but Arista&rsquo;s September advisory did list 7.0 builds as affected by an earlier flaw. If you run 7.0, confirm your status with Arista rather than relying on the absence of a row. With the July flaw, Arista said it had not assessed end-of-support releases, so if you are on an older train, ask Arista&rsquo;s Technical Assistance Center (TAC) about upgrade paths.</p>
<p>VCO is the management plane for VeloCloud SD-WAN. It holds configuration, device inventory and credentials for the Edges it manages, so treat it as a tier-0 system even if it doesn&rsquo;t look like one in your asset inventory.</p>
<h2 id="what-the-vulnerability-does">What the vulnerability does</h2>
<p>Arista&rsquo;s description in the CVE record: on an affected system, &ldquo;any authenticated user can gain operator privileges and masquerade as a local &lsquo;superuser operator&rsquo; user&rdquo;. That includes functionality meant for internal use only and not meant to be reachable remotely. The record classifies it as CWE-89 (SQL injection).</p>
<p>The conditions, per the record&rsquo;s configuration notes:</p>
<ul>
<li>VCO is exposed by default, and &ldquo;there is no configuration that can prevent the exposure&rdquo;.</li>
<li>The attacker needs network access to the VCO web interface.</li>
<li>The attacker needs a valid authenticated session on the VCO portal. The minimum role is <strong>Enterprise Read Only</strong>.</li>
</ul>
<p>That last condition matters more than it first appears. In a multi-tenant VCO, a read-only user at one customer or one business unit could reach operator-level control over the whole orchestrator. Every tenant login, including service accounts, shared logins and accounts left behind by former staff or partners, is a possible starting point.</p>
<p>Arista scores it CVSS 3.1 <strong>9.9</strong> (<code>AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H</code>) and CVSS 4.0 <strong>9.4</strong>. The <code>PR:L</code> (low privileges required) is the only thing keeping it off 10.0. The record says the issue was discovered externally. Arista published Security Advisory 0199 and the CVE record on October 22, 2026.</p>
<h2 id="exploitation-status">Exploitation status</h2>
<p>The evidence of exploitation comes from CISA rather than from Arista&rsquo;s CVE text:</p>
<ul>
<li><strong>KEV:</strong> CISA added CVE-2026-114625 on October 23, 2026, with a due date of <strong>October 26, 2026</strong> for US federal civilian agencies. The entry has the forensic triage flag set, so under <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04</a> agencies must also check for compromise, not just patch. Known ransomware use is listed as &ldquo;Unknown&rdquo;.</li>
<li><strong>SSVC:</strong> CISA&rsquo;s Vulnrichment entry in the CVE record, timestamped October 22, rates Exploitation <strong>active</strong>, Automatable <strong>no</strong> and Technical Impact <strong>total</strong>. If you use the <a href="/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/">SSVC deployer model</a>, &ldquo;active&rdquo; plus &ldquo;total&rdquo; puts an internet-facing VCO at the top of the queue even though the flaw needs a login.</li>
</ul>
<p>Arista&rsquo;s records for the July and September VCO flaws said explicitly that they were &ldquo;known to be actively exploited&rdquo;. The October record does not, and at the time of writing we found no public detail on who is exploiting this one or how. Treat attribution and scale as unknown.</p>
<h2 id="third-exploited-vco-flaw-since-july">Third exploited VCO flaw since July</h2>
<p>CISA&rsquo;s KEV catalog now lists three VeloCloud Orchestrator vulnerabilities, all added in 2026:</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Arista advisory</th>
          <th>Added to KEV</th>
          <th>Type (CWE)</th>
          <th>Credentials needed</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-16812</td>
          <td>0144, July 27</td>
          <td>July 27</td>
          <td>OS command injection (CWE-78)</td>
          <td>None</td>
      </tr>
      <tr>
          <td>CVE-2026-93952</td>
          <td>0183, September 22</td>
          <td>September 22</td>
          <td>Improper input validation (CWE-20)</td>
          <td>None, but needs certificate-based Edge authentication and the public part of an Edge certificate</td>
      </tr>
      <tr>
          <td>CVE-2026-114625</td>
          <td>0199, October 22</td>
          <td>October 23</td>
          <td>SQL injection (CWE-89)</td>
          <td>Any portal user (Enterprise Read Only or above)</td>
      </tr>
  </tbody>
</table>
<p>Each fix moved the floor. We checked a set of common builds against the affected ranges in all three CVE records:</p>
<table>
  <thead>
      <tr>
          <th>Build</th>
          <th>Still affected by</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>5.2.3.14 (July fix)</td>
          <td>CVE-2026-93952, CVE-2026-114625</td>
      </tr>
      <tr>
          <td>5.2.3.16 (September fix)</td>
          <td>CVE-2026-114625</td>
      </tr>
      <tr>
          <td>5.2.3.18</td>
          <td>None of the three</td>
      </tr>
      <tr>
          <td>6.1.3.4 (July fix)</td>
          <td>CVE-2026-93952, CVE-2026-114625</td>
      </tr>
      <tr>
          <td>6.1.3.10</td>
          <td>None of the three</td>
      </tr>
      <tr>
          <td>6.4.2.4 (July fix)</td>
          <td>CVE-2026-93952, CVE-2026-114625</td>
      </tr>
      <tr>
          <td>6.4.2.8 (September fix)</td>
          <td>CVE-2026-114625</td>
      </tr>
      <tr>
          <td>6.4.2.12</td>
          <td>None of the three</td>
      </tr>
      <tr>
          <td>7.0.0.1 (July fix)</td>
          <td>CVE-2026-93952</td>
      </tr>
  </tbody>
</table>
<p>That table was generated by comparing each build to the <code>affected</code> version ranges in the three CVE JSON records from the CVEProject/cvelistV5 repository. It reflects what the records say, not testing against a VCO. Arista&rsquo;s September advisory did not list a fixed 6.1 or 7.0 build for CVE-2026-93952 when it was published, so for those trains check Arista&rsquo;s current advisory before assuming you&rsquo;re clear.</p>
<p>The practical point: an orchestrator that was patched once and then left alone has been inside at least one exploited vulnerability window for most of the last three months. That changes the hunting question from &ldquo;was I hit last week?&rdquo; to &ldquo;has this box been clean since July?&rdquo;</p>
<h2 id="what-to-do-now">What to do now</h2>
<ol>
<li><strong>Find every VCO you run.</strong> Include lab, DR and acquired-company instances. Record version, train and whether the web interface is reachable from the internet.</li>
<li><strong>Restrict the web interface</strong> to trusted administrative networks now, before the upgrade window. This is Arista&rsquo;s first interim control. It does not remove the risk from users who already have access.</li>
<li><strong>Upgrade</strong> to 5.2.3.18, 6.1.3.10 or 6.4.2.12 or later. If you are on 7.0 or an unsupported train, open a case with Arista TAC.</li>
<li><strong>Preserve evidence first</strong> if anything below looks wrong. Arista&rsquo;s September advisory asks for web access logs, application and system logs, database logs and file-system timestamps to be kept before remediation. An upgrade can overwrite exactly the files you need.</li>
<li><strong>Cut back portal accounts.</strong> Remove dormant tenant users, rotate shared credentials, and check that partner and MSP accounts are still needed. This flaw turns every one of them into a path to operator rights.</li>
<li><strong>Block outbound traffic</strong> from the VCO host that normal operation doesn&rsquo;t need. Arista lists this in the October workarounds.</li>
</ol>
<p>For the wider prioritization question, our guide to <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">matching the KEV feed against a scanner export</a> shows how to catch the next one automatically.</p>
<h2 id="how-to-hunt">How to hunt</h2>
<p>Arista&rsquo;s October record lists the same detection areas as its earlier VCO advisories: access from known malicious IPs, unexpected outbound network activity from the VCO host, backdoor daemons and web shells, and unexpected administrator activity. It does not publish indicators specific to CVE-2026-114625.</p>
<p>Arista did publish concrete indicators with the September flaw, CVE-2026-93952 (Security Advisory 0183). Arista has not said they are linked to the October issue. Because most builds vulnerable today were also exposed to CVE-2026-93952, they are a sensible baseline:</p>
<table>
  <thead>
      <tr>
          <th>Indicator (from advisory 0183)</th>
          <th>Where to look</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>/usr/local/sbin/.vcnode.js</code></td>
          <td>File system</td>
      </tr>
      <tr>
          <td><code>/usr/local/sbin/vc-sysmond</code>, known malicious MD5 <code>dc78e206eaeadec59fc5801fe4556bd0</code></td>
          <td>File system</td>
      </tr>
      <tr>
          <td><code>/etc/systemd/system/vc-sysmon.service</code></td>
          <td>systemd units</td>
      </tr>
      <tr>
          <td><code>x-vc-opt</code> HTTP header</td>
          <td>nginx access logs</td>
      </tr>
      <tr>
          <td>142.93.149.77, 104.248.126.159</td>
          <td>Web and firewall logs</td>
      </tr>
  </tbody>
</table>
<p>A starting point on the VCO host, run as root. This is illustrative and was not tested against a VCO; adjust log paths to your build:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span><span style="color:#75715e"># Persistence files named in Arista advisory 0183</span>
</span></span><span style="display:flex;"><span>ls -la /usr/local/sbin/.vcnode.js /usr/local/sbin/vc-sysmond <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>       /etc/systemd/system/vc-sysmon.service 2&gt;/dev/null
</span></span><span style="display:flex;"><span>md5sum /usr/local/sbin/vc-sysmond 2&gt;/dev/null
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Header and source IPs from the same advisory in nginx logs</span>
</span></span><span style="display:flex;"><span>zgrep -il <span style="color:#e6db74">&#39;x-vc-opt&#39;</span> /var/log/nginx/*access* 2&gt;/dev/null
</span></span><span style="display:flex;"><span>zgrep -hE <span style="color:#e6db74">&#39;142\.93\.149\.77|104\.248\.126\.159&#39;</span> /var/log/nginx/*access* 2&gt;/dev/null | head
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Unexpected outbound connections from the orchestrator</span>
</span></span><span style="display:flex;"><span>ss -tnp state established
</span></span></code></pre></div><p>Beyond file indicators, Arista&rsquo;s advisories point to behavior worth reviewing in the VCO&rsquo;s own audit trail:</p>
<ul>
<li>operator-level actions taken by accounts that only hold tenant roles, or by the local superuser operator at times nobody on your team was working</li>
<li>sensitive configuration changes or privileged maintenance actions with no matching change ticket</li>
<li>database exports, new archive files, or unexpected access to device inventory, certificates and keys</li>
<li>configuration pushed to Edges that you did not approve</li>
</ul>
<p>If you find a compromise, Arista&rsquo;s guidance from the earlier advisories applies: rotate credentials, validate the state of every managed Edge, and restore or replace the orchestrator from a trusted source. Patching a compromised VCO does not remove an attacker who is already there.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We could not retrieve Arista&rsquo;s Security Advisory 0199 page directly. The affected versions, conditions, workarounds and description above come from the CVE record Arista published as the CNA, which carries the advisory&rsquo;s content. We have not tested any VCO build, and the hunting commands are a starting point, not a validated detection. There is no public technical write-up of the exploit at the time of writing.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Status</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Description, &ldquo;superuser operator&rdquo;, Enterprise Read Only minimum role, exposed by default, affected and fixed versions, hosted and Dedicated patched, CVSS 9.9 / 9.4, CWE-89, external discovery, workarounds</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/114xxx/CVE-2026-114625.json">CVE-2026-114625 record (Arista as CNA)</a>; <a href="https://www.arista.com/en/support/advisories-notices/security-advisory/24838-security-advisory-0199">Arista Security Advisory 0199</a></td>
          <td>Verified (CVE record)</td>
      </tr>
      <tr>
          <td>KEV addition October 23, due October 26, forensic triage flag, ransomware use Unknown</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a> (JSON feed, catalog version 2026.10.23)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC: Exploitation active, Automatable no, Technical Impact total</td>
          <td>CISA ADP Vulnrichment container in the CVE-2026-114625 record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-16812: July 27, OS command injection, no credentials needed, fixed builds</td>
          <td><a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/16xxx/CVE-2026-16812.json">CVE-2026-16812 record</a>; <a href="https://bleepingcomputer.com/news/security/arista-patches-velocloud-orchestrator-zero-day-exploited-in-attacks">BleepingComputer, July 27, 2026</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-93952: conditions, affected and fixed builds, indicators of compromise, evidence preservation and post-compromise guidance</td>
          <td><a href="https://www.arista.com/en/support/advisories-notices/security-advisory/24765-security-advisory-0183">Arista Security Advisory 0183</a>; <a href="https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/93xxx/CVE-2026-93952.json">CVE-2026-93952 record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Build-by-build &ldquo;still affected by&rdquo; table</td>
          <td>Computed from the three CVE records&rsquo; version ranges</td>
          <td>Verified (computed, not tested on VCO)</td>
      </tr>
      <tr>
          <td>Status of the 7.0 train for CVE-2026-114625</td>
          <td>Not listed in the CVE record</td>
          <td>Qualified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Collabora CODE CVE-2026-110447: Patch Nextcloud, Then Hunt</title><link>https://firewallsync.com/posts/collabora-code-cve-2026-110447-nextcloud-patch-and-hunt/</link><guid>https://firewallsync.com/posts/collabora-code-cve-2026-110447-nextcloud-patch-and-hunt/</guid><pubDate>Sat, 24 Oct 2026 04:30:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>CVE-2026-110447 is an unauthenticated OS command injection in proxy.php of the Collabora Online Built-in CODE Server app for Nextcloud (richdocumentscode) before 26.4.303. It was exploited from September, a month before the CVE was published, and CISA added it to KEV on October 23. Update or remove the app, then hunt for web shells.</description><enclosure url="https://firewallsync.com/posts/collabora-code-cve-2026-110447-nextcloud-patch-and-hunt/cover_hu8653931782960325198.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/collabora-code-cve-2026-110447-nextcloud-patch-and-hunt/cover_hu8653931782960325198.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/collabora-code-cve-2026-110447-nextcloud-patch-and-hunt/cover_hu8653931782960325198.jpg" alt="A panelled door standing ajar in a pitch-black room, warm light spilling through the gap and across the floor" width="1200" height="630"></p><p>If your Nextcloud server runs the <strong>Collabora Online - Built-in CODE Server</strong> app (<code>richdocumentscode</code>) at any version before 26.4.303, treat it as exposed to unauthenticated remote code execution. Update the app to the current release or remove it today, then hunt for web shells: the fix shipped on September 11, 2026, attacks were reported from September 10, and the CVE was not published until October 14. CISA added CVE-2026-110447 to its Known Exploited Vulnerabilities (KEV) catalog on October 23, with a due date of October 26 for US federal civilian agencies.</p>
<h2 id="are-you-affected">Are you affected?</h2>
<p>The CVE record is specific: the flaw is &ldquo;not present in standalone Collabora Online servers; only this deployment with CODE Server is affected.&rdquo;</p>
<table>
  <thead>
      <tr>
          <th>Deployment</th>
          <th>Affected?</th>
          <th>Basis</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>richdocumentscode</code> app before 26.4.303</td>
          <td>Yes</td>
          <td>CVE record</td>
      </tr>
      <tr>
          <td><code>richdocumentscode</code> 26.4.303 or later</td>
          <td>No</td>
          <td>CVE record (fixed version)</td>
      </tr>
      <tr>
          <td>Nextcloud Office (<code>richdocuments</code>) pointed at a separately run Collabora Online server (Docker, DEB/RPM, appliance)</td>
          <td>No</td>
          <td>CVE record; Collabora advisory as reported</td>
      </tr>
      <tr>
          <td>Nextcloud All-in-One</td>
          <td>No</td>
          <td>Collabora, as reported by BleepingComputer</td>
      </tr>
      <tr>
          <td>Nextcloud 29 or older with the app installed</td>
          <td>Upgrade Nextcloud or remove the app; 26.4.303 needs Nextcloud 30 or later</td>
          <td>App metadata (<code>min-version=&quot;30&quot;</code>); advisory as reported</td>
      </tr>
  </tbody>
</table>
<p>The two app names are easy to confuse. <code>richdocuments</code> is the Nextcloud Office connector. <code>richdocumentscode</code> is the optional app that bundles a CODE (Collabora Online Development Edition) AppImage and starts it on the Nextcloud host. Only the second is in scope.</p>
<p>Check which you have from the Nextcloud directory:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span><span style="color:#75715e"># Run as the web server user; adjust the user and path for your install</span>
</span></span><span style="display:flex;"><span>sudo -u www-data php occ app:list | grep -E <span style="color:#e6db74">&#39;richdocuments(code)?:&#39;</span>
</span></span></code></pre></div><h2 id="the-vulnerability">The vulnerability</h2>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Value</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE</td>
          <td>CVE-2026-110447</td>
      </tr>
      <tr>
          <td>Product</td>
          <td>Collabora Online - Built-in CODE Server (<code>richdocumentscode</code>), a Nextcloud app</td>
      </tr>
      <tr>
          <td>Affected</td>
          <td>All versions before 26.4.303</td>
      </tr>
      <tr>
          <td>Weakness</td>
          <td>CWE-78, OS command injection, in <code>proxy.php</code></td>
      </tr>
      <tr>
          <td>Condition</td>
          <td>The CODE server is not running, &ldquo;e.g., the product was recently restarted&rdquo;</td>
      </tr>
      <tr>
          <td>CVSS 4.0</td>
          <td>9.2 Critical, <code>CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N</code></td>
      </tr>
      <tr>
          <td>CISA SSVC</td>
          <td>Exploitation: active; Automatable: yes; Technical Impact: total</td>
      </tr>
      <tr>
          <td>KEV</td>
          <td>Added October 23, 2026; due October 26, 2026; forensic triage: yes</td>
      </tr>
  </tbody>
</table>
<p>The <code>AT:P</code> (attack requirements present) in the vector reflects that precondition: the bundled server has to be down. That is less of a barrier than it sounds. Restarts, app upgrades and crashes all stop it, and in the vulnerable code an unauthenticated <code>?status</code> request to <code>proxy.php</code> is what asks for the server to be started when it is not running. If you are new to v4.0 vectors, our guide to <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">reading a CVSS v4.0 vector string</a> walks through each metric.</p>
<p>The CVE record also says this is &ldquo;a different vulnerability than CVE-2025-66208&rdquo;, so a host patched for that earlier issue is not covered.</p>
<h3 id="what-the-fix-changed">What the fix changed</h3>
<p>The CVE record describes the bug only as command injection in <code>proxy.php</code>. The app&rsquo;s public repository shows more. This section is our reading of that source, not a vendor statement.</p>
<p>In 26.4.302, <code>startCoolwsd()</code> built the launch command as one string. Two values derived from the request, the HTTP <code>Host</code> header and the request URI (used to build a font-configuration URL), were each passed through PHP&rsquo;s <code>escapeshellarg()</code>, which wraps them in single quotes. The whole command was then placed inside a double-quoted <code>bash -c &quot;...&quot;</code> string and handed to the shell. Single quotes have no special meaning inside double quotes, so shell command substitution in those values was still evaluated by the outer shell.</p>
<p>The 26.4.303 change, committed on September 9, wraps the entire inner command in <code>escapeshellarg()</code> before passing it to <code>bash -c</code>. The current code goes further: it passes an argument array to <code>proc_open()</code> and runs a fixed script that receives the options as <code>&quot;$@&quot;</code>, so no request value is ever parsed by a shell.</p>
<p>Two practical consequences follow. First, the attacker-controlled input sits in ordinary request fields, so web server access logs and any proxy logs that record the <code>Host</code> header are your best evidence. Second, the vulnerable <code>proxy.php</code> in 26.4.302 includes no Nextcloud bootstrap at all, which fits Collabora&rsquo;s warning that disabling the app does not help: if the web server serves the file, it runs.</p>
<h2 id="timeline">Timeline</h2>
<table>
  <thead>
      <tr>
          <th>Date (2026)</th>
          <th>Event</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>September 9</td>
          <td>Fix committed to <code>proxy.php</code></td>
          <td>richdocumentscode repository</td>
      </tr>
      <tr>
          <td>September 10</td>
          <td>Earliest attacks reported by administrators</td>
          <td>BleepingComputer, eSecurityPlanet</td>
      </tr>
      <tr>
          <td>September 11</td>
          <td>26.4.303 released; GitHub issue #354 opened by affected admins</td>
          <td>Repository tag and changelog; WZ-IT</td>
      </tr>
      <tr>
          <td>October 3</td>
          <td>Community warning on the Nextcloud forum to update immediately</td>
          <td>BleepingComputer</td>
      </tr>
      <tr>
          <td>October 14</td>
          <td>CVE published; Collabora advisory GHSA-g3vx-3m76-fghh</td>
          <td>CVE record</td>
      </tr>
      <tr>
          <td>October 23</td>
          <td>CISA adds the CVE to KEV; app 26.4.502 released</td>
          <td>KEV catalog; repository changelog</td>
      </tr>
      <tr>
          <td>October 26</td>
          <td>KEV due date for US federal civilian agencies</td>
          <td>KEV catalog</td>
      </tr>
  </tbody>
</table>
<p>A month passed between a public fix and a CVE. Collabora, according to BleepingComputer, attributed much of the delay to GitHub&rsquo;s CVE-processing backlog. Whatever the reason, anyone who waited for a CVE or a scanner signature before updating was exposed for the entire window, which is why the hunt below matters more than the patch date. Our explainer on <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">CISA BOD 26-04</a> covers what KEV&rsquo;s &ldquo;forensic triage: yes&rdquo; flag asks federal agencies to do. Private teams can borrow the same approach.</p>
<h2 id="what-to-do-now">What to do now</h2>
<ol>
<li>
<p><strong>Decide whether you need the built-in server.</strong> If you only installed it to try Nextcloud Office, remove it:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>sudo -u www-data php occ app:remove richdocumentscode
</span></span><span style="display:flex;"><span>ls -d apps/richdocumentscode 2&gt;/dev/null <span style="color:#f92672">&amp;&amp;</span> echo <span style="color:#e6db74">&#34;directory still present&#34;</span>
</span></span></code></pre></div><p>Confirm the directory is actually gone. A leftover <code>proxy.php</code> is the risk.</p>
</li>
<li>
<p><strong>Otherwise update it.</strong> The latest release at the time of writing is 26.4.502 (October 23). Anything at or above 26.4.303 contains the fix.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>sudo -u www-data php occ app:update richdocumentscode
</span></span><span style="display:flex;"><span>sudo -u www-data php occ app:list | grep <span style="color:#e6db74">&#39;richdocumentscode:&#39;</span>
</span></span></code></pre></div></li>
<li>
<p><strong>If you are on Nextcloud 29 or older</strong>, 26.4.303 will not install, because the app now requires Nextcloud 30. Remove the app or upgrade Nextcloud first. Those Nextcloud releases are end of life.</p>
</li>
<li>
<p><strong>Block the file at the edge as a stopgap</strong> if neither is possible in the next few hours. Denying requests to <code>/apps/richdocumentscode/proxy.php</code> at the reverse proxy breaks the built-in server, but it removes the unauthenticated entry point. This is a stopgap of our own, not vendor guidance, and it does nothing about a shell that is already there.</p>
</li>
</ol>
<h2 id="how-to-hunt">How to hunt</h2>
<p>Run these checks whether or not you have already patched. The indicators below come from administrators&rsquo; reports in the app&rsquo;s GitHub issue #354, as summarized by eSecurityPlanet and WZ-IT; we could not read the issue directly, so treat them as a starting point, not a complete list.</p>
<h3 id="files">Files</h3>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>cd /var/www/nextcloud   <span style="color:#75715e"># your Nextcloud root</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#75715e"># Reported web shell names, then any PHP file that is not part of the release</span>
</span></span><span style="display:flex;"><span>find apps/richdocumentscode -maxdepth <span style="color:#ae81ff">2</span> <span style="color:#ae81ff">\(</span> -name <span style="color:#e6db74">&#39;shx*.php&#39;</span> -o -name <span style="color:#e6db74">&#39;shrf4645.php&#39;</span> <span style="color:#ae81ff">\)</span>
</span></span><span style="display:flex;"><span>sudo -u www-data php occ integrity:check-app richdocumentscode
</span></span></code></pre></div><p><code>integrity:check-app</code> compares the app directory with the signed release and lists extra files as <code>EXTRA_FILE</code>. A clean result after an update is good news; a hit means you go to the response steps below. Also look for recently modified PHP files elsewhere under <code>apps/</code> and in the data directory, since an attacker with a shell can write anywhere the web server account can.</p>
<h3 id="web-server-logs">Web server logs</h3>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span><span style="color:#75715e"># Requests to proxy.php that carry shell syntax, raw or URL-encoded</span>
</span></span><span style="display:flex;"><span>zgrep -hiE <span style="color:#e6db74">&#39;richdocumentscode/proxy\.php&#39;</span> /var/log/nginx/access.log* /var/log/apache2/access.log* 2&gt;/dev/null <span style="color:#ae81ff">\
</span></span></span><span style="display:flex;"><span><span style="color:#ae81ff"></span>  | grep -iE <span style="color:#e6db74">&#39;\$\(|%24%28|\$\{IFS\}|%24%7BIFS|%2e%2e&#39;</span>
</span></span></code></pre></div><p>We ran the <code>find</code> and log-search commands above against sample files with GNU findutils and grep to confirm they match the reported patterns; adapt the paths to your install. The reported patterns are <code>$(</code> or <code>%24%28</code> (command substitution), <code>${IFS}</code> (used to avoid spaces) and <code>%2e%2e</code> (path traversal), along with bursts of <code>proxy.php?status</code> requests. Because the injected value can travel in the <code>Host</code> header, check whether your log format records it; the default combined format does not. If it does not, look at reverse proxy or load balancer logs that do. Go back to at least September 10, 2026.</p>
<h3 id="host-signals">Host signals</h3>
<ul>
<li>Outbound connections from the web server to <code>webhook.site</code> or other request-capture services, which reported payloads used to send back system information.</li>
<li>Failed <code>sudo</code> attempts by <code>www-data</code> or your equivalent web server user in the auth log.</li>
<li>New cron entries, SSH keys or processes owned by the web server user.</li>
</ul>
<p>Reported post-exploitation activity included searches for password manager files, cryptocurrency wallets and keystores. Published reporting we reviewed does not confirm data exfiltration or root access.</p>
<h2 id="if-you-find-a-web-shell">If you find a web shell</h2>
<p>The web server account can read Nextcloud&rsquo;s <code>config/config.php</code>, which holds the database password, the instance secret and any object storage, SMTP or other service credentials configured there. It can also read every file in the data directory.</p>
<ol>
<li>Preserve the suspicious files and logs before you delete anything.</li>
<li>Isolate the host, then rebuild it or restore from a backup taken before the first suspicious request. Deleting the shell does not remove anything else the attacker set up.</li>
<li>Rotate every secret in <code>config.php</code> and any credential the web server account could reach.</li>
<li>Tell users whose files were on the server, and follow your own incident and notification obligations.</li>
</ol>
<h2 id="prioritizing-this-against-everything-else">Prioritizing this against everything else</h2>
<p>This one does not need a scoring debate. It has an active-exploitation SSVC value, it is in KEV, it needs no credentials, and the vulnerable component is reachable on the same host and port as Nextcloud. If you use the three signals together, as we argue in <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>, every one of them points the same way. CISA&rsquo;s SSVC values are explained in our guide to <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">reading Vulnrichment SSVC data</a>.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We could not read Collabora&rsquo;s advisory (GHSA-g3vx-3m76-fghh), the Nextcloud forum warning or GitHub issue #354 directly. Statements attributed to them come from the CVE record, CISA&rsquo;s KEV entry and press coverage, and are labelled as such. The root-cause description is our reading of the public source diff. We have not tested exploitation, and the indicator list comes from early victim reports, so absence of these indicators is not proof that a server is clean.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Affected before 26.4.303; OS command injection in <code>proxy.php</code> when the CODE server isn&rsquo;t running; exploited in September and October 2026; distinct from CVE-2025-66208; standalone servers not affected; CWE-78; CVSS 4.0 9.2 and vector; published October 14</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-110447">CVE-2026-110447 record</a> (read via the cvelistV5 repository)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC: Exploitation active, Automatable yes, Technical Impact total</td>
          <td>CISA ADP container in the CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV addition October 23, due October 26, forensic triage flag, BOD 26-04 required action</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a> (JSON feed, catalog version 2026.10.23)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Fix committed September 9; 26.4.303 released September 11; 26.4.502 released October 23; Nextcloud 30 minimum; <code>Host</code> header and request URI in the launch command; standalone <code>proxy.php</code>; <code>?status</code> starts the server</td>
          <td><a href="https://github.com/CollaboraOnline/richdocumentscode">richdocumentscode repository</a> (tags 26.4.302, 26.4.303, 26.4.502 and CHANGELOG)</td>
          <td>Verified from source; root-cause reading is editorial</td>
      </tr>
      <tr>
          <td>&ldquo;Disabling the app is insufficient&rdquo;; Docker, package and All-in-One installs not affected; attacks from September 10; October 3 forum warning; Collabora&rsquo;s comments on CVE delay</td>
          <td><a href="https://www.bleepingcomputer.com/news/security/hackers-exploit-collabora-rce-flaw-to-backdoor-nextcloud-servers/">BleepingComputer</a></td>
          <td>Qualified: secondary report of vendor and admin statements</td>
      </tr>
      <tr>
          <td>Web shell names, log patterns, webhook.site callbacks, sudo attempts, data searched for</td>
          <td><a href="https://www.esecurityplanet.com/threats/news-collabora-nextcloud-cve-2026-110447/">eSecurityPlanet</a>; <a href="https://wz-it.com/en/blog/nextcloud-built-in-code-server-cve-2026-110447-rce-webshell/">WZ-IT</a></td>
          <td>Qualified: secondary summaries of GitHub issue #354</td>
      </tr>
      <tr>
          <td><code>occ integrity:check-app</code> reports extra files</td>
          <td>WZ-IT</td>
          <td>Qualified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Atlassian CVE-2026-21589 File Read: Patch and Hunt</title><link>https://firewallsync.com/posts/atlassian-cve-2026-21589-data-center-file-read-patch-and-hunt/</link><guid>https://firewallsync.com/posts/atlassian-cve-2026-21589-data-center-file-read-patch-and-hunt/</guid><pubDate>Fri, 23 Oct 2026 04:30:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>CVE-2026-21589 lets an unauthenticated attacker read files inside the web root of eight Atlassian Data Center products, including Jira, Confluence and Bitbucket. CISA added it to KEV on October 15. Here are the fixed versions, the interim mitigations, how to search access logs, and which secrets to rotate if a read succeeded.</description><enclosure url="https://firewallsync.com/posts/atlassian-cve-2026-21589-data-center-file-read-patch-and-hunt/cover_hu8381364802700144461.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/atlassian-cve-2026-21589-data-center-file-read-patch-and-hunt/cover_hu8381364802700144461.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/atlassian-cve-2026-21589-data-center-file-read-patch-and-hunt/cover_hu8381364802700144461.jpg" alt="Two worn white card-catalog cabinets with brass label holders and handles, several drawers pulled partly open" width="1200" height="630"></p><p>CVE-2026-21589 lets an unauthenticated attacker read files from the web root of self-managed Jira, Confluence, Bitbucket, Bamboo, Crowd, Crucible and Fisheye. If you run any of them, upgrade to a fixed release now, check your access logs back to early October, and rotate credentials stored in configuration files if a read may have succeeded. CISA added the flaw to its <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">Known Exploited Vulnerabilities catalog</a> on October 15, 2026.</p>
<p>Atlassian Cloud is not part of the job: Atlassian says it has already patched affected Cloud products.</p>
<h2 id="the-vulnerability">The vulnerability</h2>
<p>Atlassian&rsquo;s CVE record describes an arbitrary file access flaw (CWE-552, files or directories accessible to external parties) that &ldquo;allows an unauthenticated attacker to access specific files within the web application root directory.&rdquo; Atlassian scores it CVSS 4.0 <strong>9.3</strong>, vector <code>CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:H/SI:H/SA:H</code>.</p>
<p>Read that vector carefully (our <a href="/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/">CVSS v4.0 walkthrough</a> covers each metric). The vulnerable product loses confidentiality only (<code>VC:H</code>, <code>VI:N</code>, <code>VA:N</code>). The high scores for <em>subsequent</em> systems (<code>SC:H/SI:H/SA:H</code>) reflect what stolen files can unlock elsewhere: credentials in a configuration file are worth more than the file.</p>
<p>Two limits from the record matter for triage:</p>
<ul>
<li>The attacker needs the <strong>exact name and path</strong> of the file. The flaw &ldquo;does not allow attackers to enumerate or list directory contents.&rdquo;</li>
<li>Atlassian says &ldquo;in some configurations, there may be some sensitive files that make this highly severe.&rdquo; Whether a given instance is a nuisance or a breach depends on what sits in its web root.</li>
</ul>
<h3 id="root-cause-per-watchtowr">Root cause, per watchTowr</h3>
<p>watchTowr Labs diffed vulnerable and patched Jira, Confluence and Bitbucket builds and, as summarized by Rapid7, found a path traversal in Atlassian&rsquo;s web-resource handling where double-colon (<code>::</code>) sequences can act as path separators during request processing. In watchTowr&rsquo;s testing the attacker could not leave the Tomcat context but could read files throughout the web root, including the protected <code>WEB-INF/web.xml</code>.</p>
<p>The most serious example in the research is Crowd. In a Crowd deployment connected to Jira, reading <code>WEB-INF/classes/crowd.properties</code> exposed application credentials. watchTowr used them to create a user and add it to <code>jira-administrators</code>. Rapid7 notes that Crowd&rsquo;s IP allowlisting for applications can block that second step.</p>
<h2 id="timeline">Timeline</h2>
<table>
  <thead>
      <tr>
          <th>Date (2026)</th>
          <th>Event</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>October 5</td>
          <td>Atlassian publishes the advisory and fixed releases; CVE record published</td>
          <td>CVE record, Rapid7</td>
      </tr>
      <tr>
          <td>October 6</td>
          <td>watchTowr publishes its technical analysis and a checker script</td>
          <td>Rapid7, Help Net Security</td>
      </tr>
      <tr>
          <td>October 6</td>
          <td>Previdian reports exploitation attempts hitting its honeypots</td>
          <td>Help Net Security</td>
      </tr>
      <tr>
          <td>October 15</td>
          <td>CISA adds the CVE to KEV; CISA&rsquo;s SSVC entry records exploitation as active</td>
          <td>KEV catalog, CVE record (CISA ADP)</td>
      </tr>
      <tr>
          <td>October 18</td>
          <td>KEV due date for US federal civilian agencies</td>
          <td>KEV catalog</td>
      </tr>
  </tbody>
</table>
<p>Reports differ on what was published. Help Net Security says watchTowr built but did not publish an exploit and released a vulnerability checker. Rapid7 says file-read proof-of-concept scripts and a Nuclei template are publicly available. Either way, detailed root-cause analysis has been public since October 6, and you should assume working exploits exist.</p>
<p>CISA&rsquo;s SSVC values in the CVE record are Exploitation <strong>active</strong>, Automatable <strong>no</strong>, Technical Impact <strong>total</strong>. &ldquo;Automatable: no&rdquo; fits the need to know file paths per target; it is not a reason to wait. Our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment guide</a> explains how to read these fields.</p>
<h2 id="what-to-patch">What to patch</h2>
<p>Upgrade to one of these releases or later. Versions are from Atlassian&rsquo;s CVE record; Atlassian&rsquo;s Jira ticket says versions past end of life may also be affected, so an unsupported release needs a move to a fixed supported line, not a hunt for a patch.</p>
<table>
  <thead>
      <tr>
          <th>Product</th>
          <th>Fixed releases</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Jira Software Data Center</td>
          <td>9.12.40, 10.3.26, 11.3.12</td>
      </tr>
      <tr>
          <td>Jira Service Management Data Center</td>
          <td>5.12.40, 10.3.26, 11.3.12</td>
      </tr>
      <tr>
          <td>Confluence Data Center</td>
          <td>9.2.26, 10.2.19</td>
      </tr>
      <tr>
          <td>Bitbucket Data Center</td>
          <td>9.4.26, 10.2.8, 10.5.1</td>
      </tr>
      <tr>
          <td>Bamboo Data Center</td>
          <td>10.2.24, 12.1.12</td>
      </tr>
      <tr>
          <td>Crowd Data Center</td>
          <td>6.3.7, 7.0.3, 7.1.7, 7.2.4</td>
      </tr>
      <tr>
          <td>Crucible</td>
          <td>4.9.15</td>
      </tr>
      <tr>
          <td>Fisheye</td>
          <td>4.9.15</td>
      </tr>
  </tbody>
</table>
<p>Don&rsquo;t forget the quiet ones. Crowd and Bamboo often run on separate hosts owned by a different team, and Crowd is where the published credential-theft path lives.</p>
<h3 id="if-you-cannot-patch-today">If you cannot patch today</h3>
<p>Atlassian offers three interim mitigations, described in its per-product Jira tickets:</p>
<ol>
<li><strong>A WAF or reverse-proxy rule</strong> (all products) that blocks URLs where <code>..</code> sits next to <code>/</code>, <code>\</code> or <code>::</code>, including URL-encoded forms. Atlassian asks you to test that encoded variants are blocked.</li>
<li><strong>A Tomcat <code>RewriteValve</code></strong> (Confluence, Jira Software, Jira Service Management, Bamboo, Crowd): add <code>&lt;Valve className=&quot;org.apache.catalina.valves.rewrite.RewriteValve&quot; /&gt;</code> inside the application&rsquo;s <code>&lt;Context&gt;</code> in <code>conf/server.xml</code>, install Atlassian&rsquo;s <code>rewrite.config</code> under the application&rsquo;s <code>WEB-INF</code>, and restart each node in turn.</li>
<li><strong>A <code>urlrewrite.xml</code> rule</strong> for Bitbucket.</li>
</ol>
<p>Take the rule content from Atlassian&rsquo;s ticket for your product, not from a third-party blog. Atlassian is direct about the value: the mitigations are &ldquo;limited and not a replacement for patching.&rdquo; Where possible, remove unpatched instances from internet reach until they are upgraded.</p>
<h2 id="how-to-hunt">How to hunt</h2>
<p>Patching closes the hole; it does not tell you whether someone already read a file. Rapid7 recommends checking for compromise even after patching and describes a log search: decode each request line up to twice, then look for <code>..</code> directly next to <code>/</code>, <code>\</code> or <code>::</code>.</p>
<p>Here is a small implementation of that approach. It is <strong>our</strong> script, not Atlassian&rsquo;s published regex, and it will flag some benign requests. We ran it with Python 3 against a five-line synthetic log; the request paths in that test were made up to exercise the matcher and are not real exploit strings.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Flag access-log lines with &#39;..&#39; next to &#39;/&#39;, &#39;</span><span style="color:#ae81ff">\\</span><span style="color:#e6db74">&#39; or &#39;::&#39; after up to two
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">rounds of URL decoding. Illustrates the approach Rapid7 describes for
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">CVE-2026-21589; it is not Atlassian&#39;s published regex.
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">Usage: python3 scan_access_log.py access_log.txt&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> re
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> urllib.parse <span style="color:#f92672">import</span> unquote
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>PATTERN <span style="color:#f92672">=</span> re<span style="color:#f92672">.</span>compile(<span style="color:#e6db74">r</span><span style="color:#e6db74">&#34;(\.\.(/|</span><span style="color:#ae81ff">\\</span><span style="color:#e6db74">|::))|((/|</span><span style="color:#ae81ff">\\</span><span style="color:#e6db74">|::)\.\.)&#34;</span>)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">suspicious</span>(line):
</span></span><span style="display:flex;"><span>    candidates <span style="color:#f92672">=</span> [line]
</span></span><span style="display:flex;"><span>    decoded <span style="color:#f92672">=</span> line
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> _ <span style="color:#f92672">in</span> range(<span style="color:#ae81ff">2</span>):
</span></span><span style="display:flex;"><span>        decoded <span style="color:#f92672">=</span> unquote(decoded)
</span></span><span style="display:flex;"><span>        candidates<span style="color:#f92672">.</span>append(decoded)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> any(PATTERN<span style="color:#f92672">.</span>search(c) <span style="color:#66d9ef">for</span> c <span style="color:#f92672">in</span> candidates)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], errors<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;replace&#34;</span>) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> n, line <span style="color:#f92672">in</span> enumerate(f, <span style="color:#ae81ff">1</span>):
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">if</span> suspicious(line):
</span></span><span style="display:flex;"><span>            print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>n<span style="color:#e6db74">}</span><span style="color:#e6db74">: </span><span style="color:#e6db74">{</span>line<span style="color:#f92672">.</span>rstrip()<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>On the synthetic log it flagged a raw <code>::..::</code> request, a double-encoded <code>%252e%252e%252f</code> request and an encoded backslash variant, and ignored a filename containing <code>report..final</code>.</p>
<p>When you triage hits:</p>
<ul>
<li><strong>Start from October 5.</strong> Scan back at least to the advisory date; extend further if your logs allow, because the CVE record lists every release before the fixes as affected.</li>
<li><strong>Look at the status code and size.</strong> A <code>200</code> with a body on a traversal-shaped request deserves more attention than a <code>404</code>.</li>
<li><strong>Check every node and the proxy.</strong> In a Data Center cluster each node has its own Tomcat access log; a WAF or load balancer in front may log what the nodes did not.</li>
<li><strong>Note what the path aimed at.</strong> Requests for <code>WEB-INF</code> paths, configuration files or <code>.properties</code> files tell you which secrets to rotate.</li>
</ul>
<h2 id="if-a-read-may-have-succeeded">If a read may have succeeded</h2>
<p>Treat any secret in a web-root file as exposed, then contain and rotate in this order:</p>
<ol>
<li><strong>Crowd application credentials</strong> from <code>crowd.properties</code>, and review Crowd&rsquo;s application IP allowlists.</li>
<li><strong>Recently created users and group changes</strong>, especially additions to <code>jira-administrators</code> or other admin groups since October 5.</li>
<li><strong>Database credentials, API tokens and integration secrets</strong> held in configuration files the requests could have reached.</li>
<li><strong>Downstream systems</strong> those credentials unlock: build agents behind Bamboo, repositories behind Bitbucket, SSO-linked applications behind Crowd.</li>
</ol>
<p>Rapid7&rsquo;s advice matches: contain first, then rotate exposed credentials and secrets. If you work to US federal timelines, this flaw also falls under <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04&rsquo;s forensic triage requirements</a>, which the required action of every KEV entry added in October cites.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We have not tested the vulnerability or Atlassian&rsquo;s mitigations. The root-cause details come from watchTowr&rsquo;s research as reported by Rapid7 and Help Net Security; we did not reach watchTowr&rsquo;s original post or Atlassian&rsquo;s main advisory page during research, and rely on the CVE record and Atlassian&rsquo;s public Jira tickets for versions and mitigations. No public source we found names the files requested in observed attacks.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Description, eight products, unauthenticated web-root file read, no enumeration, CWE-552, CVSS 4.0 9.3 and vector, fixed releases, published October 5</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-21589">CVE-2026-21589 record</a> (read via the cvelistV5 repository)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>SSVC: Exploitation active, Automatable no, Technical Impact total</td>
          <td>CISA ADP container in the CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV addition October 15, due date October 18, BOD 26-04 forensic triage wording</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a> (JSON feed, catalog version 2026.10.21)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Cloud already patched; WAF, RewriteValve and urlrewrite mitigations; &ldquo;limited and not a replacement for patching&rdquo;; EOL versions may be affected</td>
          <td><a href="https://jira.atlassian.com/browse/JRASERVER-79546">Atlassian JRASERVER-79546</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td><code>::</code> traversal root cause; web-root reads; crowd.properties to jira-administrators path; log-search approach</td>
          <td><a href="https://www.rapid7.com/blog/post/etr-cve-2026-21589-critical-unauthenticated-arbitrary-file-access-in-atlassian-products/">Rapid7 analysis</a> summarizing watchTowr</td>
          <td>Qualified: secondary account of watchTowr research</td>
      </tr>
      <tr>
          <td>Exploitation attempts reported by Previdian on October 6; watchTowr released a checker, not an exploit</td>
          <td><a href="https://www.helpnetsecurity.com/?p=387226">Help Net Security</a></td>
          <td>Qualified: Rapid7 reports public file-read PoCs</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>BeyondTrust CVE-2026-102764: Patch, Then Hunt for SAWFLY</title><link>https://firewallsync.com/posts/beyondtrust-remote-support-cve-2026-102764-patch-and-hunt/</link><guid>https://firewallsync.com/posts/beyondtrust-remote-support-cve-2026-102764-patch-and-hunt/</guid><pubDate>Thu, 22 Oct 2026 18:41:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>A BeyondTrust Remote Support authentication bypass was exploited as a zero-day from late August 2026. SaaS instances are patched; self-hosted appliances need the BT26-05-RS patch, and patching does not remove an attacker who is already in. Here is what to patch, what to hunt for and what to rotate.</description><enclosure url="https://firewallsync.com/posts/beyondtrust-remote-support-cve-2026-102764-patch-and-hunt/cover_hu10350490808426910252.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/beyondtrust-remote-support-cve-2026-102764-patch-and-hunt/cover_hu10350490808426910252.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/beyondtrust-remote-support-cve-2026-102764-patch-and-hunt/cover_hu10350490808426910252.jpg" alt="A bright, empty data centre hall with black and white equipment cabinets under overhead trays of blue network cables" width="1200" height="630"></p><p>BeyondTrust Remote Support appliances running version 26.2.3 or earlier have an authentication bypass, CVE-2026-102764, that was exploited as a zero-day for about six weeks before it was disclosed. If you run a self-hosted appliance, apply the BT26-05-RS patch now. Then assume the appliance may have been compromised and check it, because the actor behind the attacks installed a web shell, a tunnel and a root backdoor that a patch does not remove.</p>
<p><strong>Scope.</strong> On October 22, 2026 we read the CVE record, the KEV catalog (version 2026.10.21), BeyondTrust&rsquo;s blog post on BT26-05 (last updated October 16), Google Threat Intelligence Group&rsquo;s analysis, the Canadian Centre for Cyber Security&rsquo;s alert AL26-025 and Rapid7&rsquo;s analysis. BeyondTrust&rsquo;s advisory page for BT26-05 did not load for us, so patch and version details below come from the CVE record, the Cyber Centre and Google, and we say so where it matters. We did not test the vulnerability.</p>
<h2 id="the-vulnerability">The vulnerability</h2>
<p>The CVE record, published by BeyondTrust on October 15, 2026, says internal components of Remote Support &ldquo;did not adequately validate the origin of the messages it received&rdquo;, which &ldquo;could allow an attacker to perform file operations on the appliance&rdquo;. It is classified as CWE-290, authentication bypass by spoofing, and affects Remote Support versions up to and including 26.2.3.</p>
<p>BeyondTrust scores it CVSS 4.0 9.3 (<code>AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H</code>): reachable over the network, no privileges, no user interaction. CISA&rsquo;s enrichment in the record rates it exploitation active, automatable yes, technical impact total. The Cyber Centre describes the impact as unauthorized file operations &ldquo;including arbitrary file uploads, potentially leading to full appliance compromise&rdquo;.</p>
<p>That combination matters more than the score. Remote Support appliances sit at the edge, broker privileged sessions into internal systems and can store credentials. An unauthenticated file upload on such a device is a direct path to the systems it administers.</p>
<h2 id="timeline">Timeline</h2>
<table>
  <thead>
      <tr>
          <th>Date (2026)</th>
          <th>Event</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>August 29</td>
          <td>The zero-day is used against one BeyondTrust-hosted Remote Support instance and one self-hosted appliance</td>
          <td>BeyondTrust; Google says &ldquo;late August&rdquo;</td>
      </tr>
      <tr>
          <td>September 28</td>
          <td>BeyondTrust finds activity on the hosted instance and contains it, it says within 27 minutes of first detection</td>
          <td>BeyondTrust</td>
      </tr>
      <tr>
          <td>September 30</td>
          <td>Mandiant shares findings from the self-hosted customer&rsquo;s investigation</td>
          <td>BeyondTrust</td>
      </tr>
      <tr>
          <td>October 8</td>
          <td>Patch released to SaaS and self-hosted customers</td>
          <td>BeyondTrust</td>
      </tr>
      <tr>
          <td>October 9</td>
          <td>SaaS tenants patched</td>
          <td>Google, Canadian Cyber Centre</td>
      </tr>
      <tr>
          <td>October 15</td>
          <td>Coordinated disclosure: CVE, BeyondTrust blog, Google analysis</td>
          <td>CVE record, BeyondTrust, Google</td>
      </tr>
      <tr>
          <td>October 16</td>
          <td>CISA adds the CVE to KEV; due date October 19; forensic triage required</td>
          <td>KEV catalog</td>
      </tr>
  </tbody>
</table>
<p>As of its last update on October 16, BeyondTrust said it knew of only those two affected customers and had seen no adversary activity since August. That is a statement about what BeyondTrust has observed, not a guarantee about every appliance.</p>
<h2 id="what-to-patch">What to patch</h2>
<table>
  <thead>
      <tr>
          <th>Deployment</th>
          <th>Action</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Remote Support SaaS</td>
          <td>No patch action; BeyondTrust says SaaS was patched automatically</td>
          <td>BeyondTrust, Cyber Centre, Google</td>
      </tr>
      <tr>
          <td>Self-hosted Remote Support 24.3.1 to 26.2.3</td>
          <td>Apply the BT26-05-RS patch (Google says from the <code>/appliance</code> interface)</td>
          <td>Cyber Centre, Google</td>
      </tr>
      <tr>
          <td>Self-hosted Remote Support older than 24.3.1</td>
          <td>Upgrade to a supported version, then apply the patch</td>
          <td>Cyber Centre, Google</td>
      </tr>
      <tr>
          <td>Self-hosted Privileged Remote Access</td>
          <td>Apply BeyondTrust&rsquo;s precautionary PRA patch; BeyondTrust found no exploit path in PRA</td>
          <td>BeyondTrust; Rapid7 says confirm the right patch with BeyondTrust Support</td>
      </tr>
  </tbody>
</table>
<p>The Cyber Centre also advises checking that automatic updates are enabled where appropriate. If you have more than one appliance, inventory them first; a forgotten secondary or lab instance is a common gap.</p>
<h2 id="what-the-attackers-did">What the attackers did</h2>
<p>Google Threat Intelligence Group attributes the exploitation to ARMORY CASTLE, &ldquo;a suspected PRC-nexus espionage threat actor (formerly tracked as UNC5221)&rdquo;. BeyondTrust attributes it, with moderate to high confidence, to the same group under its earlier name. Google&rsquo;s analysis describes:</p>
<ul>
<li><strong>SAWFLY</strong>, an obfuscated PHP web shell named <code>mangoapi.php</code>, placed under the appliance&rsquo;s Laravel <code>public</code> directory, that returns HTTP 404 unless a specific POST parameter (<code>_c94mfe</code>) is present.</li>
<li><strong>BARECLAW</strong>, a tunneler at <code>/tmp/ingredi_ws</code> providing a SOCKS5 proxy over WebSocket, listening on local TCP port 19800 and exposed through an added HAProxy route, <code>/uidchanarge</code>, in <code>/ns/config/haproxy.conf</code>.</li>
<li>A copy of <code>/bin/sh</code> at <code>/usr/bin/dbus-authd</code>, owned by root with the SUID bit set.</li>
<li>Persistence through <code>/home/bomgar/sbin/post_boot.d/zz-root-persist</code>, which adds a passwordless sudo rule to <code>/ns/config/sudoers.template</code> and rebuilds <code>/etc/sudoers</code>.</li>
<li>Deletion of logs on the appliance.</li>
</ul>
<p>Google publishes these SHA-256 hashes:</p>
<table>
  <thead>
      <tr>
          <th>File</th>
          <th>SHA-256</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>mangoapi.php</code> (SAWFLY)</td>
          <td><code>dc14e215269a93fdbc4c1a80b984c23faf7af0911c6dbb85309796a6f2ae95fb</code></td>
      </tr>
      <tr>
          <td><code>ingredi_ws</code> (BARECLAW)</td>
          <td><code>1f19b9a27d79c11bda00a1e8faf74a7fdd683d38ccef1b60df2fd0307e3aa462</code></td>
      </tr>
  </tbody>
</table>
<p>Rapid7 notes that customers cannot audit the appliance&rsquo;s host file system directly and should coordinate host-level inspection with BeyondTrust Support. BeyondTrust offers a threat-hunting package through its trust portal.</p>
<h2 id="how-to-hunt-without-relying-on-hashes">How to hunt without relying on hashes</h2>
<p>BeyondTrust warns that indicators from one victim &ldquo;rarely reappear at the next&rdquo; and that evidence stored only on the appliance may be missing or manipulated. Its blog and Google&rsquo;s analysis suggest hunting for behavior instead:</p>
<ul>
<li>Successful <code>/login</code> sessions from residential broadband or commercial VPN addresses, or outside business hours.</li>
<li><code>user_changed</code> events that raise Jump Item or team roles, and any unauthorized changes to accounts, permissions or session policies.</li>
<li>Requests to unknown <code>.php</code> paths, and WebSocket upgrades from unexpected sources.</li>
<li>Web service accounts spawning shells.</li>
<li>Gaps in syslog forwarding, especially gaps that start and end within one session.</li>
</ul>
<p>The last one depends on external logs. If the appliance does not already forward syslog and session events to a SIEM, BeyondTrust&rsquo;s advice is to start now; as Rapid7 points out, logging enabled today cannot recover evidence that was already deleted.</p>
<h2 id="if-an-appliance-may-be-compromised">If an appliance may be compromised</h2>
<p>Google&rsquo;s recommended order, condensed:</p>
<ol>
<li><strong>Preserve evidence first.</strong> Snapshot virtual appliances with memory, and collect syslog, reverse-proxy or WAF and firewall logs before reboots or changes.</li>
<li><strong>Contain.</strong> Block inbound traffic at the perimeter, restrict egress, end sessions, and disable, rather than delete, Jump Clients, Jumpoints and API accounts.</li>
<li><strong>Rebuild rather than clean.</strong> Redeploy from a clean, patched image, and restore only configuration backups from before the exposure window.</li>
<li><strong>Rotate credentials after patching or rebuilding</strong>, in dependency order: credentials in the vault and Password Safe, and any credential typed into a support session during the window. BeyondTrust adds: watch for unmanaged logins and review new MFA registrations on privileged accounts.</li>
<li><strong>Restrict the admin interfaces.</strong> Limit <code>/login</code> and <code>/appliance</code> by IP and require single sign-on with phishing-resistant MFA.</li>
</ol>
<p>If you find evidence of exploitation, BeyondTrust asks customers to open a severity 1 support case citing BT26-05.</p>
<p>For federal agencies, the KEV due date was October 19 and forensic triage is required; <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">our BOD 26-04 explainer</a> covers what that triage involves and what private teams can borrow from it. For how a KEV listing should outrank a score when you prioritize, see <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>BeyondTrust&rsquo;s BT26-05 advisory page did not load for us; we relied on its blog post, the CVE record, Google and the Canadian Cyber Centre for versions and patch names. Rapid7&rsquo;s article is a secondary source and is used only for statements it attributes to BeyondTrust or Mandiant and for its own clearly labelled advice. Attribution and indicators are Google&rsquo;s and BeyondTrust&rsquo;s; we have not verified them. Details may change as the vendors update their guidance.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 22, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Description, CWE-290, affected up to 26.2.3, CVSS 4.0 9.3 and vector, published October 15</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-102764">CVE-2026-102764 record</a> (read via the cvelistV5 repository)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA SSVC: exploitation active, automatable, total impact</td>
          <td>CISA-ADP container in the CVE record</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV: added October 16, due October 19, forensic triage Yes</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a>, version 2026.10.21</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Timeline (August 29 to October 15), two affected customers, PRA precautionary patch, SaaS auto-patched, hunting behaviors, rotation advice, Sev 1 case</td>
          <td><a href="https://beyondtrust.com/blog/entry/bt26-05">BeyondTrust: ARMORY CASTLE and BT26-05</a></td>
          <td>Verified against the vendor&rsquo;s blog</td>
      </tr>
      <tr>
          <td>Attribution, malware names, file paths, HAProxy route, hashes, response steps</td>
          <td><a href="https://cloud.google.com/blog/topics/threat-intelligence/armory-castle-exploits-beyondtrust">Google Threat Intelligence: ARMORY CASTLE Exploits BeyondTrust</a></td>
          <td>Qualified: Google&rsquo;s analysis, not independently verified</td>
      </tr>
      <tr>
          <td>BT26-05-RS patch for 24.3.1 to 26.2.3; upgrade first if older; SaaS patched October 9</td>
          <td><a href="https://cyber.gc.ca/en/alerts-advisories/al26-025-critical-vulnerability-impacting-beyondtrust-remote-support-cve-2026-102764">Canadian Centre for Cyber Security alert AL26-025</a> and Google</td>
          <td>Verified across two sources</td>
      </tr>
      <tr>
          <td>Host file system can&rsquo;t be audited directly; logs enabled now can&rsquo;t recover deleted evidence</td>
          <td><a href="https://www.rapid7.com/blog/post/etr-cve-2026-102764-critical-beyondtrust-remote-support-authentication-bypass-exploited-in-the-wild">Rapid7 analysis</a></td>
          <td>Qualified: secondary source</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Plex Media Server CVE-2026-96656 Chain: Patch and Hunt</title><link>https://firewallsync.com/posts/plex-media-server-cve-2026-96656-exploited-chain-patch-and-hunt/</link><guid>https://firewallsync.com/posts/plex-media-server-cve-2026-96656-exploited-chain-patch-and-hunt/</guid><pubDate>Thu, 22 Oct 2026 18:40:00 +0000</pubDate><dc:creator>John Marshall</dc:creator><category>Security Operations</category><description>Four Plex Media Server flaws fixed in 1.43.3.10861 are now in CISA's KEV catalog, and Plex says the chain is being widely exploited after a public proof of concept. Here is what each CVE does, which deadlines apply, and what to check if your server was exposed.</description><enclosure url="https://firewallsync.com/posts/plex-media-server-cve-2026-96656-exploited-chain-patch-and-hunt/cover_hu5450766343746298430.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/plex-media-server-cve-2026-96656-exploited-chain-patch-and-hunt/cover_hu5450766343746298430.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/plex-media-server-cve-2026-96656-exploited-chain-patch-and-hunt/cover_hu5450766343746298430.jpg" alt="A strip of red film curling across a pale surface toward a dark film reel standing on its edge" width="1200" height="630"></p><p>Plex Media Server builds older than 1.43.3.10861 contain four vulnerabilities that the US Cybersecurity and Infrastructure Security Agency (CISA) added to its Known Exploited Vulnerabilities (KEV) catalog on October 20, 2026. Plex says that, chained together, they let an unauthenticated remote attacker run code on the server, and that exploitation is now wide. The fix has been available since September. If you run Plex anywhere, at home, in a lab or on a company network, update it, then check whether it was compromised before you updated.</p>
<p><strong>Scope.</strong> We read the four CVE records in the CVE List, CISA&rsquo;s machine-readable advisory VA-26-266-01, the KEV catalog (version 2026.10.21) and Plex&rsquo;s security thread on its forums, including the October 19, 2026 update, on October 22, 2026. We did not test the vulnerabilities or the proof of concept. For how KEV deadlines work, see <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">our explainer on BOD 26-04</a>.</p>
<h2 id="what-is-affected-and-what-fixes-it">What is affected and what fixes it</h2>
<p>Every one of the four CVE records lists Plex Media Server before 1.43.3.10861 as affected and 1.43.3.10861 as unaffected. CISA, acting as the CVE Numbering Authority, published the records and the advisory on September 23, 2026.</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Weakness</th>
          <th>What the record says</th>
          <th>CVSS v3.1 / v4.0 (CVE record)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-96652</td>
          <td>SSRF (CWE-918)</td>
          <td>Any <code>X-Plex-Token</code> value lets an attacker put a full URL in the <code>protocol</code> parameter of <code>/player/timeline</code> and make the server POST to a destination of their choice</td>
          <td>4.3 / 5.3</td>
      </tr>
      <tr>
          <td>CVE-2026-96654</td>
          <td>URL injection (CWE-84)</td>
          <td>URL values in <code>searchOne</code> are not neutralized, so an attacker can call other plugins&rsquo; functions with their own parameters</td>
          <td>6.5 / 6.9</td>
      </tr>
      <tr>
          <td>CVE-2026-96655</td>
          <td>SSRF (CWE-918)</td>
          <td>An authenticated user can make the server request arbitrary internal or external addresses through the <code>/video/:/transcode/universal</code> path parameter</td>
          <td>4.3 / 5.3</td>
      </tr>
      <tr>
          <td>CVE-2026-96656</td>
          <td>File path control (CWE-73)</td>
          <td>An admin user can write files that may be executed on load. The <code>TranscoderH264Options</code> preference is appended verbatim to x264&rsquo;s options on every transcode, and at startup all <code>.so</code> files are loaded without signature, execute-bit or symbol checks</td>
          <td>7.2 / 8.6</td>
      </tr>
  </tbody>
</table>
<p>The individual scores look moderate. That is the trap: each score rates one flaw on its own, while the risk comes from the chain. Plex&rsquo;s update says that chained, the four &ldquo;allow an unauthenticated remote attacker&rdquo; to execute code as the server process. Plex has not published the chain step by step, and we have not reproduced it. Our reading of the CVE descriptions is that the request-forgery and injection flaws provide the access that the file-write flaw, which on its own needs admin rights, requires. Treat that as interpretation, not a confirmed sequence.</p>
<p>The same advisory lists a fifth flaw, CVE-2026-96651, a path traversal in <code>/system/agents/media/get</code> that can read files including the <code>PlexOnlineToken</code>, which the advisory says &ldquo;grants control of the Plex account and server&rdquo;. It is not in KEV, and CISA&rsquo;s enrichment dated September 23 records no exploitation. The same build fixes it.</p>
<h2 id="what-changed-from-routine-update-to-active-exploitation">What changed: from routine update to active exploitation</h2>
<table>
  <thead>
      <tr>
          <th>Date (2026)</th>
          <th>Event</th>
          <th>Source</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>September 1</td>
          <td>Plex posts a security notice telling users to update to Plex Media Server 1.43.3 and Plex Desktop 1.115.0</td>
          <td>Plex forum thread</td>
      </tr>
      <tr>
          <td>September 23</td>
          <td>CISA publishes the CVE records and advisory VA-26-266-01, fixed in 1.43.3.10861</td>
          <td>CVE records, CSAF advisory</td>
      </tr>
      <tr>
          <td>October 16</td>
          <td>A public proof of concept is released, according to Plex</td>
          <td>Plex forum update</td>
      </tr>
      <tr>
          <td>October 19</td>
          <td>Plex updates its notice: the flaws are &ldquo;being actively and widely exploited&rdquo;, attackers are installing cryptocurrency miners and proxy software, and unpatched internet-facing servers should be assumed compromised</td>
          <td>Plex forum update</td>
      </tr>
      <tr>
          <td>October 20</td>
          <td>CISA adds the four CVEs to KEV; its enrichment marks exploitation as active for all four</td>
          <td>KEV catalog, CVE records</td>
      </tr>
  </tbody>
</table>
<h3 id="kev-details">KEV details</h3>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Added</th>
          <th>Due date (federal agencies)</th>
          <th>Forensic triage</th>
          <th>CISA SSVC: Automatable / Technical impact</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-96652</td>
          <td>Oct 20</td>
          <td>Nov 3</td>
          <td>No</td>
          <td>No / Partial</td>
      </tr>
      <tr>
          <td>CVE-2026-96654</td>
          <td>Oct 20</td>
          <td>Oct 23</td>
          <td>No</td>
          <td>Yes / Partial</td>
      </tr>
      <tr>
          <td>CVE-2026-96655</td>
          <td>Oct 20</td>
          <td>Nov 3</td>
          <td>No</td>
          <td>No / Partial</td>
      </tr>
      <tr>
          <td>CVE-2026-96656</td>
          <td>Oct 20</td>
          <td>Oct 23</td>
          <td>Yes</td>
          <td>No / Total</td>
      </tr>
  </tbody>
</table>
<p>Each KEV entry says the flaw can be chained with the other three. The deadlines bind US federal civilian agencies under BOD 26-04; for everyone else they are a signal of urgency. CISA&rsquo;s catalog does not explain why two flaws got three days and two got fourteen.</p>
<h2 id="what-to-do-now">What to do now</h2>
<h3 id="1-update-and-confirm-the-build-number">1. Update and confirm the build number</h3>
<p>Update Plex Media Server to 1.43.3.10861 or later, then confirm the version under <strong>Settings → General</strong> in Plex Web, as Plex&rsquo;s notice describes. &ldquo;1.43.3&rdquo; alone is not enough to be sure: Plex&rsquo;s later update gives the full build number as the threshold. Plex also tells desktop users to update Plex Desktop.</p>
<h3 id="2-decide-whether-the-server-was-exposed">2. Decide whether the server was exposed</h3>
<p>Plex&rsquo;s guidance is that a server running an older build that was reachable from the internet &ldquo;may be compromised&rdquo;. Exposure usually came from Plex&rsquo;s Remote Access feature or a port forward. Plex specifically advises against forwarding port 32400 directly and suggests turning remote access off if you don&rsquo;t need it.</p>
<h3 id="3-look-for-the-signs-plex-lists">3. Look for the signs Plex lists</h3>
<p>Plex&rsquo;s October 19 update names these indicators:</p>
<ul>
<li>Processes you don&rsquo;t recognize running as the Plex user.</li>
<li>Unfamiliar <code>.so</code> (Linux) or <code>.dll</code> (Windows) files in the Plex <strong>Codecs</strong> folder.</li>
<li>Outbound connections to <code>proxystream.st</code>.</li>
<li>High CPU use while nothing is playing.</li>
<li>Transcoder options you did not set in <code>Preferences.xml</code>, which ties directly to CVE-2026-96656&rsquo;s <code>TranscoderH264Options</code> behavior.</li>
</ul>
<p>The location of the Codecs folder and <code>Preferences.xml</code> depends on your operating system and install method; Plex&rsquo;s support documentation lists them. A clean result is not proof of a clean server: these are the indicators Plex has published so far, and an attacker can remove files.</p>
<h3 id="4-if-you-find-something-or-cant-rule-it-out">4. If you find something, or can&rsquo;t rule it out</h3>
<p>Plex&rsquo;s update recommends rebuilding from a clean installation, clearing the Codecs folder and reviewing <code>Preferences.xml</code>. It also says to sign out of all devices and change the Plex account password, and to treat any network shares or credentials the Plex service account could reach as potentially exposed. Because CVE-2026-96651 can expose the server token, signing out all devices after updating is sensible even without other signs; Plex&rsquo;s notice does not explicitly address token rotation.</p>
<h3 id="for-companies">For companies</h3>
<p>If Plex appears in an asset inventory or scanner export, check for it on employee-run lab machines and NAS devices too. Media servers often run under accounts with broad read access to shared storage, which is what turns a hobby service into a foothold. Our guide to <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">matching the KEV feed against a scanner export</a> shows how to flag entries like these automatically. If you are new to SSRF, <a href="/posts/ssrf-explained-how-server-side-request-forgery-works-and-how-to-prevent-it/">our SSRF explainer</a> covers why &ldquo;make the server send a request&rdquo; is so often the first link in a chain.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We did not run the proof of concept or confirm the chain&rsquo;s steps. Plex&rsquo;s statements about exploitation, the proof-of-concept date and the indicators come from its forum post and are not independently verified. We did not read the researcher&rsquo;s write-up referenced in the CVE records. KEV entries and Plex&rsquo;s guidance may change; check the sources below for updates.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 22, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Affected versions (before 1.43.3.10861), descriptions, CWE and CVSS scores of the four CVEs; published September 23, 2026 by CISA as CNA</td>
          <td>CVE records <a href="https://www.cve.org/CVERecord?id=CVE-2026-96652">CVE-2026-96652</a>, <a href="https://www.cve.org/CVERecord?id=CVE-2026-96654">-96654</a>, <a href="https://www.cve.org/CVERecord?id=CVE-2026-96655">-96655</a>, <a href="https://www.cve.org/CVERecord?id=CVE-2026-96656">-96656</a> (CVE List, read via the cvelistV5 repository)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Advisory VA-26-266-01: risk evaluation, fix in 1.43.3.10861, CVE-2026-96651 details</td>
          <td><a href="https://raw.githubusercontent.com/cisagov/CSAF/develop/csaf_files/IT/white/2026/va-26-266-01.json">CISA CSAF advisory VA-26-266-01</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV additions on October 20, due dates, forensic triage flags, chaining note</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a>, version 2026.10.21 (JSON via cisagov/kev-data)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA SSVC values (exploitation active; automatable; technical impact)</td>
          <td>CISA-ADP containers in the CVE records</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>September 1 notice; October 19 update on wide exploitation, public proof of concept on October 16, indicators and remediation steps</td>
          <td><a href="https://forums.plex.tv/t/942319">Plex forum: Important Security Update for Plex Media Server v1.43.2 and earlier</a></td>
          <td>Qualified: vendor statements, not independently verified</td>
      </tr>
      <tr>
          <td>The chain&rsquo;s step-by-step mechanics</td>
          <td>Our reading of the CVE descriptions</td>
          <td>Qualified: not confirmed by Plex or reproduced</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>SSVC Deployer Model Explained: How Exploitation, Exposure, Automatable and Human Impact Set Patch Priority</title><link>https://firewallsync.com/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/</link><guid>https://firewallsync.com/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/</guid><pubDate>Sun, 18 Oct 2026 06:06:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>The CERT/CC SSVC deployer model turns four inputs into Defer, Scheduled, Out-of-cycle or Immediate. This guide reproduces all 72 rows of its example decision table, shows which inputs CISA's public CVE data already supplies, and flags where the model's own pages disagree.</description><enclosure url="https://firewallsync.com/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/cover_hu9859181224393853406.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/cover_hu9859181224393853406.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/ssvc-deployer-patch-priority-model-explained-exploitation-exposure-automatable-and-human-impact/cover_hu9859181224393853406.jpg" alt="A snowy, nearly empty downtown street at dusk with red traffic lights, a glass skywalk bridge between tall office buildings and one car in the distance" width="1200" height="630"></p><p>Most teams prioritize patches by a severity score, then argue about the exceptions. CERT/CC&rsquo;s SSVC takes a different route: it asks four questions about the vulnerability and your own systems, and a fixed lookup table returns a priority. This article explains the <strong>deployer</strong> version of that model, reproduces its full decision table so you can check it yourself, and shows how much of the answer CISA&rsquo;s public CVE data can and cannot provide.</p>
<p><strong>Scope and method.</strong> We read the CERT/CC SSVC documentation on October 18, 2026: the &ldquo;Prioritizing Patch Deployment&rdquo; page (the deployer model), and the reference pages for Human Impact, Safety Impact and Mission Impact. We extracted the 72-row decision table from the page and checked it programmatically (counts, completeness and consistency, described below). We also read the CISA Vulnrichment SSVC values and CISA Known Exploited Vulnerabilities (KEV) entries in six CVE records from the CVE Services API on October 18, 2026. The CERT/CC pages show no publication date, so we cite the read date. We did not test SSVC in a production vulnerability-management program, and this is research-based analysis, not firsthand operating experience. If you only want to read CISA&rsquo;s own three values, see our <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment guide</a>; this article covers the other decision a deployer faces.</p>
<h2 id="the-short-answer">The short answer</h2>
<ul>
<li><strong>Who it is for.</strong> The &ldquo;deployer&rdquo;: an organization that receives a patch or mitigation from a supplier and decides how fast to apply it to systems it runs. CERT/CC recommends the unit of work be the pairing of a remediation or mitigation with one product-version instance.</li>
<li><strong>Four inputs:</strong> Exploitation, System Exposure, Automatable, Human Impact.</li>
<li><strong>Four outputs:</strong> Defer, Scheduled, Out-of-cycle, Immediate.</li>
<li><strong>How many inputs you supply yourself:</strong> two (System Exposure and Human Impact). Exploitation and Automatable can come from public sources such as CISA&rsquo;s data, if you accept them.</li>
</ul>
<h2 id="the-four-inputs">The four inputs</h2>
<table>
  <thead>
      <tr>
          <th>Input (version)</th>
          <th>Values</th>
          <th>What it asks</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Exploitation (v1.1.0)</td>
          <td>None, Public PoC, Active</td>
          <td>None: no evidence of active exploitation and no public proof of concept. Public PoC: a typical public PoC exists (for example in Metasploit or ExploitDB) or the vulnerability has a well-known method of exploitation. Active: shared, observable, reliable evidence of use in the wild by real attackers, with credible public reporting</td>
      </tr>
      <tr>
          <td>System Exposure (v1.0.1)</td>
          <td>Small, Controlled, Open</td>
          <td>Small: local service or program, highly controlled network. Controlled: networked service with some access restrictions or mitigations already in place. Open: internet or another widely accessible network where access cannot plausibly be restricted</td>
      </tr>
      <tr>
          <td>Automatable (v2.0.0)</td>
          <td>No, Yes</td>
          <td>Can an attacker reliably automate steps 1 to 4 of the kill chain (reconnaissance, weaponization, delivery, exploitation)?</td>
      </tr>
      <tr>
          <td>Human Impact (v2.0.2)</td>
          <td>Low, Medium, High, Very High</td>
          <td>A combination of Safety Impact and Mission Impact, explained below</td>
      </tr>
  </tbody>
</table>
<p>&ldquo;PoC&rdquo; means proof of concept. Note the definition of <strong>Controlled</strong>: CERT/CC says a successful mitigation must reliably interrupt the adversary&rsquo;s attack, which requires that the attack be detectable both reliably and quickly enough to respond. A firewall rule nobody monitors does not necessarily qualify.</p>
<h3 id="human-impact-the-lookup-table">Human Impact: the lookup table</h3>
<p>Human Impact combines Safety Impact (v2.0.1: Negligible, Marginal, Critical, Catastrophic, based on IEC 61508 with modifications) with Mission Impact (v2.0.0: Degraded, MEF Support Crippled, MEF Failure, Mission Failure). &ldquo;MEF&rdquo; means Mission Essential Function. This is the lookup table published on the Human Impact page:</p>
<table>
  <thead>
      <tr>
          <th>Safety Impact</th>
          <th>Degraded</th>
          <th>MEF Support Crippled</th>
          <th>MEF Failure</th>
          <th>Mission Failure</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Negligible</td>
          <td>Low</td>
          <td>Low</td>
          <td>Medium</td>
          <td>Very High</td>
      </tr>
      <tr>
          <td>Marginal</td>
          <td>Low</td>
          <td>Low</td>
          <td>Medium</td>
          <td>Very High</td>
      </tr>
      <tr>
          <td>Critical</td>
          <td>Medium</td>
          <td>High</td>
          <td>High</td>
          <td>Very High</td>
      </tr>
      <tr>
          <td>Catastrophic</td>
          <td>Very High</td>
          <td>Very High</td>
          <td>Very High</td>
          <td>Very High</td>
      </tr>
  </tbody>
</table>
<p><strong>A documentation inconsistency you should know about.</strong> The same page also gives prose definitions of each Human Impact level, and they do not match the table in four cells:</p>
<table>
  <thead>
      <tr>
          <th>Safety / Mission</th>
          <th>Table says</th>
          <th>Prose says</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Marginal / Degraded</td>
          <td>Low</td>
          <td>Medium</td>
      </tr>
      <tr>
          <td>Marginal / MEF Support Crippled</td>
          <td>Low</td>
          <td>Medium</td>
      </tr>
      <tr>
          <td>Marginal / MEF Failure</td>
          <td>Medium</td>
          <td>High</td>
      </tr>
      <tr>
          <td>Critical / Degraded</td>
          <td>Medium</td>
          <td>High</td>
      </tr>
  </tbody>
</table>
<p>In addition, the prose does not mention Critical combined with MEF Failure at all, while the table gives High. We use the table, because it is the full 16-cell mapping and the prose definitions do not cover every combination. This is our judgment, not a CERT/CC statement. Before adopting the model, check which one your tooling implements, and check the current CERT/CC page, as this may be corrected.</p>
<h2 id="the-full-decision-table">The full decision table</h2>
<p>This is the example decision table from the CERT/CC deployer page (identified there as <code>ssvc:DT_DP:1.0.0</code>), reorganized so each row shows the outcome for all four Human Impact levels. The original lists 72 rows (3 Exploitation values × 3 System Exposure values × 2 Automatable values × 4 Human Impact values); we confirmed that each of the 72 combinations appears exactly once.</p>
<table>
  <thead>
      <tr>
          <th>Exploitation</th>
          <th>System Exposure</th>
          <th>Automatable</th>
          <th>Human Impact: Low</th>
          <th>Medium</th>
          <th>High</th>
          <th>Very High</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>None</td>
          <td>Small</td>
          <td>No</td>
          <td>Defer</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>None</td>
          <td>Small</td>
          <td>Yes</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>None</td>
          <td>Controlled</td>
          <td>No</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>None</td>
          <td>Controlled</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>None</td>
          <td>Open</td>
          <td>No</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>None</td>
          <td>Open</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Small</td>
          <td>No</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Small</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Controlled</td>
          <td>No</td>
          <td>Defer</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Controlled</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Open</td>
          <td>No</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Public PoC</td>
          <td>Open</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Small</td>
          <td>No</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Small</td>
          <td>Yes</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Controlled</td>
          <td>No</td>
          <td>Scheduled</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Controlled</td>
          <td>Yes</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Open</td>
          <td>No</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Immediate</td>
      </tr>
      <tr>
          <td>Active</td>
          <td>Open</td>
          <td>Yes</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Immediate</td>
          <td>Immediate</td>
      </tr>
  </tbody>
</table>
<h3 id="what-the-table-shows-our-counts">What the table shows (our counts)</h3>
<p>We counted outcomes across the 72 rows and checked the shape of the table:</p>
<table>
  <thead>
      <tr>
          <th>Outcome</th>
          <th>Rows out of 72</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Defer</td>
          <td>7</td>
      </tr>
      <tr>
          <td>Scheduled</td>
          <td>42</td>
      </tr>
      <tr>
          <td>Out-of-cycle</td>
          <td>20</td>
      </tr>
      <tr>
          <td>Immediate</td>
          <td>3</td>
      </tr>
  </tbody>
</table>
<ul>
<li><strong>Active exploitation never yields Defer.</strong> This matches a note on the CERT/CC page.</li>
<li><strong>Immediate needs Active exploitation and Open exposure.</strong> All three Immediate rows are Active, Open and one of: Automatable No with Very High Human Impact, or Automatable Yes with High or Very High Human Impact.</li>
<li><strong>No exploitation evidence is mostly Defer or Scheduled.</strong> The one exception is None, Open, Automatable Yes and Very High Human Impact, which is Out-of-cycle. The CERT/CC page makes the same observation.</li>
<li><strong>The table is monotonic.</strong> We compared every pair of rows: raising any single input, with the others unchanged or also raised, never lowered the priority in the table (zero violations out of all comparable pairs).</li>
</ul>
<h2 id="what-cisas-public-data-can-feed-in">What CISA&rsquo;s public data can feed in</h2>
<p>CISA&rsquo;s Vulnrichment adds an SSVC block to many CVE records, with three decision points: Exploitation, Automatable and Technical Impact. Exploitation and Automatable share names with two deployer inputs. We read six records on October 18, 2026; values were stamped by CISA on the dates shown and could have changed since:</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Product family (see our article)</th>
          <th>CISA SSVC timestamp</th>
          <th>Exploitation</th>
          <th>Automatable</th>
          <th>In KEV (date added)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-88771</td>
          <td><a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">Citrix NetScaler</a></td>
          <td>2026-09-27</td>
          <td>active</td>
          <td>yes</td>
          <td>Yes (2026-09-27)</td>
      </tr>
      <tr>
          <td>CVE-2026-102489</td>
          <td><a href="/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/">Zammad</a></td>
          <td>2026-10-02</td>
          <td>active</td>
          <td>no</td>
          <td>Yes (2026-10-02)</td>
      </tr>
      <tr>
          <td>CVE-2026-105207</td>
          <td><a href="/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/">ZITADEL</a></td>
          <td>2026-10-06</td>
          <td>none</td>
          <td>yes</td>
          <td>Not in the record</td>
      </tr>
      <tr>
          <td>CVE-2026-76504</td>
          <td><a href="/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/">Cisco Catalyst SD-WAN Manager</a></td>
          <td>2026-09-30</td>
          <td>active</td>
          <td>yes</td>
          <td>Yes (2026-09-30)</td>
      </tr>
      <tr>
          <td>CVE-2026-104286</td>
          <td><a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">FortiMail</a></td>
          <td>2026-10-02</td>
          <td>active</td>
          <td>yes</td>
          <td>Yes (2026-10-01)</td>
      </tr>
      <tr>
          <td>CVE-2026-88779</td>
          <td><a href="/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/">NetScaler SAML</a></td>
          <td>2026-10-04</td>
          <td>active</td>
          <td>yes</td>
          <td>Yes (2026-10-04)</td>
      </tr>
  </tbody>
</table>
<p>Three cautions apply before you reuse these as deployer inputs:</p>
<ol>
<li><strong>Definitions may differ.</strong> CISA labels its blocks with the role &ldquo;CISA Coordinator&rdquo; and tree version 2.0.3. We did not compare CISA&rsquo;s decision-point versions against the deployer model&rsquo;s versions line by line. The names and the value sets (none/poc/active; no/yes) match, but confirm before automating.</li>
<li><strong>A timestamp is a snapshot.</strong> CVE-2026-105207 was marked <code>none</code> on October 6; &ldquo;none&rdquo; says no evidence at that time. Recheck KEV and the vendor advisory.</li>
<li><strong>A missing block is missing data.</strong> No CISA block does not mean low risk. See the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV feed guide</a> for matching KEV to your inventory.</li>
</ol>
<p>Technical Impact is part of CISA&rsquo;s own tree but is not one of the deployer model&rsquo;s four inputs. Do not feed it in as if it were.</p>
<h2 id="worked-examples-same-flaw-different-answers">Worked examples: same flaw, different answers</h2>
<p>Suppose the affected product is running in your environment. We take Exploitation and Automatable from the table above, then show the full range of outcomes across your two inputs. All outcomes are read from the decision table.</p>
<table>
  <thead>
      <tr>
          <th>CVE (Exploitation / Automatable)</th>
          <th>Best case: Small exposure, Low Human Impact</th>
          <th>Worst case: Open exposure, Very High Human Impact</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-88771 (Active / Yes)</td>
          <td>Scheduled</td>
          <td>Immediate</td>
      </tr>
      <tr>
          <td>CVE-2026-102489 (Active / No)</td>
          <td>Scheduled</td>
          <td>Immediate</td>
      </tr>
      <tr>
          <td>CVE-2026-105207 (None / Yes)</td>
          <td>Defer</td>
          <td>Out-of-cycle</td>
      </tr>
  </tbody>
</table>
<p>More useful are the in-between cases. For CVE-2026-88771 (Active, Automatable Yes):</p>
<table>
  <thead>
      <tr>
          <th>System Exposure</th>
          <th>Low</th>
          <th>Medium</th>
          <th>High</th>
          <th>Very High</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Small</td>
          <td>Scheduled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Controlled</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
      </tr>
      <tr>
          <td>Open</td>
          <td>Out-of-cycle</td>
          <td>Out-of-cycle</td>
          <td>Immediate</td>
          <td>Immediate</td>
      </tr>
  </tbody>
</table>
<p>Read that second table carefully. An internet-facing instance (Open) supporting a function with Human Impact Medium is Out-of-cycle, not Immediate; the model reserves Immediate for High or Very High impact on an Open system. An instance with Controlled exposure is Out-of-cycle at every Human Impact level. This is the model&rsquo;s output, not an endorsement of those timings for your organization.</p>
<p>For CVE-2026-102489 (Active, Automatable No) an Open system with Medium Human Impact is Out-of-cycle, and a Controlled system with Medium Human Impact is only Scheduled, even though the flaw is listed in KEV. That is the kind of result a severity score alone would not give you, and the kind a risk owner may want to override. SSVC does not stop you from doing so.</p>
<h2 id="using-it-in-practice">Using it in practice</h2>
<ol>
<li><strong>Pre-assign Human Impact by asset class.</strong> Record, once, the Safety Impact and Mission Impact level for each class of system, then look up Human Impact in the table. Re-deciding it per CVE is where inconsistency creeps in.</li>
<li><strong>Define Exposure per instance, not per product.</strong> The same product can be Open in one place and Small in another. Apply CERT/CC&rsquo;s strict wording for Controlled.</li>
<li><strong>Treat mitigations as input changes.</strong> CERT/CC says a mitigation that successfully changes a decision point can lower the priority; it also warns that mitigations can fail, for example a firewall rule that is later changed. If you rely on one, record who owns it.</li>
<li><strong>Keep legal deadlines separate.</strong> Binding timelines, such as those in <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04</a> for federal agencies, are not the same thing as SSVC priorities.</li>
<li><strong>Compare with your current method.</strong> Our comparison of <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS, EPSS and KEV</a> shows what each signal answers; SSVC&rsquo;s deployer table consumes an exploitation signal and then adds your context.</li>
<li><strong>Reassess when inputs change.</strong> CERT/CC says remediation is not a final state, because rollbacks or new deployments can reintroduce a vulnerability.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>We read the CERT/CC example table; we did not read the underlying repository files or test CERT/CC&rsquo;s calculator tool.</li>
<li>The CERT/CC pages show no dates, so we cannot say when the table or the Human Impact inconsistency was last changed.</li>
<li>The CVE examples assume the affected product is present in your environment and are not statements about any real organization.</li>
<li>Values in live CVE records can change after October 18, 2026.</li>
<li>The outcome names describe urgency, not a service-level agreement. CERT/CC does not give hour or day limits for them on the pages we read, so we give none.</li>
<li>We did not evaluate how well SSVC performs compared with other methods.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>CERT/CC pages and CVE records read October 18, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Deployer model unit of work, four decision points, four outcomes and their descriptions; mitigation and remediation notes</td>
          <td><a href="https://certcc.github.io/SSVC/howto/deployer_tree/">CERT/CC SSVC: Prioritizing Patch Deployment</a></td>
          <td>Verified: page shows no date</td>
      </tr>
      <tr>
          <td>72-row deployer decision table (ssvc:DT_DP:1.0.0), notes on Active and None; our counts (7, 42, 20, 3), completeness and monotonicity checks</td>
          <td>Same page; our calculation from the extracted table</td>
          <td>Qualified: counts and checks are ours; the table is presented by CERT/CC as an example</td>
      </tr>
      <tr>
          <td>Exploitation, System Exposure and Automatable values and definitions</td>
          <td>Same page; <a href="https://certcc.github.io/SSVC/reference/decision_points/exploitation/">Exploitation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Human Impact definitions, Safety and Mission Impact value names, and the 16-cell lookup table</td>
          <td><a href="https://certcc.github.io/SSVC/reference/decision_points/human_impact/">CERT/CC SSVC: Human Impact</a>, <a href="https://certcc.github.io/SSVC/reference/decision_points/safety_impact/">Safety Impact</a>, <a href="https://certcc.github.io/SSVC/reference/decision_points/mission_impact/">Mission Impact</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Four cells where the Human Impact prose and table differ; Critical with MEF Failure missing from the prose</td>
          <td>Same Human Impact page; our comparison</td>
          <td>Qualified: our comparison; the use of the table is our judgment</td>
      </tr>
      <tr>
          <td>Vulnrichment values and KEV dates for six CVEs</td>
          <td>CVE Services records, e.g. <a href="https://cveawg.mitre.org/api/cve/CVE-2026-88771">CVE-2026-88771</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2026-102489">CVE-2026-102489</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2026-105207">CVE-2026-105207</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2026-76504">CVE-2026-76504</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2026-104286">CVE-2026-104286</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2026-88779">CVE-2026-88779</a></td>
          <td>Qualified: live records read October 18, 2026; values can change</td>
      </tr>
      <tr>
          <td>Worked-example outcomes</td>
          <td>Table above, our lookup</td>
          <td>Qualified: assumes the product is present and uses our stated exposure and impact assumptions</td>
      </tr>
      <tr>
          <td>Practical steps and their rationale</td>
          <td>Our analysis</td>
          <td>Qualified: not tested in a production program</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>How to Read a CVSS v4.0 Vector String, Metric by Metric, with Real CVE Records</title><link>https://firewallsync.com/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/</link><guid>https://firewallsync.com/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/</guid><pubDate>Sat, 17 Oct 2026 06:06:00 +0000</pubDate><dc:creator>Craig Leslie</dc:creator><category>Security Operations</category><description>A CVSS v4.0 score is only as meaningful as the vector behind it. This guide decodes every metric in a v4.0 vector string, shows how the same flaw scores 8.7 or 9.4 depending on one metric group, and explains why a published score usually assumes the worst case for threat data.</description><enclosure url="https://firewallsync.com/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/cover_hu18057320311752076734.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/cover_hu18057320311752076734.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/how-to-read-a-cvss-v4-0-vector-string-metric-by-metric-with-real-cve-records/cover_hu18057320311752076734.jpg" alt="Glowing green analog cockpit gauges and instruments on a dark helicopter panel at night, with blurred city lights in the background" width="1200" height="630"></p><p>A CVSS v4.0 score such as 9.3 tells you very little until you can read the vector string behind it. This article decodes the string metric by metric, using FIRST&rsquo;s specification and real CVE records published in September and October 2026, and then shows three things that change what a number means: which metric groups were used, which default values were silently applied, and which version produced the score.</p>
<p><strong>Scope and method.</strong> We read version 1.2 of the <a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document</a> on October 17, 2026. The example vectors and published scores come from CVE Services records (<code>cveawg.mitre.org</code>), read the same day. We recalculated every score with the open-source Red Hat <code>cvss</code> Python library (version 3.6). Its results matched all four published v4.0 base scores we checked (9.5, 8.7, 9.4 and 9.3). Scores for modified vectors, such as &ldquo;what if Exploit Maturity were Unreported&rdquo;, are our own recalculations and are labelled as such. This is research-based analysis, not an independent reimplementation of the scoring formula.</p>
<h2 id="the-short-answer-how-a-vector-is-built">The short answer: how a vector is built</h2>
<p>The specification defines the string like this:</p>
<ul>
<li>It begins with <code>CVSS:4.0</code>, then each metric follows after a <code>/</code> as <code>NAME:VALUE</code>, using the abbreviations in the spec (case sensitive).</li>
<li>All <strong>eleven Base metrics</strong> are mandatory. Threat, Environmental and Supplemental metrics are optional, and an omitted metric is treated as Not Defined (<code>X</code>).</li>
<li>Metrics must appear in the order of the spec&rsquo;s Table 23. &ldquo;Every other ordering is invalid.&rdquo; A metric may not appear twice.</li>
</ul>
<p>Here is the spec&rsquo;s own example, with each segment decoded:</p>
<p><code>CVSS:4.0/AV:N/AC:L/AT:N/PR:H/UI:N/VC:L/VI:L/VA:N/SC:N/SI:N/SA:N</code></p>
<table>
  <thead>
      <tr>
          <th>Segment</th>
          <th>Metric</th>
          <th>Meaning in this example</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>AV:N</code></td>
          <td>Attack Vector: Network</td>
          <td>Exploitable across networks, up to the whole Internet</td>
      </tr>
      <tr>
          <td><code>AC:L</code></td>
          <td>Attack Complexity: Low</td>
          <td>No target-specific circumvention of built-in defenses needed</td>
      </tr>
      <tr>
          <td><code>AT:N</code></td>
          <td>Attack Requirements: None</td>
          <td>Does not depend on special deployment conditions</td>
      </tr>
      <tr>
          <td><code>PR:H</code></td>
          <td>Privileges Required: High</td>
          <td>Attacker needs administrative-level privileges first</td>
      </tr>
      <tr>
          <td><code>UI:N</code></td>
          <td>User Interaction: None</td>
          <td>No other human has to act</td>
      </tr>
      <tr>
          <td><code>VC:L</code> <code>VI:L</code> <code>VA:N</code></td>
          <td>Vulnerable system confidentiality, integrity, availability</td>
          <td>Low, Low, None</td>
      </tr>
      <tr>
          <td><code>SC:N</code> <code>SI:N</code> <code>SA:N</code></td>
          <td>Subsequent system confidentiality, integrity, availability</td>
          <td>None: no impact beyond the vulnerable system</td>
      </tr>
  </tbody>
</table>
<p>We scored this vector at <strong>5.1 (Medium)</strong>. The spec states the vector but not a score, so that number is ours.</p>
<h2 id="the-base-metrics">The Base metrics</h2>
<p><strong>Exploitability metrics</strong> describe the &ldquo;vulnerable system&rdquo;, the thing that contains the flaw:</p>
<table>
  <thead>
      <tr>
          <th>Metric</th>
          <th>Values</th>
          <th>What it asks</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Attack Vector (AV)</td>
          <td>Network (N), Adjacent (A), Local (L), Physical (P)</td>
          <td>How far away can the attacker be? Larger reach, higher score</td>
      </tr>
      <tr>
          <td>Attack Complexity (AC)</td>
          <td>Low (L), High (H)</td>
          <td>Must the attacker defeat built-in security measures, for example exploit mitigations or by obtaining a target-specific secret?</td>
      </tr>
      <tr>
          <td>Attack Requirements (AT)</td>
          <td>None (N), Present (P)</td>
          <td>Does success depend on conditions that arise from deployment, such as a race condition or being on the network path?</td>
      </tr>
      <tr>
          <td>Privileges Required (PR)</td>
          <td>None (N), Low (L), High (H)</td>
          <td>What access must the attacker already have?</td>
      </tr>
      <tr>
          <td>User Interaction (UI)</td>
          <td>None (N), Passive (P), Active (A)</td>
          <td>Does another person have to take part, and did they have to override a protection?</td>
      </tr>
  </tbody>
</table>
<p><strong>Impact metrics</strong> come in two sets: three for the vulnerable system (VC, VI, VA) and three for any <strong>subsequent system</strong> (SC, SI, SA). The specification says all impacts outside the vulnerable system belong in the subsequent set; if there are none, they are None. Each takes High (H), Low (L) or None (N). This replaces the single &ldquo;Scope&rdquo; switch of earlier versions, which the spec calls the property &ldquo;formerly known as Scope&rdquo;.</p>
<p>Two scoring rules from the spec save arguments. First, Base metrics assume the attacker has &ldquo;perfect knowledge&rdquo; of the vulnerability. Second, a specific configuration needed for an attack does not lower the score: the vulnerable system is assessed as if it were in that configuration.</p>
<h2 id="worked-example-1-a-vector-with-attack-requirements">Worked example 1: a vector with Attack Requirements</h2>
<p><strong>CVE-2026-88771</strong>, a Citrix NetScaler flaw published on September 27, 2026 by NetScaler as the CVE Numbering Authority (CNA, the organization that assigns and describes the record), carries:</p>
<p><code>CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H</code> with a published base score of <strong>9.5 (Critical)</strong>.</p>
<p>Reading it: reachable over the network, low complexity, <strong>but</strong> <code>AT:P</code> means success depends on specific deployment conditions, no privileges or user action needed, and High impact on all three properties of both the vulnerable and subsequent systems. We covered the flaw itself in our <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">NetScaler patch-and-hunt article</a>. Our recalculation of this vector gave 9.5, the same as the published figure. Note that the record&rsquo;s metric lists Exploit Maturity and the Supplemental metrics as NOT_DEFINED, so this is a Base-only score.</p>
<h2 id="worked-example-2-one-flaw-two-scores">Worked example 2: one flaw, two scores</h2>
<p><strong>CVE-2026-102489</strong>, a Zammad flaw published on September 30, 2026 by the CNA DIVD, carries two CVSS v4.0 entries in the same record. They differ in exactly one respect: the subsequent-system impacts.</p>
<table>
  <thead>
      <tr>
          <th>Entry (scenario label in the record)</th>
          <th>SC / SI / SA</th>
          <th>Published base score</th>
          <th>Our recalculation</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>General</td>
          <td>Low / Low / Low</td>
          <td>8.7 (High)</td>
          <td>8.7</td>
      </tr>
      <tr>
          <td>&ldquo;Chained with CVE-2026-102490&rdquo;</td>
          <td>High / High / High</td>
          <td>9.4 (Critical)</td>
          <td>9.4</td>
      </tr>
  </tbody>
</table>
<p>Both vectors share <code>AV:N/AC:L/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H</code> and end in <code>/E:A/AU:Y/V:C</code>. A score is therefore a function of assumptions, here whether an attacker can chain the flaw with a second one. That is useful information, but it also means &ldquo;the CVSS score&rdquo; for this CVE is not one number. Read the scenario label before you copy a score into a ticket. For context on the flaws, see our <a href="/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/">Zammad article</a>.</p>
<h2 id="threat-environmental-and-supplemental-metrics">Threat, Environmental and Supplemental metrics</h2>
<table>
  <thead>
      <tr>
          <th>Group</th>
          <th>Who supplies it</th>
          <th>Metrics</th>
          <th>Effect on score</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Threat</td>
          <td>The consumer, from threat intelligence</td>
          <td>Exploit Maturity (E): Attacked (A), Proof-of-Concept (P), Unreported (U), Not Defined (X)</td>
          <td>Changes it</td>
      </tr>
      <tr>
          <td>Environmental</td>
          <td>The consumer, about their own environment</td>
          <td>Security requirements (CR, IR, AR: High, Medium, Low) and Modified Base metrics (MAV, MAC and so on)</td>
          <td>Changes it</td>
      </tr>
      <tr>
          <td>Supplemental</td>
          <td>Provider supplies, consumer decides use</td>
          <td>Safety (S), Automatable (AU), Recovery (R), Value Density (V), Vulnerability Response Effort (RE), Provider Urgency (U)</td>
          <td><strong>Does not change it</strong></td>
      </tr>
  </tbody>
</table>
<p>Three details matter in practice:</p>
<ol>
<li><strong>Defaults assume the worst.</strong> The spec says Exploit Maturity <code>X</code> &ldquo;is equivalent to Attacked (A) for the purposes of the calculation&rdquo; and that Security Requirements <code>X</code> equals High.</li>
<li><strong>Supplemental metrics are context only.</strong> The spec says no Supplemental metric has any impact on the final score. <code>AU:Y</code>, for example, means attackers can reliably automate all four early kill-chain steps; it does not raise the number.</li>
<li><strong>Environmental metrics can lower or raise a score.</strong> The spec&rsquo;s own example: a system on an isolated network could reduce Attack Vector from Network to Adjacent, and the spec also says it is acceptable to use modified metrics to represent situations that increase the score.</li>
</ol>
<h3 id="our-recalculation-what-threat-and-environmental-choices-do">Our recalculation: what Threat and Environmental choices do</h3>
<p>Starting from the Base vector of CVE-2026-105207 (a ZITADEL flaw published October 4, 2026 by the CNA VulnCheck), <code>CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N</code>, published at <strong>9.3</strong>:</p>
<table>
  <thead>
      <tr>
          <th>Change we made</th>
          <th>Vector suffix</th>
          <th>Score (our recalculation)</th>
          <th>Rating</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>None (published Base)</td>
          <td>none</td>
          <td>9.3</td>
          <td>Critical</td>
      </tr>
      <tr>
          <td>Exploit Maturity: Unreported</td>
          <td><code>/E:U</code></td>
          <td>8.1</td>
          <td>High</td>
      </tr>
      <tr>
          <td>Modified Attack Vector: Adjacent</td>
          <td><code>/MAV:A</code></td>
          <td>8.7</td>
          <td>High</td>
      </tr>
      <tr>
          <td>Modified Attack Vector: Local</td>
          <td><code>/MAV:L</code></td>
          <td>8.6</td>
          <td>High</td>
      </tr>
  </tbody>
</table>
<p>And from the general vector of CVE-2026-102489 with the <code>E:A</code>, <code>AU:Y</code> and <code>V:C</code> segments removed (8.7, High): adding <code>/E:P</code> gives <strong>7.4 (High)</strong> and <code>/E:U</code> gives <strong>6.3 (Medium)</strong>. Omitting E, or setting it to <code>A</code>, leaves the score at 8.7, as the spec&rsquo;s default rule predicts. These are illustrations of how the metric works, not claims that the flaws are unexploited. For CVE-2026-105207, CISA&rsquo;s enrichment in the same record (an SSVC entry timestamped October 6, 2026) lists Exploitation as &ldquo;none&rdquo;. We did not determine whether that evidence meets the specification&rsquo;s definition of Unreported, so do not treat 8.1 as the right score for that CVE; treat it as the size of the effect.</p>
<h2 id="reading-the-name-of-the-score">Reading the name of the score</h2>
<p>The spec requires the label to say which groups were used and says the label &ldquo;should be used wherever a numerical CVSS value is displayed&rdquo;:</p>
<table>
  <thead>
      <tr>
          <th>Name</th>
          <th>Metric groups used</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVSS-B</td>
          <td>Base</td>
      </tr>
      <tr>
          <td>CVSS-BT</td>
          <td>Base and Threat</td>
      </tr>
      <tr>
          <td>CVSS-BE</td>
          <td>Base and Environmental</td>
      </tr>
      <tr>
          <td>CVSS-BTE</td>
          <td>Base, Threat and Environmental</td>
      </tr>
  </tbody>
</table>
<p>The spec also notes that providers such as product maintainers and the National Vulnerability Database (NVD) &ldquo;typically provide only the Base Scores&rdquo;, that is, CVSS-B. By that naming, the entries of CVE-2026-102489 above, which include <code>E:A</code>, are strictly Base-plus-Threat scores even though the record field is called <code>baseScore</code>. We point this out because a field name in a CVE record does not guarantee which groups the number reflects: read the vector.</p>
<h2 id="v40-versus-v31-do-not-compare-the-numbers">v4.0 versus v3.1: do not compare the numbers</h2>
<p>CVE-2026-105207 carries both a v4.0 entry (9.3) and a v3.1 entry (9.8, vector <code>CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H</code>) from the same CNA. The qualitative bands line up (both Critical), and the spec says its score mapping was designed to be backwards compatible with v3.x severity band boundaries. But the v4.0 score comes from a different model: the spec describes expert-ranked groups of vectors called MacroVectors with interpolation inside each one, not the v3.x equations. Compare like with like: v4.0 against v4.0, v3.1 against v3.1, and the band only across versions.</p>
<h2 id="a-small-script-to-check-vectors">A small script to check vectors</h2>
<p>This reads one vector per line and prints the score or the error. It needs <code>pip install cvss</code>:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> sys
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> cvss <span style="color:#f92672">import</span> CVSS4
</span></span><span style="display:flex;"><span><span style="color:#f92672">from</span> cvss.exceptions <span style="color:#f92672">import</span> CVSSError
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">for</span> line <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>stdin:
</span></span><span style="display:flex;"><span>    vec <span style="color:#f92672">=</span> line<span style="color:#f92672">.</span>strip()
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> <span style="color:#f92672">not</span> vec:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">try</span>:
</span></span><span style="display:flex;"><span>        c <span style="color:#f92672">=</span> CVSS4(vec)
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">except</span> CVSSError <span style="color:#66d9ef">as</span> e:
</span></span><span style="display:flex;"><span>        print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;INVALID  </span><span style="color:#e6db74">{</span>vec<span style="color:#e6db74">}</span><span style="color:#ae81ff">\n</span><span style="color:#e6db74">         </span><span style="color:#e6db74">{</span>e<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">continue</span>
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;</span><span style="color:#e6db74">{</span>c<span style="color:#f92672">.</span>base_score<span style="color:#e6db74">:</span><span style="color:#e6db74">&gt;4</span><span style="color:#e6db74">}</span><span style="color:#e6db74">  </span><span style="color:#e6db74">{</span>c<span style="color:#f92672">.</span>severity<span style="color:#e6db74">:</span><span style="color:#e6db74">&lt;8</span><span style="color:#e6db74">}</span><span style="color:#e6db74">  </span><span style="color:#e6db74">{</span>vec<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>)
</span></span></code></pre></div><p>We ran it on the vectors in this article. Excerpt of the output:</p>
<pre tabindex="0"><code> 9.5  Critical  CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H
 9.3  Critical  CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N
 8.1  High      CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:U
INVALID  CVSS:4.0/AV:F/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N
         Invalid metric value in CVSS4 vector &#34;AV:F&#34;
INVALID  CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/SC:N/SI:N/SA:N
         Missing mandatory metrics &#34;VA&#34;
INVALID  AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N
         Malformed CVSS4 vector &#34;AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N&#34; is missing mandatory prefix or uses unsupported CVSS version
</code></pre><p>The three INVALID vectors are the spec&rsquo;s own invalid examples (bad value, missing mandatory metric, missing prefix). One finding worth knowing: the spec&rsquo;s example of a <strong>mis-ordered</strong> vector is invalid, but when we fed this library a vector that merely swaps <code>AC</code> and <code>AV</code> it accepted it and returned 6.9 (Medium). So a library accepting a string is not proof the string is spec-valid. If you ingest vectors from many sources, add your own check that the metric order matches Table 23. We tested one library, one version, one swapped pair; we did not test other implementations.</p>
<h2 id="a-practical-checklist">A practical checklist</h2>
<ol>
<li><strong>Read the prefix.</strong> <code>CVSS:4.0</code> and <code>CVSS:3.1</code> are different scales.</li>
<li><strong>Read the tail.</strong> Are there Threat (<code>E:</code>) or Environmental (<code>M...</code>, <code>CR</code>, <code>IR</code>, <code>AR</code>) segments? If so, it is not a Base-only score.</li>
<li><strong>Check the subsequent-system impacts</strong> (<code>SC</code>, <code>SI</code>, <code>SA</code>). They move scores by more than a point in our example and often encode a scenario.</li>
<li><strong>Treat Supplemental segments as notes</strong>, not score drivers.</li>
<li><strong>Record the whole vector</strong> next to the score in your ticket or scanner export, as FIRST requires publishers to give both the score and the vector.</li>
<li><strong>Do not use CVSS as an exploitation signal.</strong> Pair it with <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">EPSS and KEV</a>, and see our guide to <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">reading CISA Vulnrichment SSVC values</a>.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>The CVSS v4.0 specification gives the vector grammar and metric definitions; the exact scoring lookup tables and interpolation are in the specification&rsquo;s later sections, which we did not reproduce or independently reimplement. We rely on one library for recalculation, cross-checked against four published scores.</li>
<li>Scores for modified vectors are ours and are illustrations. The FIRST online calculator at <a href="https://www.first.org/cvss/calculator/4.0">first.org/cvss/calculator/4.0</a> is the reference tool; we did not run it for this article.</li>
<li>The CVE records are as they appeared on October 17, 2026 and can be amended by their CNAs.</li>
<li>Some CNAs, as in these records, include Threat and Supplemental values in the vector they publish.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Vector format, mandatory Base metrics, fixed order, optional groups default to Not Defined, example vector and invalid-vector examples</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document, v1.2</a> (read October 17, 2026)</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Base metric names, values and definitions; subsequent-system impacts; &ldquo;formerly Scope&rdquo;</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Exploit Maturity X equals Attacked; Security Requirements X equals High; Supplemental metrics do not change the score</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Nomenclature CVSS-B, -BT, -BE, -BTE; NVD and maintainers typically provide Base scores only</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Scoring via MacroVectors and interpolation; backwards-compatible severity bands</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 Specification Document</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-88771: published September 27, 2026; vector and 9.5</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2026-88771">CVE Services record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-102489: published September 30, 2026; two entries, 8.7 and 9.4, differing in SC/SI/SA; scenario label</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2026-102489">CVE Services record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-105207: published October 4, 2026; v4.0 9.3 and v3.1 9.8 vectors; SSVC Exploitation &ldquo;none&rdquo;</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2026-105207">CVE Services record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Recalculated scores (5.1, 6.3, 7.4, 8.1, 8.6, 8.7) and the swapped-order result (6.9)</td>
          <td>Our run of the Red Hat <code>cvss</code> library v3.6, October 17, 2026</td>
          <td>Qualified: our computation, one library; published scores matched on four records</td>
      </tr>
      <tr>
          <td>Choice of E:U as an appropriate rating for any specific CVE</td>
          <td>Not asserted</td>
          <td>Qualified: illustration only</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CVE Record States: Published, Rejected and Disputed, and What Each Means for Vulnerability Triage</title><link>https://firewallsync.com/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/</link><guid>https://firewallsync.com/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/</guid><pubDate>Fri, 16 Oct 2026 06:12:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>A CVE ID is not always a confirmed, fixable flaw. The CVE Record Format has two states, Published and Rejected, plus tags such as disputed and unsupported-when-assigned. Here is what the rules say, what real records look like, and how to handle each in a triage queue.</description><enclosure url="https://firewallsync.com/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/cover_hu9444216300449806353.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/cover_hu9444216300449806353.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/cover_hu9444216300449806353.jpg" alt="An open drawer of a pale grey metal card-index cabinet packed with index cards and divider tabs, with a wooden panel at left" width="1200" height="630"></p><p>A CVE ID (Common Vulnerabilities and Exposures identifier) looks like a settled fact: a number for one specific flaw. In practice the record behind it can be rejected, disputed, about a product nobody supports, or about a service you cannot patch. If your triage process treats every CVE ID in a scanner report the same way, it will waste effort on some and under-react to others. This guide explains the states and tags the CVE Program defines, shows real records, and suggests how to handle each case.</p>
<p><strong>Scope.</strong> This is research-based analysis plus one script we ran. On October 16, 2026 we read the CVE Record Format schema (the CVEProject/cve-schema repository on GitHub, main branch), its tag definitions, and the CVE Program&rsquo;s CNA Operational Rules version 4.2.0 (approved August 20, 2026, effective August 25, 2026). We fetched records from the CVE Services API and queried the NVD (National Vulnerability Database) API. We did not read the CVE Program&rsquo;s policy for disputing a CVE record, which the rules reference. For how CISA adds triage fields in an ADP container, see <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC values</a>.</p>
<h2 id="who-writes-a-cve-record">Who writes a CVE record</h2>
<p>The schema&rsquo;s own description says the record format is the blueprint that CVE Numbering Authorities (CNAs) and Authorized Data Publishers (ADPs) use to describe a CVE record. A CNA is an organization authorized to assign CVE IDs within a defined scope. The schema says a record has at least one <code>cna</code> container, with exactly one allowed because only one CNA assigns the ID, and any number of <code>adp</code> containers added by other organizations. The CNA container must carry certain fields; an ADP container does not have to.</p>
<p>That split matters for triage. Everything you read in the <code>cna</code> container is the assigning organization&rsquo;s statement. Anything in an <code>adp</code> container, such as CISA&rsquo;s SSVC fields, was added by someone else.</p>
<h2 id="the-two-states-published-and-rejected">The two states: PUBLISHED and REJECTED</h2>
<p>The schema defines a record as either &ldquo;Published&rdquo; (when a CNA populates the data for the ID) or &ldquo;Rejected&rdquo;, and the <code>state</code> field in <code>cveMetadata</code> has only those two values. IDs that have been reserved but not yet published have no public record in this format that we could find; we did not test how the API presents them.</p>
<table>
  <thead>
      <tr>
          <th>State</th>
          <th>What the record contains</th>
          <th>Key fields</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>PUBLISHED</td>
          <td>The CNA&rsquo;s description, affected products, references and more</td>
          <td><code>datePublished</code>, <code>descriptions</code>, <code>affected</code>, <code>references</code>, optional <code>tags</code></td>
      </tr>
      <tr>
          <td>REJECTED</td>
          <td>No vulnerability details; an explanation instead</td>
          <td><code>dateRejected</code>, required <code>rejectedReasons</code>, optional <code>replacedBy</code> (an array of CVE IDs)</td>
      </tr>
  </tbody>
</table>
<p>Source: the schema&rsquo;s <code>cveMetadataPublished</code>, <code>cveMetadataRejected</code> and <code>cnaRejectedContainer</code> definitions.</p>
<h3 id="why-records-get-rejected">Why records get rejected</h3>
<p>The CNA Operational Rules (version 4.2.0) say:</p>
<ul>
<li>CNAs <strong>must</strong> reject unused or unpublished CVE IDs (rule 4.5.3.5).</li>
<li>CNAs <strong>may</strong> reject a published record. Common reasons include determining that no vulnerability exists, using an incorrect CVE ID, a duplicate assignment, or combining multiple IDs (4.5.3.6).</li>
<li>A rejection must use the CVE Program&rsquo;s formats and must include an explanation (4.5.3.7).</li>
<li>When IDs are merged, the redundant published IDs must be rejected (4.5.3.8).</li>
</ul>
<h3 id="two-real-rejected-records">Two real rejected records</h3>
<p>We fetched both from the CVE Services API on October 16, 2026.</p>
<table>
  <thead>
      <tr>
          <th>CVE ID</th>
          <th>Assigner</th>
          <th>Reserved</th>
          <th>Rejected</th>
          <th>Reason text in the record</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2024-1992</td>
          <td>Wordfence</td>
          <td>2024-02-28</td>
          <td>2024-03-20</td>
          <td>&ldquo;Rejected as duplicate of CVE-2024-2306&rdquo;</td>
      </tr>
      <tr>
          <td>CVE-2019-1508</td>
          <td>Microsoft</td>
          <td>2018-11-26</td>
          <td>2020-01-24</td>
          <td>&ldquo;DO NOT USE THIS CANDIDATE NUMBER&hellip; This candidate was in a CNA pool that was not assigned to any issues during 2019.&rdquo;</td>
      </tr>
  </tbody>
</table>
<p>Neither record&rsquo;s <code>replacedBy</code> field was present, even though CVE-2024-1992&rsquo;s reason names a replacement in free text. If you automate &ldquo;follow the replacement&rdquo;, parse the reason text as a fallback and do not assume the structured field is populated. We examined only these two records, so we cannot say how common that is.</p>
<h2 id="tags-disputed-unsupported-when-assigned-exclusively-hosted-service">Tags: disputed, unsupported-when-assigned, exclusively-hosted-service</h2>
<p>The schema defines tag lists. CNAs can use three tags, and ADPs only one.</p>
<table>
  <thead>
      <tr>
          <th>Tag</th>
          <th>Who can set it</th>
          <th>Schema definition (paraphrased)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>disputed</code></td>
          <td>CNA or ADP</td>
          <td>One party disagrees with another party&rsquo;s assertion that an issue in software is a vulnerability</td>
      </tr>
      <tr>
          <td><code>unsupported-when-assigned</code></td>
          <td>CNA</td>
          <td>When the CVE request was received, the product was already end-of-life or not supported by the vendor; only for records where all affected products or version lines are end-of-life</td>
      </tr>
      <tr>
          <td><code>exclusively-hosted-service</code></td>
          <td>CNA</td>
          <td>All known affected software or hardware exists only in the affected hosted service; not used if on-premises products are also affected</td>
      </tr>
  </tbody>
</table>
<p>Sources: the schema&rsquo;s <code>cna-tags.json</code> and <code>adp-tags.json</code>, and CNA rules 4.2.17.1 (CNAs must use <code>unsupported-when-assigned</code> where all affected products no longer receive fixes), 4.5.3.10 and 5.1.11.</p>
<p>On disputes, the rules say CNAs and the Secretariat <strong>must</strong> use the <code>disputed</code> tag <strong>and</strong> provide an explanation in the record (4.5.3.10), and that CVE Program members <strong>must</strong> show the disputed status and the explanation when displaying CVE record data (4.5.3.11). The disputed tag marks a disagreement; it is not a verdict. The CNA rules point disputes to a separate CVE Program policy that we did not read.</p>
<h3 id="what-these-look-like-in-real-records">What these look like in real records</h3>
<table>
  <thead>
      <tr>
          <th>CVE ID</th>
          <th>State and tag</th>
          <th>What the record says</th>
          <th>Other checks on October 16, 2026</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-14683</td>
          <td>PUBLISHED, <code>disputed</code>, assigned by VulDB</td>
          <td>A local memory-allocation issue in HdrHistogram up to 2.2.2. Text includes &ldquo;It is still unclear if this vulnerability genuinely exists. This issue is disputed due to the potential lack of crossing of security boundaries and the pre-requisites for a successful attack.&rdquo;</td>
          <td>Published 2026-07-04, updated 2026-07-16; has a CISA-ADP container; NVD lists it as disputed with vulnStatus &ldquo;Deferred&rdquo;; EPSS 0.0016 (percentile 0.04694)</td>
      </tr>
      <tr>
          <td>CVE-2020-35235</td>
          <td>PUBLISHED, <code>unsupported-when-assigned</code>, assigned by mitre</td>
          <td>elFinder access-control flaw in the secure-file-manager WordPress plugin through 2.5 that lets any authenticated user run the upload command to achieve remote code execution; description notes it affects only products no longer supported by the maintainer</td>
          <td>Published 2020-12-14; EPSS 0.18284 (percentile 0.97186)</td>
      </tr>
  </tbody>
</table>
<p>The affected-products list of CVE-2020-35235 says vendor &ldquo;n/a&rdquo;, product &ldquo;n/a&rdquo;, version &ldquo;n/a&rdquo; with status &ldquo;affected&rdquo;, so a scanner cannot match it by version from the CVE record alone.</p>
<p>Two points follow. First, the disputed record in our example questions whether the issue is a vulnerability at all. Its EPSS score is low, but a low score is a probability estimate, not proof that the dispute is right. Second, an end-of-life tag does not mean low risk: EPSS, which estimates the probability of observed exploitation activity in the next 30 days, put CVE-2020-35235 at the 97th percentile in the file scored 2026-10-15. See our <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed guide</a> for how to read that column.</p>
<h2 id="the-affected-status-field-affected-unaffected-unknown">The affected-status field: affected, unaffected, unknown</h2>
<p>Inside a published record, each product version has a status of <code>affected</code>, <code>unaffected</code> or <code>unknown</code>. The schema says <code>unknown</code> means it is unknown or unspecified whether a version is affected, for reasons including that no investigation has been done or the vendor has not disclosed the status. The CNA rules require every record to identify at least one product as affected or unknown (5.1.4), say CNAs should resolve unknown status within a reasonable timeframe, and add that a permanent unknown may be appropriate, for instance when a product is suspected of being affected but has not been investigated (5.1.6). A record may not contain only unaffected products (5.1.8).</p>
<p>For triage: <code>unknown</code> is not a clean bill of health, and a version listed as <code>unaffected</code> is only as good as the CNA&rsquo;s analysis.</p>
<h2 id="how-big-are-these-categories">How big are these categories?</h2>
<p>We queried the NVD API on October 16, 2026. NVD is a separate database that ingests CVE records, so its numbers describe NVD&rsquo;s data, not the CVE List itself.</p>
<table>
  <thead>
      <tr>
          <th>Measure (NVD API, October 16, 2026)</th>
          <th>Result</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Total CVE records returned</td>
          <td>408,789</td>
      </tr>
      <tr>
          <td>Records returned with the <code>noRejected</code> parameter</td>
          <td>390,296</td>
      </tr>
      <tr>
          <td>Difference (implied rejected records)</td>
          <td>18,493 (4.5% of the total)</td>
      </tr>
      <tr>
          <td>Records with <code>cveTag=disputed</code></td>
          <td>1,509</td>
      </tr>
  </tbody>
</table>
<p>We inferred the rejected count by subtraction; we confirmed that CVE-2019-1508 has vulnStatus &ldquo;Rejected&rdquo; without the parameter and returns no results with it, and that CVE-2024-1992 has vulnStatus &ldquo;Rejected&rdquo;. The EPSS file we used elsewhere held 389,824 scored CVEs, close to the 390,296 non-rejected count but not identical. We did not investigate the 472-record gap, and the two files were queried hours apart.</p>
<h2 id="a-triage-playbook">A triage playbook</h2>
<p>These steps are our suggestions, not part of any CVE Program rule, and we did not test them in a production queue.</p>
<table>
  <thead>
      <tr>
          <th>What the scanner or feed shows</th>
          <th>What to check</th>
          <th>Suggested handling</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE state is REJECTED</td>
          <td><code>rejectedReasons</code>, then <code>replacedBy</code> or the reason text</td>
          <td>Close the finding as stale; if the reason names another CVE ID, make sure that ID is on your list</td>
      </tr>
      <tr>
          <td>Published, tagged <code>disputed</code></td>
          <td>The explanation in the record, the vendor&rsquo;s advisory, your own exposure</td>
          <td>Keep it in a &ldquo;needs decision&rdquo; state; record who decided and why. Do not auto-close and do not auto-escalate</td>
      </tr>
      <tr>
          <td>Tagged <code>unsupported-when-assigned</code></td>
          <td>Whether you still run the product</td>
          <td>If you do, the remedy is usually removal or replacement, since no vendor fix may exist. Check threat signals such as <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV</a> and EPSS before assuming it is low priority</td>
      </tr>
      <tr>
          <td>Tagged <code>exclusively-hosted-service</code></td>
          <td>Whether you use the hosted service</td>
          <td>There may be nothing for you to patch; ask the vendor for confirmation instead</td>
      </tr>
      <tr>
          <td>Affected status <code>unknown</code> for your version</td>
          <td>Vendor advisory</td>
          <td>Treat as unresolved, not safe</td>
      </tr>
      <tr>
          <td>In the CISA KEV catalog</td>
          <td>The KEV entry and its notes</td>
          <td>KEV is evidence of exploitation; see <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a></td>
      </tr>
  </tbody>
</table>
<h2 id="a-tested-script-to-read-state-and-tags">A tested script to read state and tags</h2>
<p>This script fetches a record from the CVE Services API and prints the fields above. It uses only the Python standard library. The endpoint is a public API; we did not need a key for single-record lookups, but bulk queries returned a message that a <code>CVE-API-ORG</code> header is required.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">triage_view</span>(cve_id):
</span></span><span style="display:flex;"><span>    url <span style="color:#f92672">=</span> <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;https://cveawg.mitre.org/api/cve/</span><span style="color:#e6db74">{</span>cve_id<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>
</span></span><span style="display:flex;"><span>    rec <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(url, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">30</span>))
</span></span><span style="display:flex;"><span>    meta <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;cveMetadata&#34;</span>]
</span></span><span style="display:flex;"><span>    cna <span style="color:#f92672">=</span> rec[<span style="color:#e6db74">&#34;containers&#34;</span>][<span style="color:#e6db74">&#34;cna&#34;</span>]
</span></span><span style="display:flex;"><span>    out <span style="color:#f92672">=</span> {<span style="color:#e6db74">&#34;cve&#34;</span>: meta[<span style="color:#e6db74">&#34;cveId&#34;</span>], <span style="color:#e6db74">&#34;state&#34;</span>: meta[<span style="color:#e6db74">&#34;state&#34;</span>],
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;assigner&#34;</span>: meta<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;assignerShortName&#34;</span>),
</span></span><span style="display:flex;"><span>           <span style="color:#e6db74">&#34;updated&#34;</span>: meta<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;dateUpdated&#34;</span>)}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> meta[<span style="color:#e6db74">&#34;state&#34;</span>] <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;REJECTED&#34;</span>:
</span></span><span style="display:flex;"><span>        out[<span style="color:#e6db74">&#34;rejectedReasons&#34;</span>] <span style="color:#f92672">=</span> [r[<span style="color:#e6db74">&#34;value&#34;</span>] <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> cna<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;rejectedReasons&#34;</span>, [])]
</span></span><span style="display:flex;"><span>        out[<span style="color:#e6db74">&#34;replacedBy&#34;</span>] <span style="color:#f92672">=</span> cna<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;replacedBy&#34;</span>, [])
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">else</span>:
</span></span><span style="display:flex;"><span>        out[<span style="color:#e6db74">&#34;cnaTags&#34;</span>] <span style="color:#f92672">=</span> cna<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;tags&#34;</span>, [])
</span></span><span style="display:flex;"><span>        out[<span style="color:#e6db74">&#34;adpTags&#34;</span>] <span style="color:#f92672">=</span> [t <span style="color:#66d9ef">for</span> a <span style="color:#f92672">in</span> rec[<span style="color:#e6db74">&#34;containers&#34;</span>]<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;adp&#34;</span>, [])
</span></span><span style="display:flex;"><span>                          <span style="color:#66d9ef">for</span> t <span style="color:#f92672">in</span> a<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;tags&#34;</span>, [])]
</span></span><span style="display:flex;"><span>        out[<span style="color:#e6db74">&#34;statuses&#34;</span>] <span style="color:#f92672">=</span> sorted({v[<span style="color:#e6db74">&#34;status&#34;</span>] <span style="color:#66d9ef">for</span> p <span style="color:#f92672">in</span> cna<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;affected&#34;</span>, [])
</span></span><span style="display:flex;"><span>                                  <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> p<span style="color:#f92672">.</span>get(<span style="color:#e6db74">&#34;versions&#34;</span>, [])})
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> out
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> cid <span style="color:#f92672">in</span> sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>:]:
</span></span><span style="display:flex;"><span>        print(json<span style="color:#f92672">.</span>dumps(triage_view(cid)))
</span></span></code></pre></div><p>Run on October 16, 2026 against five IDs:</p>
<pre tabindex="0"><code>{&#34;cve&#34;: &#34;CVE-2024-1992&#34;, &#34;state&#34;: &#34;REJECTED&#34;, &#34;assigner&#34;: &#34;Wordfence&#34;, &#34;updated&#34;: &#34;2024-03-20T16:44:07.699Z&#34;, &#34;rejectedReasons&#34;: [&#34;Rejected as duplicate of CVE-2024-2306&#34;], &#34;replacedBy&#34;: []}
{&#34;cve&#34;: &#34;CVE-2019-1508&#34;, &#34;state&#34;: &#34;REJECTED&#34;, &#34;assigner&#34;: &#34;microsoft&#34;, &#34;updated&#34;: &#34;2020-01-24T21:14:20.000Z&#34;, &#34;rejectedReasons&#34;: [&#34;DO NOT USE THIS CANDIDATE NUMBER. ConsultIDs: none. Reason: This candidate was in a CNA pool that was not assigned to any issues during 2019. Notes: none&#34;], &#34;replacedBy&#34;: []}
{&#34;cve&#34;: &#34;CVE-2026-14683&#34;, &#34;state&#34;: &#34;PUBLISHED&#34;, &#34;assigner&#34;: &#34;VulDB&#34;, &#34;updated&#34;: &#34;2026-07-16T07:18:18.631Z&#34;, &#34;cnaTags&#34;: [&#34;disputed&#34;], &#34;adpTags&#34;: [], &#34;statuses&#34;: [&#34;affected&#34;]}
{&#34;cve&#34;: &#34;CVE-2020-35235&#34;, &#34;state&#34;: &#34;PUBLISHED&#34;, &#34;assigner&#34;: &#34;mitre&#34;, &#34;updated&#34;: &#34;2024-08-04T17:02:07.522Z&#34;, &#34;cnaTags&#34;: [&#34;unsupported-when-assigned&#34;], &#34;adpTags&#34;: [], &#34;statuses&#34;: [&#34;affected&#34;]}
{&#34;cve&#34;: &#34;CVE-2021-44228&#34;, &#34;state&#34;: &#34;PUBLISHED&#34;, &#34;assigner&#34;: &#34;apache&#34;, &#34;updated&#34;: &#34;2025-10-21T23:25:23.121Z&#34;, &#34;cnaTags&#34;: [], &#34;adpTags&#34;: [], &#34;statuses&#34;: [&#34;affected&#34;]}
</code></pre><p>Known limits of the script: it assumes the record has a <code>cna</code> container and reads tags only from <code>tags</code>; it does not follow the version ranges inside <code>affected</code>; and it makes one request per ID, so do not run it over a whole scanner export without caching and rate limiting. A real pipeline should also handle HTTP errors, which we did not test.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We examined two rejected records, one disputed record and one end-of-life record out of hundreds of thousands, chosen for illustration, so they do not show how typical records look. Record contents and NVD counts change over time. We did not read the CVE Program&rsquo;s dispute policy and did not investigate NVD&rsquo;s own status labels such as &ldquo;Deferred&rdquo;.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 16, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Records have one <code>cna</code> container and any number of <code>adp</code> containers; the two states are Published and Rejected; rejected records carry <code>rejectedReasons</code> and optional <code>replacedBy</code>; status values affected, unaffected, unknown</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/CVE_Record_Format.json">CVE Record Format schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Tag definitions for disputed, unsupported-when-assigned and exclusively-hosted-service; ADPs may use only disputed</td>
          <td><a href="https://github.com/CVEProject/cve-schema/blob/main/schema/tags/cna-tags.json">CNA tags</a>, <a href="https://github.com/CVEProject/cve-schema/blob/main/schema/tags/adp-tags.json">ADP tags</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Rules on rejecting, disputed tagging and display, EOL tagging, unknown status</td>
          <td><a href="https://www.cve.org/Resources/Roles/Cnas/CNA_Rules_v4.2.0.pdf">CNA Operational Rules v4.2.0</a>, effective August 25, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2024-1992 and CVE-2019-1508 are rejected, with the dates and reasons shown</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2024-1992">CVE Services API record</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2019-1508">second record</a></td>
          <td>Verified on October 16, 2026</td>
      </tr>
      <tr>
          <td>CVE-2026-14683 is tagged disputed with the quoted text; CVE-2020-35235 is tagged unsupported-when-assigned</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2026-14683">CVE-2026-14683</a>, <a href="https://cveawg.mitre.org/api/cve/CVE-2020-35235">CVE-2020-35235</a></td>
          <td>Verified on October 16, 2026</td>
      </tr>
      <tr>
          <td>NVD totals: 408,789 records, 390,296 with noRejected, 1,509 tagged disputed; vulnStatus of the examples</td>
          <td><a href="https://services.nvd.nist.gov/rest/json/cves/2.0?cveTag=disputed&amp;resultsPerPage=1">NVD CVE API 2.0</a></td>
          <td>Qualified: NVD counts on October 16, 2026; the rejected count is our subtraction</td>
      </tr>
      <tr>
          <td>EPSS values for CVE-2020-35235 and CVE-2026-14683; 389,824 scored CVEs</td>
          <td><a href="https://epss.empiricalsecurity.com/epss_scores-current.csv.gz">EPSS daily CSV</a>, file scored 2026-10-15T12:03:56Z</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>EPSS definition: probability of observed exploitation activity in the next 30 days</td>
          <td><a href="https://www.first.org/epss/faq">FIRST EPSS FAQ</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Script output</td>
          <td>Run by us on October 16, 2026 against the five IDs shown</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Triage playbook</td>
          <td>Our analysis</td>
          <td>Qualified: not tested in production</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>EPSS Data Feed Columns: epss vs percentile, Joined to Scans</title><link>https://firewallsync.com/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/</link><guid>https://firewallsync.com/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/</guid><pubDate>Fri, 16 Oct 2026 06:11:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>FIRST publishes EPSS scores as a free daily CSV with three columns. Here is what epss and percentile each promise, how the October 15, 2026 file is distributed, how it lines up with CISA's KEV catalog, and a tested Python script that adds both columns to a scanner export.</description><enclosure url="https://firewallsync.com/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/cover_hu2864331919486454589.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/cover_hu2864331919486454589.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/cover_hu2864331919486454589.jpg" alt="Four aces (diamonds, spades, hearts, clubs) fanned out face up on a pale surface beside one face-down card with a red patterned back" width="1200" height="630"></p><p>The Exploit Prediction Scoring System (EPSS) is a model that estimates how likely it is that exploitation activity for a given CVE will be observed in the next 30 days. FIRST, the Forum of Incident Response and Security Teams, publishes the scores for every scored CVE once a day, free of charge. This guide covers the data file itself: its columns, what each promises, where it breaks, how today&rsquo;s numbers are distributed, and a script that adds EPSS to a vulnerability scanner export.</p>
<p><strong>Scope.</strong> This is research-based analysis plus one script we ran. On October 16, 2026 we read FIRST&rsquo;s EPSS pages (Get the Data, FAQ, Using EPSS, How EPSS Works) and the API documentation, downloaded the current CSV, called the API, and downloaded CISA&rsquo;s KEV JSON and the CVE Services records mentioned below. All counts are from the file scored 2026-10-15T12:03:56Z and change daily. We did not evaluate the model&rsquo;s accuracy; that is covered by the research FIRST links from its site. For how EPSS compares with CVSS and KEV as a decision signal, read <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>; for the KEV file, see <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">the KEV JSON feed guide</a>.</p>
<h2 id="what-the-score-is-and-what-it-is-not">What the score is, and what it is not</h2>
<p>FIRST&rsquo;s FAQ defines EPSS as a data-driven model that estimates the probability a vulnerability will be exploited in the wild within the next 30 days, producing a score between 0 and 1 for every publicly disclosed CVE, updated daily. The scores are generated by Empirical Security and published through FIRST&rsquo;s EPSS Special Interest Group.</p>
<p>FIRST is explicit about the limits. The FAQ says EPSS does not measure how much damage successful exploitation would cause, whether the CVE affects your environment, or any compensating controls. (Technical impact is one of the decision points in CISA&rsquo;s <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">Vulnrichment SSVC values</a>.) Its measurement is also narrower than &ldquo;exploited&rdquo;: the score estimates the likelihood that one of EPSS&rsquo;s data partners will detect, record and share evidence of exploitation <em>activity</em>. FIRST&rsquo;s How EPSS Works page adds that these sources record exploitation activity, not necessarily successful exploitation; honeypots are designed to be attacked, and intrusion prevention systems block what they see.</p>
<h2 id="three-ways-to-get-the-data">Three ways to get the data</h2>
<p>FIRST&rsquo;s <a href="https://www.first.org/epss/data">Get the Data</a> page lists three routes.</p>
<table>
  <thead>
      <tr>
          <th>Route</th>
          <th>Use it for</th>
          <th>Notes from FIRST&rsquo;s page</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Daily CSV, <code>https://epss.empiricalsecurity.com/epss_scores-current.csv.gz</code></td>
          <td>Bulk work: databases, joins against an asset inventory, dashboards</td>
          <td>Gzip-compressed; the URL is stable but redirects, so curl needs <code>-L</code>; contains every scored CVE</td>
      </tr>
      <tr>
          <td>API, <code>https://api.first.org/data/v1/epss</code></td>
          <td>Looking up one CVE or a small batch</td>
          <td>Not for bulk download or keeping a local copy in sync. Maintained by FIRST. The API documentation labels the endpoint BETA</td>
      </tr>
      <tr>
          <td>GitHub archive, <code>empiricalsec/epss_scores</code></td>
          <td>Historical analysis</td>
          <td>One file per day, from 2021-04-14</td>
      </tr>
  </tbody>
</table>
<p>The API documentation lists these query parameters: <code>cve</code> (comma-separated, up to 2,000 characters including commas), <code>date</code> (YYYY-MM-DD, since April 14, 2021), <code>epss-gt</code> / <code>epss-lt</code>, <code>percentile-gt</code> / <code>percentile-lt</code>, <code>days</code>, <code>q</code> (partial CVE ID search) and <code>scope=time-series</code> for the last 30 days. We tested two of these on October 16, 2026:</p>
<ul>
<li>A lookup for CVE-2021-44228 (Log4Shell) returned epss 0.99999 and percentile 1.0 for 2026-10-15.</li>
<li>The same CVE with <code>date=2022-02-28</code> returned epss 0.94402, percentile 0.99944. That value comes from a different model version than today&rsquo;s, which matters for the next section.</li>
</ul>
<h2 id="the-three-columns">The three columns</h2>
<p>The current file begins with a comment line, then a header, then one row per CVE. Ours began:</p>
<pre tabindex="0"><code>#model_version:v2026.06.15,score_date:2026-10-15T12:03:56Z
cve,epss,percentile
CVE-1999-0001,0.04753,0.91753
</code></pre><table>
  <thead>
      <tr>
          <th>Column</th>
          <th>Type</th>
          <th>Meaning (FIRST)</th>
          <th>Watch for</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>cve</code></td>
          <td>CVE ID</td>
          <td>The CVE identifier</td>
          <td>Normalize case and whitespace before joining</td>
      </tr>
      <tr>
          <td><code>epss</code></td>
          <td>Decimal, 0 to 1</td>
          <td>Probability of observed exploitation activity in the next 30 days</td>
          <td>A probability, not a severity; 0.05 means 5%</td>
      </tr>
      <tr>
          <td><code>percentile</code></td>
          <td>Decimal, 0 to 1</td>
          <td>The proportion of all scored vulnerabilities with the same score or lower</td>
          <td>A ranking of the <code>epss</code> value, so it depends on the whole population that day</td>
      </tr>
  </tbody>
</table>
<p>The <code>#</code> comment line is part of the file in current versions. FIRST says files from the v2 model onward (published from 2022-02-04) include it, while v1 files have no comment. A parser that expects the first line to be the header will fail, so skip lines that start with <code>#</code>.</p>
<h3 id="model-versions-break-time-series">Model versions break time series</h3>
<p>FIRST lists five model versions and when each began publishing: v1 (2021-04-14), v2 (v2022.01.01, 2022-02-04), v3 (v2023.03.01, 2023-03-07), v4 (v2025.03.14, 2025-03-17) and v5 (v2026.06.15, 2026-06-15). It warns that each version produced a score shift on its start date, so a series that crosses one of them &ldquo;reflects a change in methodology, not a change in the vulnerability itself.&rdquo; If you graph a CVE&rsquo;s score over months, mark those dates. Our own file reported <code>v2026.06.15</code>, matching the fifth entry.</p>
<p>FIRST also says that day-to-day changes inside a version are by design: scores move because new signals arrived (for example exploit code appearing or new analysis being published), not because the model changed.</p>
<h2 id="probability-vs-percentile-with-todays-numbers">Probability vs percentile, with today&rsquo;s numbers</h2>
<p>FIRST&rsquo;s FAQ says the probability is the model&rsquo;s primary output and the percentile is a transformation of it. Because most CVEs are never exploited, the distribution is skewed toward zero. We measured it on the October 15, 2026 file:</p>
<table>
  <thead>
      <tr>
          <th>Measure (389,824 scored CVEs)</th>
          <th>Count</th>
          <th>Share of file</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Median <code>epss</code></td>
          <td>0.00649</td>
          <td>n/a</td>
      </tr>
      <tr>
          <td>Mean <code>epss</code></td>
          <td>0.0267</td>
          <td>n/a</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.01</td>
          <td>147,786</td>
          <td>37.9%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.04</td>
          <td>37,460</td>
          <td>9.6%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.05</td>
          <td>30,749</td>
          <td>7.9%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.10</td>
          <td>17,309</td>
          <td>4.4%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.50</td>
          <td>4,332</td>
          <td>1.1%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.90</td>
          <td>801</td>
          <td>0.2%</td>
      </tr>
      <tr>
          <td><code>percentile</code> at least 0.90</td>
          <td>38,987</td>
          <td>10.0%</td>
      </tr>
      <tr>
          <td><code>percentile</code> at least 0.99</td>
          <td>3,903</td>
          <td>1.0%</td>
      </tr>
  </tbody>
</table>
<p>Three readings follow. These are our interpretations of the table, not FIRST statements.</p>
<ol>
<li><strong>A &ldquo;failing grade&rdquo; intuition misleads.</strong> Only 0.2% of CVEs scored 0.9 or higher. A threshold of 0.5 ignores 98.9% of the file. FIRST makes the same point, saying a 50% threshold &ldquo;deprioritizes the 98.8% of all published vulnerabilities&rdquo; (its figure was computed on its own data and rounds slightly differently from ours).</li>
<li><strong>Small numbers can be high.</strong> FIRST says 0.10 sits around the 95th percentile; in our file 4.4% of CVEs were at or above 0.1, so about the 95.6th. The 90th percentile in our file corresponds to a probability of about 0.0385, which is consistent with FIRST&rsquo;s statement that the 90th percentile is roughly 0.04.</li>
<li><strong>Our mean and median are close to FIRST&rsquo;s.</strong> FIRST says the mean is around 2.8% and the median around 0.7%; ours are 2.7% and 0.65%. FIRST does not say which day its figures describe.</li>
</ol>
<p>FIRST also offers a way to translate a CVSS-based policy. It says that over the previous 12 months about 61,000 CVEs were published and just over 10% were rated CVSS Critical, so acting on roughly the 90th EPSS percentile (a score of about 0.04 or more) yields a similar-sized workload as acting on CVSS Critical, and that acting on CVSS High and above corresponds to about 0.008. These are FIRST&rsquo;s rules of thumb for migrating a program, which it describes as a starting point rather than an endorsement of CVSS categories. We did not reproduce them.</p>
<h3 id="combining-probabilities">Combining probabilities</h3>
<p>FIRST&rsquo;s Using EPSS page gives two formulas, both assuming exploitation events are approximately independent: the probability that at least one of <em>n</em> vulnerabilities is exploited is 1 minus the product of (1 − p), and the expected number exploited is the sum of the scores. FIRST&rsquo;s example is 100 vulnerabilities each scoring 0.05. We checked the arithmetic: 1 − 0.95^100 = 0.9941, or 99.4%, and the expected count is 5. FIRST notes that both estimates depend on calibration holding at portfolio level and should not be treated as precise forecasts. Real findings are rarely independent (many hosts share one CVE, and one campaign can use several), so use these for trend tracking, not as a promise.</p>
<h2 id="how-epss-lines-up-with-cisas-kev-catalog">How EPSS lines up with CISA&rsquo;s KEV catalog</h2>
<p>We joined the file to CISA&rsquo;s KEV JSON, catalog version 2026.10.15 (released 2026-10-15T17:00:11Z, 1,746 entries). The EPSS file was scored about five hours earlier, so an entry added in between could have been missing; none was. All 1,746 KEV CVE IDs had an EPSS row.</p>
<table>
  <thead>
      <tr>
          <th>KEV entries (n = 1,746)</th>
          <th>Count</th>
          <th>Share</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>percentile</code> at least 0.90</td>
          <td>1,470</td>
          <td>84.2%</td>
      </tr>
      <tr>
          <td><code>percentile</code> below 0.90</td>
          <td>276</td>
          <td>15.8%</td>
      </tr>
      <tr>
          <td><code>epss</code> at least 0.50</td>
          <td>870</td>
          <td>49.8%</td>
      </tr>
      <tr>
          <td><code>epss</code> below 0.01</td>
          <td>61</td>
          <td>3.5%</td>
      </tr>
      <tr>
          <td>Median <code>epss</code></td>
          <td>0.4912</td>
          <td>n/a</td>
      </tr>
  </tbody>
</table>
<p>KEV is about 0.45% of the scored file (1,746 of 389,824), which is in line with FIRST&rsquo;s statement that KEV lists roughly 0.5% of published CVEs. Two cautions on reading this table:</p>
<ul>
<li><strong>The two measure different things.</strong> KEV records that CISA has evidence a CVE was exploited. EPSS estimates activity in the next 30 days from many signals. A KEV CVE with a low EPSS score is not a contradiction; the old attack may be quiet now, and per FIRST the score reflects current signals.</li>
<li><strong>Overlap is not proof of accuracy.</strong> FIRST lists VulnCheck&rsquo;s KEV among its data partners; its page does not say CISA&rsquo;s catalog is an input. We cannot tell from the public pages how much of the KEV overlap is built into the training data, so do not read the 84.2% as an independent validation.</li>
</ul>
<p>The practical takeaway matches <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">our earlier ordering</a>: a KEV listing is evidence that exploitation has happened, and should not be downgraded because EPSS is low.</p>
<h2 id="a-tested-script-to-add-epss-to-a-scanner-export">A tested script to add EPSS to a scanner export</h2>
<p>This script downloads the current file, normalizes the CVE ID in each scanner row, and appends <code>epss</code> and <code>epss_percentile</code>. It uses only the Python standard library.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> gzip<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>EPSS_URL <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://epss.empiricalsecurity.com/epss_scores-current.csv.gz&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_epss</span>(url<span style="color:#f92672">=</span>EPSS_URL):
</span></span><span style="display:flex;"><span>    raw <span style="color:#f92672">=</span> gzip<span style="color:#f92672">.</span>decompress(urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(url)<span style="color:#f92672">.</span>read())<span style="color:#f92672">.</span>decode(<span style="color:#e6db74">&#34;utf-8&#34;</span>)
</span></span><span style="display:flex;"><span>    lines <span style="color:#f92672">=</span> raw<span style="color:#f92672">.</span>splitlines()
</span></span><span style="display:flex;"><span>    header <span style="color:#f92672">=</span> lines[<span style="color:#ae81ff">0</span>] <span style="color:#66d9ef">if</span> lines[<span style="color:#ae81ff">0</span>]<span style="color:#f92672">.</span>startswith(<span style="color:#e6db74">&#34;#&#34;</span>) <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>
</span></span><span style="display:flex;"><span>    body <span style="color:#f92672">=</span> lines[<span style="color:#ae81ff">1</span>:] <span style="color:#66d9ef">if</span> header <span style="color:#66d9ef">else</span> lines
</span></span><span style="display:flex;"><span>    scores <span style="color:#f92672">=</span> {r[<span style="color:#e6db74">&#34;cve&#34;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#f92672">.</span>upper(): (float(r[<span style="color:#e6db74">&#34;epss&#34;</span>]), float(r[<span style="color:#e6db74">&#34;percentile&#34;</span>]))
</span></span><span style="display:flex;"><span>              <span style="color:#66d9ef">for</span> r <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(body)}
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> header, scores
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">main</span>(scanner_csv, out_csv):
</span></span><span style="display:flex;"><span>    header, scores <span style="color:#f92672">=</span> load_epss()
</span></span><span style="display:flex;"><span>    print(<span style="color:#e6db74">&#34;EPSS file header:&#34;</span>, header <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;(none)&#34;</span>, <span style="color:#e6db74">&#34;| CVEs scored:&#34;</span>, len(scores))
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> open(scanner_csv, newline<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;&#34;</span>) <span style="color:#66d9ef">as</span> f, open(out_csv, <span style="color:#e6db74">&#34;w&#34;</span>, newline<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;&#34;</span>) <span style="color:#66d9ef">as</span> g:
</span></span><span style="display:flex;"><span>        rd <span style="color:#f92672">=</span> csv<span style="color:#f92672">.</span>DictReader(f)
</span></span><span style="display:flex;"><span>        wr <span style="color:#f92672">=</span> csv<span style="color:#f92672">.</span>DictWriter(g, fieldnames<span style="color:#f92672">=</span>rd<span style="color:#f92672">.</span>fieldnames <span style="color:#f92672">+</span> [<span style="color:#e6db74">&#34;epss&#34;</span>, <span style="color:#e6db74">&#34;epss_percentile&#34;</span>])
</span></span><span style="display:flex;"><span>        wr<span style="color:#f92672">.</span>writeheader()
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">for</span> row <span style="color:#f92672">in</span> rd:
</span></span><span style="display:flex;"><span>            hit <span style="color:#f92672">=</span> scores<span style="color:#f92672">.</span>get(row[<span style="color:#e6db74">&#34;cve&#34;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#f92672">.</span>upper())
</span></span><span style="display:flex;"><span>            row[<span style="color:#e6db74">&#34;epss&#34;</span>], row[<span style="color:#e6db74">&#34;epss_percentile&#34;</span>] <span style="color:#f92672">=</span> hit <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> (<span style="color:#e6db74">&#34;&#34;</span>, <span style="color:#e6db74">&#34;&#34;</span>)
</span></span><span style="display:flex;"><span>            wr<span style="color:#f92672">.</span>writerow(row)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">if</span> __name__ <span style="color:#f92672">==</span> <span style="color:#e6db74">&#34;__main__&#34;</span>:
</span></span><span style="display:flex;"><span>    main(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">2</span>])
</span></span></code></pre></div><p>We ran it on October 16, 2026 against a five-row CSV we wrote, with a <code>cve</code> column:</p>
<pre tabindex="0"><code>host,cve
web-1,CVE-2021-44228
web-2,cve-2023-4863 
db-1,CVE-2024-0001
bad-1,CVE-2099-0001
web-3,CVE-1999-0001
</code></pre><p>Output:</p>
<pre tabindex="0"><code>EPSS file header: #model_version:v2026.06.15,score_date:2026-10-15T12:03:56Z | CVEs scored: 389824
host,cve,epss,epss_percentile
web-1,CVE-2021-44228,0.99999,1.0
web-2,cve-2023-4863 ,0.99979,0.9998
db-1,CVE-2024-0001,0.00949,0.60451
bad-1,CVE-2099-0001,,
web-3,CVE-1999-0001,0.04753,0.91753
</code></pre><p>Row <code>web-2</code> shows the normalization working (lowercase with a trailing space still matched), and <code>bad-1</code>, a made-up ID, comes out blank. The script prints the model version and score date; store those with the output so you can later explain why a score differs.</p>
<h3 id="pitfalls">Pitfalls</h3>
<ul>
<li><strong>Blank is not zero.</strong> A blank score means no EPSS row was found. Do not coerce it to 0 before sorting.</li>
<li><strong>Rejected CVEs are not scored.</strong> CVE-2024-1992 (rejected as a duplicate of CVE-2024-2306) and CVE-2019-1508 (rejected as an unused candidate number) were both absent from the file. If your scanner emits one, the right action is to look at the CVE record&rsquo;s state; see our companion article on <a href="/posts/cve-record-states-published-rejected-and-disputed-what-each-means-for-vulnerability-triage/">CVE record states</a>.</li>
<li><strong>Disputed CVEs are scored.</strong> CVE-2026-14683, which its CNA tagged as disputed, had epss 0.0016 and percentile 0.04694. A score does not tell you whether the dispute is valid.</li>
<li><strong>Old scores in tickets age badly.</strong> FIRST says scores are recomputed every day. Store the score date next to the value, or re-pull daily.</li>
<li><strong>Use the API sparingly.</strong> FIRST asks that bulk work use the CSV or the GitHub archive.</li>
</ul>
<h3 id="production-suggestions">Production suggestions</h3>
<p>These are our suggestions and were not tested beyond the script above: download once a day and keep each day&rsquo;s file; fail the job if the header does not contain <code>cve,epss,percentile</code> or the row count drops sharply; store the <code>model_version</code> and <code>score_date</code> comment with each load; and treat EPSS as one column next to exposure, asset criticality and KEV status. FIRST&rsquo;s Using EPSS page recommends combining EPSS with environment-specific signals because the score has no knowledge of your assets.</p>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We tested the script only on a five-row file, not a real scanner export. Counts are from one file scored on October 15, 2026, and one KEV version. We did not assess the model&rsquo;s accuracy or calibration, did not read the research papers FIRST links, and did not verify FIRST&rsquo;s own figures beyond the comparisons stated above. FIRST&rsquo;s Using EPSS page states some thresholds without a date, so treat them as of the page we read on October 16, 2026. The API documentation has a rate-limit section that we did not review, so read it before building anything that makes many calls.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 16, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>EPSS estimates the probability of exploitation in the next 30 days, a 0 to 1 score updated daily; not a complete risk score; 0.10 sits around the 95th percentile; KEV lists roughly 0.5% of published CVEs</td>
          <td><a href="https://www.first.org/epss/faq">FIRST EPSS FAQ</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Three access routes; CSV URL; three columns; percentile definition; archive back to 2021-04-14; model version start dates and the score-shift warning</td>
          <td><a href="https://www.first.org/epss/data">FIRST, Get the Data</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>API endpoint, parameters, time-series scope, BETA status</td>
          <td><a href="https://api.first.org/epss/">FIRST API documentation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Sources record exploitation activity, not necessarily success; VulnCheck KEV listed as a data partner</td>
          <td><a href="https://www.first.org/epss/how-it-works">FIRST, How EPSS Works</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Aggregation formulas; 100 vulnerabilities at 0.05 give about 99.4% and an expected count of 5; 90th percentile about 0.04; CVSS High about 0.008</td>
          <td><a href="https://www.first.org/epss/using-epss">FIRST, Using EPSS</a> and our arithmetic (1 − 0.95^100 = 0.9941)</td>
          <td>Verified; FIRST&rsquo;s thresholds are its own rules of thumb</td>
      </tr>
      <tr>
          <td>File header: model v2026.06.15, scored 2026-10-15T12:03:56Z, 389,824 CVEs; all distribution counts and medians</td>
          <td><a href="https://epss.empiricalsecurity.com/epss_scores-current.csv.gz">EPSS daily CSV</a>, downloaded October 16, 2026 (scores dated October 15, 2026)</td>
          <td>Qualified: one day&rsquo;s snapshot, our calculations</td>
      </tr>
      <tr>
          <td>KEV catalog version 2026.10.15, 1,746 entries, all scored; overlap counts</td>
          <td><a href="https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json">CISA KEV JSON</a> joined to the EPSS file by us</td>
          <td>Qualified: snapshot; overlap is not evidence of accuracy</td>
      </tr>
      <tr>
          <td>CVE-2021-44228 scores on 2026-10-15 and 2022-02-28</td>
          <td><a href="https://api.first.org/data/v1/epss?cve=CVE-2021-44228&amp;date=2022-02-28">FIRST EPSS API</a></td>
          <td>Verified on October 16, 2026</td>
      </tr>
      <tr>
          <td>CVE-2024-1992 and CVE-2019-1508 are REJECTED and absent from the file; CVE-2026-14683 is tagged disputed and scored</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2024-1992">CVE Services records</a> and the EPSS CSV</td>
          <td>Verified on October 16, 2026</td>
      </tr>
      <tr>
          <td>Script output</td>
          <td>Run by us on October 16, 2026 against the five-row CSV shown</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Production suggestions</td>
          <td>Our analysis</td>
          <td>Qualified: not tested</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CISA KEV JSON Feed: Every Field and a Scanner Match Script</title><link>https://firewallsync.com/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/</link><guid>https://firewallsync.com/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/</guid><pubDate>Thu, 15 Oct 2026 06:05:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>CISA publishes the Known Exploited Vulnerabilities catalog as a JSON file. Here is every field, what the schema does and does not promise, how due dates and the forensic triage flag behave in the data, and a tested 30-line Python script that joins the feed to a scanner CSV.</description><enclosure url="https://firewallsync.com/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/cover_hu17135229785408898193.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/cover_hu17135229785408898193.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/cover_hu17135229785408898193.jpg" alt="A computer screen photographed at an angle, showing a blue cohort grid labelled user activity by cohort and line graphs with 68K, 16K and 1.7K figures" width="1200" height="630"></p><p>The Known Exploited Vulnerabilities (KEV) catalog is CISA&rsquo;s list of vulnerabilities with evidence of exploitation in the wild. The catalog page is built for people. Teams that want to automate, for example flagging scanner findings that are in KEV, need the machine-readable JSON file and an accurate idea of what each field promises. This guide covers the file&rsquo;s structure, the traps we found in the data, and a script that joins it to a scanner export.</p>
<p><strong>Scope.</strong> This is research-based analysis plus one script we ran. On October 15, 2026 we downloaded the live JSON file, read CISA&rsquo;s KEV page, the published JSON schema and the kev-data repository README, and ran the script below against a five-row test CSV we wrote. All counts below are from that single download and will change as CISA edits the catalog. For what KEV deadlines mean for U.S. federal agencies, see our explainer on <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">CISA BOD 26-04</a>; for when to rank KEV above scores, see <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS vs EPSS vs KEV</a>.</p>
<h2 id="where-to-get-the-file-and-how-fresh-it-is">Where to get the file and how fresh it is</h2>
<p>CISA&rsquo;s <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">KEV catalog page</a> offers CSV and JSON downloads and a JSON schema. The JSON URL we used:</p>
<pre tabindex="0"><code>https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json
</code></pre><p>It downloaded without authentication or an API key using Python&rsquo;s standard library and <code>curl</code>. The file is about 1.8 MB.</p>
<p>The <a href="https://github.com/cisagov/kev-data">cisagov/kev-data repository</a> holds the same data, a CSV and the current <a href="https://github.com/cisagov/kev-data/blob/develop/known_exploited_vulnerabilities_schema.json">schema</a>, and its git history is the only practical change log: the README says the KEV &ldquo;has no inline file revision history or log that&rsquo;s easily accessible after the fact.&rdquo; Per that README the repository updates within minutes of cisa.gov, and updates &ldquo;typically happen weekdays during normal US Eastern business hours when there are new or updated KEV entries.&rdquo;</p>
<p>The top of the file carries four metadata fields. In our download:</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Value on October 15, 2026</th>
          <th>Meaning (per schema)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>catalogVersion</code></td>
          <td>2026.10.14</td>
          <td>Version string of the catalog</td>
      </tr>
      <tr>
          <td><code>dateReleased</code></td>
          <td>2026-10-14T19:27:28.0105Z</td>
          <td>Release time, UTC</td>
      </tr>
      <tr>
          <td><code>count</code></td>
          <td>1745</td>
          <td>Total number of entries</td>
      </tr>
      <tr>
          <td><code>vulnerabilities</code></td>
          <td>array of 1,745 objects</td>
          <td>The entries</td>
      </tr>
  </tbody>
</table>
<p>We checked that <code>count</code>, the array length and the number of distinct <code>cveID</code> values were all 1,745. That is worth asserting in your own pipeline: if they disagree, the download is probably truncated.</p>
<h2 id="the-entry-fields-one-by-one">The entry fields, one by one</h2>
<p>Every entry has 12 fields in our copy (<code>cveID</code>, <code>vendorProject</code>, <code>product</code>, <code>vulnerabilityName</code>, <code>dateAdded</code>, <code>shortDescription</code>, <code>requiredAction</code>, <code>dueDate</code>, <code>knownRansomwareCampaignUse</code>, <code>forensicTriage</code>, <code>notes</code>, <code>cwes</code>). The table gives CISA&rsquo;s schema definition and, where we measured something, what the data shows.</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>Schema definition (paraphrased)</th>
          <th>What to watch for</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>cveID</code></td>
          <td>CVE ID, <code>CVE-YYYY-NNNN</code>; the number part can have more than four digits</td>
          <td>Match on this. Scanners may emit lowercase or padded values, so normalize</td>
      </tr>
      <tr>
          <td><code>vendorProject</code>, <code>product</code></td>
          <td>Vendor/project and product names</td>
          <td>Free text, not a CPE (Common Platform Enumeration, the standard product naming scheme). <code>product</code> can be &ldquo;Multiple Products&rdquo;. Do not match on these</td>
      </tr>
      <tr>
          <td><code>vulnerabilityName</code></td>
          <td>Name of the vulnerability</td>
          <td>Display only</td>
      </tr>
      <tr>
          <td><code>dateAdded</code></td>
          <td>Date added to the catalog, <code>YYYY-MM-DD</code></td>
          <td>The date CISA added it, not the disclosure or first-exploitation date</td>
      </tr>
      <tr>
          <td><code>shortDescription</code></td>
          <td>Short description</td>
          <td>Display only</td>
      </tr>
      <tr>
          <td><code>requiredAction</code></td>
          <td>Required action to address it</td>
          <td>Text varies by entry; recent entries refer to BOD 26-04</td>
      </tr>
      <tr>
          <td><code>dueDate</code></td>
          <td>Date the required action is due, <code>YYYY-MM-DD</code></td>
          <td>See next section</td>
      </tr>
      <tr>
          <td><code>knownRansomwareCampaignUse</code></td>
          <td>&ldquo;Known&rdquo; if leveraged in a ransomware campaign; &ldquo;Unknown&rdquo; if CISA lacks confirmation</td>
          <td>361 Known, 1,384 Unknown in our copy</td>
      </tr>
      <tr>
          <td><code>forensicTriage</code></td>
          <td>&ldquo;Yes&rdquo; if forensic triage is required per BOD 26-04, &ldquo;No&rdquo; if not</td>
          <td>84 Yes, 1,661 No</td>
      </tr>
      <tr>
          <td><code>notes</code></td>
          <td>Additional notes</td>
          <td>Free text; see below</td>
      </tr>
      <tr>
          <td><code>cwes</code></td>
          <td>Array of CWE (Common Weakness Enumeration) IDs, <code>CWE-NNNN</code></td>
          <td>Can be empty: 175 entries have none</td>
      </tr>
  </tbody>
</table>
<p>Source for the definitions: CISA&rsquo;s <a href="https://github.com/cisagov/kev-data/blob/develop/known_exploited_vulnerabilities_schema.json">published schema</a>. The counts are our own tallies of the downloaded file.</p>
<h3 id="notes-is-not-a-single-url"><code>notes</code> is not a single URL</h3>
<p>Many entries put several links in <code>notes</code>, separated by semicolons and sometimes by empty segments. In our copy 963 of 1,745 notes contain a semicolon and 973 contain more than one <code>http</code>. Example from the entry added on October 14, 2026 for Ahsay AhsayCBS (CVE-2026-105134): the field holds a vendor announcement, two CISA BOD 26-04 links and an NVD link in one string. If you want links, split on <code>;</code>, trim, and drop empties.</p>
<h3 id="reading-knownransomwarecampaignuse-correctly">Reading <code>knownRansomwareCampaignUse</code> correctly</h3>
<p>The two values are not symmetric. &ldquo;Known&rdquo; is a positive confirmation; &ldquo;Unknown&rdquo; is the absence of one. Treating &ldquo;Unknown&rdquo; as &ldquo;no ransomware risk&rdquo; is a misreading the schema itself rules out. The count of &ldquo;Known&rdquo; entries by year added is also not a measure of ransomware activity that year, since the flag can be updated after the entry is added; we did not test how often that happens.</p>
<h2 id="duedate-is-not-a-fixed-interval"><code>dueDate</code> is not a fixed interval</h2>
<p>If you assume &ldquo;due date = added + N days&rdquo; you will be wrong for much of the file. We computed <code>dueDate − dateAdded</code> in calendar days for every entry:</p>
<table>
  <thead>
      <tr>
          <th>Year added</th>
          <th>Entries</th>
          <th>Most common gap</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>2021</td>
          <td>311</td>
          <td>181 days (189 entries), 14 days (111)</td>
      </tr>
      <tr>
          <td>2022</td>
          <td>555</td>
          <td>21 days (403)</td>
      </tr>
      <tr>
          <td>2023</td>
          <td>187</td>
          <td>21 days (177)</td>
      </tr>
      <tr>
          <td>2024</td>
          <td>186</td>
          <td>21 days (175)</td>
      </tr>
      <tr>
          <td>2025</td>
          <td>245</td>
          <td>21 days (226)</td>
      </tr>
      <tr>
          <td>2026</td>
          <td>261</td>
          <td>3 days (136), 14 days (78), 21 days (44)</td>
      </tr>
  </tbody>
</table>
<p>Across the whole file, 238 entries have a gap of exactly 181 days. We did not investigate why, and the data does not say.</p>
<p>The 2026 change lines up with BOD 26-04, CISA&rsquo;s directive dated June 10, 2026. Restricting to the 128 entries added on or after June 10, 2026 (the directive&rsquo;s date, our cut-off):</p>
<table>
  <thead>
      <tr>
          <th><code>forensicTriage</code></th>
          <th>Gap to <code>dueDate</code></th>
          <th>Entries</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Yes</td>
          <td>3 days</td>
          <td>84</td>
      </tr>
      <tr>
          <td>No</td>
          <td>3 days</td>
          <td>21</td>
      </tr>
      <tr>
          <td>No</td>
          <td>14 days</td>
          <td>23</td>
      </tr>
  </tbody>
</table>
<p>Three observations, all about this data and not about the directive&rsquo;s text:</p>
<ol>
<li>Every &ldquo;Yes&rdquo; entry in the file falls in this group, and the earliest has <code>dateAdded</code> of 2026-06-25.</li>
<li>Every &ldquo;Yes&rdquo; entry has a three-day gap. The reverse is not true: 21 &ldquo;No&rdquo; entries also have three days.</li>
<li>So <code>forensicTriage</code> and a short deadline are separate signals. A tool that infers one from the other will be wrong for those 21 entries.</li>
</ol>
<p>Whose deadline is it? The entries cite BOD 26-04, which applies to U.S. Federal Civilian Executive Branch agencies. For any other organization the date is context, not an obligation. The directive also sets timelines from asset exposure and other factors that the feed cannot represent; our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04 explainer</a> covers that.</p>
<h2 id="a-tested-join-against-a-scanner-export">A tested join against a scanner export</h2>
<p>The script below reads a CSV with columns <code>host,cve,severity</code>, downloads the feed, and prints each finding with the KEV fields appended, KEV hits first, ordered by <code>dueDate</code>. It uses only the Python standard library.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#75715e">#!/usr/bin/env python3</span>
</span></span><span style="display:flex;"><span><span style="color:#e6db74">&#34;&#34;&#34;Join a scanner CSV (columns: host,cve,severity) to the CISA KEV JSON feed.&#34;&#34;&#34;</span>
</span></span><span style="display:flex;"><span><span style="color:#f92672">import</span> csv<span style="color:#f92672">,</span> json<span style="color:#f92672">,</span> sys<span style="color:#f92672">,</span> urllib.request
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>URL <span style="color:#f92672">=</span> <span style="color:#e6db74">&#34;https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json&#34;</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> urllib<span style="color:#f92672">.</span>request<span style="color:#f92672">.</span>urlopen(URL, timeout<span style="color:#f92672">=</span><span style="color:#ae81ff">30</span>) <span style="color:#66d9ef">as</span> r:
</span></span><span style="display:flex;"><span>    feed <span style="color:#f92672">=</span> json<span style="color:#f92672">.</span>load(r)
</span></span><span style="display:flex;"><span>kev <span style="color:#f92672">=</span> {v[<span style="color:#e6db74">&#34;cveID&#34;</span>]<span style="color:#f92672">.</span>upper(): v <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> feed[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>print(<span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;catalog </span><span style="color:#e6db74">{</span>feed[<span style="color:#e6db74">&#39;catalogVersion&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">, released </span><span style="color:#e6db74">{</span>feed[<span style="color:#e6db74">&#39;dateReleased&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">, &#34;</span>
</span></span><span style="display:flex;"><span>      <span style="color:#e6db74">f</span><span style="color:#e6db74">&#34;count field </span><span style="color:#e6db74">{</span>feed[<span style="color:#e6db74">&#39;count&#39;</span>]<span style="color:#e6db74">}</span><span style="color:#e6db74">, entries </span><span style="color:#e6db74">{</span>len(kev)<span style="color:#e6db74">}</span><span style="color:#e6db74">&#34;</span>, file<span style="color:#f92672">=</span>sys<span style="color:#f92672">.</span>stderr)
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>rows <span style="color:#f92672">=</span> []
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">with</span> open(sys<span style="color:#f92672">.</span>argv[<span style="color:#ae81ff">1</span>], newline<span style="color:#f92672">=</span><span style="color:#e6db74">&#34;&#34;</span>) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">for</span> row <span style="color:#f92672">in</span> csv<span style="color:#f92672">.</span>DictReader(f):
</span></span><span style="display:flex;"><span>        cve <span style="color:#f92672">=</span> row[<span style="color:#e6db74">&#34;cve&#34;</span>]<span style="color:#f92672">.</span>strip()<span style="color:#f92672">.</span>upper()
</span></span><span style="display:flex;"><span>        hit <span style="color:#f92672">=</span> kev<span style="color:#f92672">.</span>get(cve)
</span></span><span style="display:flex;"><span>        rows<span style="color:#f92672">.</span>append({
</span></span><span style="display:flex;"><span>            <span style="color:#f92672">**</span>row, <span style="color:#e6db74">&#34;cve&#34;</span>: cve, <span style="color:#e6db74">&#34;in_kev&#34;</span>: bool(hit),
</span></span><span style="display:flex;"><span>            <span style="color:#e6db74">&#34;dateAdded&#34;</span>: hit[<span style="color:#e6db74">&#34;dateAdded&#34;</span>] <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>,
</span></span><span style="display:flex;"><span>            <span style="color:#e6db74">&#34;dueDate&#34;</span>: hit[<span style="color:#e6db74">&#34;dueDate&#34;</span>] <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>,
</span></span><span style="display:flex;"><span>            <span style="color:#e6db74">&#34;ransomware&#34;</span>: hit[<span style="color:#e6db74">&#34;knownRansomwareCampaignUse&#34;</span>] <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>,
</span></span><span style="display:flex;"><span>            <span style="color:#e6db74">&#34;forensicTriage&#34;</span>: hit[<span style="color:#e6db74">&#34;forensicTriage&#34;</span>] <span style="color:#66d9ef">if</span> hit <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;&#34;</span>,
</span></span><span style="display:flex;"><span>        })
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>rows<span style="color:#f92672">.</span>sort(key<span style="color:#f92672">=</span><span style="color:#66d9ef">lambda</span> r: (<span style="color:#f92672">not</span> r[<span style="color:#e6db74">&#34;in_kev&#34;</span>], r[<span style="color:#e6db74">&#34;dueDate&#34;</span>] <span style="color:#f92672">or</span> <span style="color:#e6db74">&#34;9999&#34;</span>))
</span></span><span style="display:flex;"><span>w <span style="color:#f92672">=</span> csv<span style="color:#f92672">.</span>DictWriter(sys<span style="color:#f92672">.</span>stdout, fieldnames<span style="color:#f92672">=</span>list(rows[<span style="color:#ae81ff">0</span>]))
</span></span><span style="display:flex;"><span>w<span style="color:#f92672">.</span>writeheader(); w<span style="color:#f92672">.</span>writerows(rows)
</span></span></code></pre></div><p>We ran it on October 15, 2026 against this five-row file. Row 4 has a lowercase ID with a trailing space, and the last CVE ID is invented to represent a finding that is not in KEV:</p>
<pre tabindex="0"><code>host,cve,severity
web-01,CVE-2026-105134,High
web-01,CVE-2020-29583,Critical
db-02,CVE-2026-105134,High
db-02,cve-2021-44228 ,Critical
db-02,CVE-2099-0001,Low
</code></pre><p>Output (the <code>catalog ... entries</code> line goes to stderr):</p>
<pre tabindex="0"><code>catalog 2026.10.14, released 2026-10-14T19:27:28.0105Z, count field 1745, entries 1745
host,cve,severity,in_kev,dateAdded,dueDate,ransomware,forensicTriage
db-02,CVE-2021-44228,Critical,True,2021-12-10,2021-12-24,Known,No
web-01,CVE-2020-29583,Critical,True,2021-11-03,2022-05-03,Unknown,No
web-01,CVE-2026-105134,High,True,2026-10-14,2026-10-17,Unknown,Yes
db-02,CVE-2026-105134,High,True,2026-10-14,2026-10-17,Unknown,Yes
db-02,CVE-2099-0001,Low,False,,,,
</code></pre><p>Read this carefully, because it shows the pitfalls above:</p>
<ul>
<li><strong>Sorting by <code>dueDate</code> is not the same as sorting by urgency.</strong> The 2021 and 2022 entries sort first because their due dates are in the past, yet the only &ldquo;Yes&rdquo; triage flags are on the entry added the day before we ran this. Sort by what you actually want, for example <code>forensicTriage</code> first, then <code>dateAdded</code> descending.</li>
<li><strong>Normalization worked.</strong> The lowercase, space-padded <code>cve-2021-44228 </code> matched Log4Shell (Apache Log4j2), which is in KEV with <code>dateAdded</code> 2021-12-10.</li>
<li><strong>A miss is not a clean bill of health.</strong> <code>CVE-2099-0001</code> is simply absent from KEV. KEV only lists vulnerabilities CISA has evidence were exploited; it says nothing about the rest.</li>
<li><strong>The script does not match products.</strong> If your scanner reports a CVE on a product version that a vendor has already patched or backported, the join still flags it. Confirm against the vendor advisory.</li>
</ul>
<h3 id="making-it-production-ready">Making it production-ready</h3>
<p>These are our suggestions; we did not build or test them beyond the script above.</p>
<ol>
<li><strong>Fail loudly on a bad download.</strong> Assert <code>len(vulnerabilities) == count</code> and that the file parses before replacing yesterday&rsquo;s copy.</li>
<li><strong>Pull daily on weekdays</strong> (the kev-data README says updates cluster in US Eastern business hours) and keep each day&rsquo;s file, or use the kev-data git history, so you can diff what was added.</li>
<li><strong>Detect changes beyond additions.</strong> The README mentions &ldquo;new or updated&rdquo; entries, so compare the whole entry, not just the set of CVE IDs.</li>
<li><strong>Keep the flags separate in your ticket system.</strong> Store <code>dateAdded</code>, <code>dueDate</code>, <code>forensicTriage</code> and <code>knownRansomwareCampaignUse</code> as distinct fields rather than a single priority, for the reasons in the table above.</li>
<li><strong>Pair with other data.</strong> KEV answers &ldquo;has this been exploited?&rdquo; For a likelihood estimate on everything else, join the <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed</a> on the same CVE ID. Add the exploitability details in <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">CISA Vulnrichment SSVC values</a> and your own exposure data, and watch the <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">backlog dynamics</a> so KEV items do not get buried.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We tested the script only against a five-row file, not a real scanner export, and only on Python 3 with the standard library. Counts come from one download on October 15, 2026. We did not determine why 238 entries have a 181-day gap, how often fields are edited after addition, or which CISA process sets each <code>dueDate</code>. The CISA catalog page does not define the fields or state an update frequency, so those points rely on the schema file and the kev-data README.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 15, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Catalog version 2026.10.14, release time, count 1,745 equals array length and unique CVE IDs</td>
          <td><a href="https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json">CISA KEV JSON feed</a>, downloaded October 15, 2026</td>
          <td>Verified (snapshot)</td>
      </tr>
      <tr>
          <td>Field definitions, including the Known/Unknown and forensicTriage semantics and the <code>cwes</code> format</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/develop/known_exploited_vulnerabilities_schema.json">CISA KEV JSON schema</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Weekday update pattern; repository synchronised within minutes; no inline revision history</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/develop/README.md">cisagov/kev-data README</a></td>
          <td>Verified; the CISA catalog page itself states no update frequency</td>
      </tr>
      <tr>
          <td>JSON, CSV and schema links on CISA&rsquo;s page</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Counts of Known/Unknown, forensicTriage Yes/No, empty cwes, notes patterns, due-date gap tables, 128 entries since June 10, 2026</td>
          <td>Our calculation on the downloaded JSON</td>
          <td>Qualified: depends on the October 15, 2026 snapshot; June 10, 2026 cut-off is the BOD 26-04 page date</td>
      </tr>
      <tr>
          <td>BOD 26-04 applies to U.S. federal civilian agencies</td>
          <td><a href="https://www.cisa.gov/news-events/directives/bod-26-04-prioritizing-security-updates-based-risk">CISA, BOD 26-04</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Ahsay AhsayCBS CVE-2026-105134 added 2026-10-14, due 2026-10-17, triage Yes</td>
          <td>KEV JSON as above</td>
          <td>Verified (snapshot)</td>
      </tr>
      <tr>
          <td>Script output</td>
          <td>Run by us on October 15, 2026 against the five-row CSV shown</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Production suggestions</td>
          <td>Our analysis</td>
          <td>Qualified: not tested</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>CISA Vulnrichment SSVC: Read CVE Record Values with jq</title><link>https://firewallsync.com/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/</link><guid>https://firewallsync.com/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/</guid><pubDate>Wed, 14 Oct 2026 06:07:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>CISA adds three SSVC fields (Exploitation, Automatable, Technical Impact) to many CVE records. Here is where they sit in the JSON, how to pull them with one jq command, what each value means, and how CISA's published decision table turns them into Track, Track*, Attend or Act.</description><enclosure url="https://firewallsync.com/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/cover_hu5517198160202342200.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/cover_hu5517198160202342200.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/cover_hu5517198160202342200.jpg" alt="A black compass and a silver map-measuring instrument with a dial, resting on an old black-and-white street map" width="1200" height="630"></p><p>If a CVE record has been analyzed by CISA, it contains three small fields that answer a triage question faster than most advisories do: is this being exploited, can attackers automate it, and what do they get if it works. They sit inside the record under CISA&rsquo;s Authorized Data Publisher (ADP) container, and you can read them with one <code>curl</code> and one <code>jq</code> command. This article shows where the fields are, what each value means, what the data cannot tell you, and how CISA&rsquo;s own decision table converts the three values plus one input from you into one of four actions.</p>
<p>We read CISA&rsquo;s Vulnrichment README and flowcharts, the SSVC project&rsquo;s definitions of the decision points, and CISA&rsquo;s published decision tree file, and we ran the commands below against the live CVE Services API on October 14, 2026. This is research-based explanation plus our own command-line checks of public records; we did not scan any systems.</p>
<h2 id="the-short-answer">The short answer</h2>
<ul>
<li><strong>What it is:</strong> CISA&rsquo;s Vulnrichment project enriches public CVE records. For each CVE CISA analyzes it adds three SSVC (Stakeholder-Specific Vulnerability Categorization) decision points: Exploitation, Automatable and Technical Impact. For some it also adds a CWE weakness identifier or a CVSS score.</li>
<li><strong>Where to read it:</strong> the <code>adp</code> array in the CVE record, in the entry whose <code>providerMetadata.shortName</code> is <code>CISA-ADP</code>.</li>
<li><strong>What to do with it:</strong> combine it with your own exposure and asset criticality. CISA&rsquo;s decision table needs a fourth input, which only you can supply.</li>
</ul>
<h2 id="where-the-data-lives">Where the data lives</h2>
<p>A CVE record in the CVE Record Format has a CNA container, written by the organization that assigned the CVE, and zero or more ADP containers, written by other authorized publishers. CISA&rsquo;s README says CISA never overwrites the CNA&rsquo;s data in the CNA container; it writes its additions to its own ADP container.</p>
<p>You can fetch any record from the CVE Services API and filter for CISA&rsquo;s entry. This is the command we ran (it needs <code>curl</code> and <code>jq</code>):</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>curl -s https://cveawg.mitre.org/api/cve/CVE-2024-4947 | jq <span style="color:#e6db74">&#39;[
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  .containers.adp[]?
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  | select(.providerMetadata.shortName == &#34;CISA-ADP&#34;)
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  | .metrics[]?
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  | select(.other.type == &#34;ssvc&#34;)
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  | .other.content
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">  | {version, timestamp, options: (.options | add)}
</span></span></span><span style="display:flex;"><span><span style="color:#e6db74">]&#39;</span>
</span></span></code></pre></div><p>On October 14, 2026 it returned, for CVE-2024-4947:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>[
</span></span><span style="display:flex;"><span>  {
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#e6db74">&#34;2.0.3&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2024-05-23T20:01:07.846921Z&#34;</span>,
</span></span><span style="display:flex;"><span>    <span style="color:#f92672">&#34;options&#34;</span>: {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;Exploitation&#34;</span>: <span style="color:#e6db74">&#34;active&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;Automatable&#34;</span>: <span style="color:#e6db74">&#34;no&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;Technical Impact&#34;</span>: <span style="color:#e6db74">&#34;total&#34;</span>
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>  }
</span></span><span style="display:flex;"><span>]
</span></span></code></pre></div><p>If CISA has not analyzed a CVE, the command returns <code>[]</code>. Replace <code>&quot;ssvc&quot;</code> with <code>&quot;kev&quot;</code> and <code>.other.content</code> with <code>.other.content.dateAdded</code> to read the date CISA&rsquo;s ADP container recorded the CVE being added to the Known Exploited Vulnerabilities (KEV) catalog; for CVE-2024-4947 that was <code>2024-05-20</code>.</p>
<p>Two fields deserve attention:</p>
<ul>
<li><strong><code>timestamp</code></strong> is when CISA produced that SSVC result. The record above is from May 2024. Exploitation can change after that date.</li>
<li><strong><code>version</code></strong> follows a major.minor.patch pattern. CISA&rsquo;s README says major.minor is the SSVC version and the patch number is the version of CISA&rsquo;s decision tree, so <code>2.0.3</code> means SSVC 2.0 and CISA Coordinator tree version 3. The README asks consumers to check this before validating or processing the JSON.</li>
</ul>
<h2 id="what-the-three-values-mean">What the three values mean</h2>
<table>
  <thead>
      <tr>
          <th>Decision point</th>
          <th>Values CISA publishes</th>
          <th>What it means</th>
          <th>Source definition</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Exploitation</td>
          <td><code>none</code>, <code>poc</code>, <code>active</code></td>
          <td>The state of exploitation at analysis time: no evidence and no public proof of concept (PoC); a public PoC or well-known exploitation method; or shared, observable, reliable evidence of exploitation in the wild</td>
          <td>SSVC project, Exploitation 1.1.0</td>
      </tr>
      <tr>
          <td>Automatable</td>
          <td><code>no</code>, <code>yes</code></td>
          <td><code>yes</code> means attackers can reliably automate steps 1 to 4 of the kill chain: reconnaissance, weaponization, delivery and exploitation</td>
          <td>SSVC project, Automatable 2.0.0</td>
      </tr>
      <tr>
          <td>Technical Impact</td>
          <td><code>partial</code>, <code>total</code></td>
          <td><code>total</code> means the attacker gains total control over the behavior of the affected software, or full disclosure of all information on the affected system; <code>partial</code> means limited influence or visibility</td>
          <td>SSVC project, Technical Impact 1.0.0</td>
      </tr>
  </tbody>
</table>
<p>Some details change how you should read them:</p>
<ul>
<li><strong>Exploitation is present-tense, not predictive.</strong> The SSVC documentation says it measures current exploitation, not predicted future exploitation, and points to predictive systems such as <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS</a> for that. A <code>none</code> value means no exploitation evidence at the time of analysis, not that exploitation will not happen.</li>
<li><strong>The definitions of <code>poc</code> differ between versions.</strong> The SSVC project says Exploitation version 1.0.0 also counted privately attested exploitation and widespread hearsay under <code>poc</code>; version 1.1.0 limits it to public PoCs and well-known methods. CISA&rsquo;s Coordinator tree file we read (v2.0.3) still carries the broader 1.0.0-style wording. Treat <code>poc</code> as &ldquo;exploit knowledge exists&rdquo;, not as a measured exploit-development effort.</li>
<li><strong>Automatable defaults to yes.</strong> The SSVC documentation says that if nothing is known, assume Yes, and that remote code execution or command injection should generally be scored Yes. Reasons for No include a component that is not discoverable on the network, weaponization that needs manual work per target, delivery blocked by common network controls, or exploit-prevention features that make exploitation unreliable.</li>
<li><strong>Technical Impact is relative to the vulnerable component.</strong> The SSVC documentation says control is judged relative to the affected component where the flaw resides. It does not describe how much of your network the attacker then reaches.</li>
</ul>
<h3 id="how-cisa-picks-which-cves-get-more-data">How CISA picks which CVEs get more data</h3>
<p>CISA&rsquo;s README says that CVEs rated Total Technical Impact, Automatable, or with Exploitation <code>poc</code> or <code>active</code> get further analysis for a CWE identifier and a CVSS score, where there is enough information, and that CISA may provide those even for lower-rated CVEs. The README also says that if the CNA later provides its own CWE, CVSS or CPE data, CISA removes its own, and if both exist the CNA&rsquo;s data takes precedence. As of December 10, 2024, CISA no longer adds CPE strings.</p>
<h2 id="four-records-four-profiles">Four records, four profiles</h2>
<p>We pulled these records from the CVE Services API on October 14, 2026. The three values are as CISA recorded them at the timestamp shown.</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>CISA SSVC (Exploitation / Automatable / Technical Impact)</th>
          <th>CISA analysis timestamp</th>
          <th>KEV block in the record</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2024-4947</td>
          <td>active / no / total</td>
          <td>2024-05-23</td>
          <td>Yes, added 2024-05-20</td>
      </tr>
      <tr>
          <td>CVE-2024-25522</td>
          <td>poc / yes / total</td>
          <td>2024-05-24</td>
          <td>No</td>
      </tr>
      <tr>
          <td>CVE-2024-34974</td>
          <td>none / yes / partial</td>
          <td>2024-05-10</td>
          <td>No</td>
      </tr>
      <tr>
          <td>CVE-2021-44228 (Log4Shell)</td>
          <td>active / yes / total</td>
          <td>2025-02-04</td>
          <td>Yes, added 2021-12-10</td>
      </tr>
  </tbody>
</table>
<p>CVE-2024-25522 shows why you need more than one field. CISA&rsquo;s ADP container also carries a CVSS v3.1 base score of 9.4 for it (vector <code>CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:L</code>), yet its SSVC profile is only <code>poc</code>, with no exploitation evidence. A near-critical severity score and an unexploited status can coexist; they answer different questions. Our companion article on <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS, EPSS and KEV</a> covers how to combine them.</p>
<p>For records published in September 2026, the same command returned <code>active / yes / total</code> for CVE-2026-76504 (KEV block dated 2026-09-30, and covered in our <a href="/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/">Cisco SD-WAN Manager article</a>) and for CVE-2026-88771 (KEV block dated 2026-09-27, covered in our <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">NetScaler article</a>).</p>
<h2 id="from-three-values-to-a-decision">From three values to a decision</h2>
<p>CISA&rsquo;s SSVC page says its tree uses five decision points (exploitation status, technical impact, automatable, mission prevalence and public well-being impact) and ends in one of four outcomes. It describes them as follows:</p>
<table>
  <thead>
      <tr>
          <th>Outcome</th>
          <th>CISA&rsquo;s description</th>
          <th>Remediation timing CISA says to apply</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Track</td>
          <td>No action required at this time; keep monitoring and reassess if new information emerges</td>
          <td>Standard update timelines</td>
      </tr>
      <tr>
          <td>Track*</td>
          <td>Specific characteristics that may require closer monitoring for changes</td>
          <td>Standard update timelines</td>
      </tr>
      <tr>
          <td>Attend</td>
          <td>Needs supervisory-level attention, which may include requesting help or information and publishing an internal or external notification</td>
          <td>Sooner than standard timelines</td>
      </tr>
      <tr>
          <td>Act</td>
          <td>Requires attention from the organization&rsquo;s internal, supervisory-level and leadership-level individuals; leadership coordinates the response</td>
          <td>As soon as possible</td>
      </tr>
  </tbody>
</table>
<p>Vulnrichment fills in three of the five. The other two, Mission Prevalence and Public Well-being Impact, are combined in CISA&rsquo;s tree file into a single input called Mission and Well-being, with three values:</p>
<ul>
<li><strong>Low:</strong> Mission Prevalence is Minimal and Public Well-being Impact is Minimal.</li>
<li><strong>Medium:</strong> Mission Prevalence is Support and Public Well-being is Minimal or Material, or Mission Prevalence is Minimal or Support and Public Well-being is Material.</li>
<li><strong>High:</strong> Mission Prevalence is Essential, or Public Well-being Impact is Irreversible.</li>
</ul>
<p>Our reading, not stated in the README, is that these two are left out of the public record because they depend on who you are and what the component does in your environment, which a public CVE record cannot know.</p>
<p>CISA&rsquo;s tree file lists all 36 combinations. The table below is the full result for the three Vulnrichment values and each Mission and Well-being level. We extracted it directly from the file (<code>CISA-Coordinator.json</code>, title &ldquo;CISA Coordinator v2.0.3&rdquo;) and checked that each of the 12 combinations has exactly one outcome per level.</p>
<table>
  <thead>
      <tr>
          <th>Exploitation</th>
          <th>Automatable</th>
          <th>Technical Impact</th>
          <th>Mission and Well-being: low</th>
          <th>medium</th>
          <th>high</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>none</td>
          <td>no</td>
          <td>partial</td>
          <td>Track</td>
          <td>Track</td>
          <td>Track</td>
      </tr>
      <tr>
          <td>none</td>
          <td>no</td>
          <td>total</td>
          <td>Track</td>
          <td>Track</td>
          <td>Track*</td>
      </tr>
      <tr>
          <td>none</td>
          <td>yes</td>
          <td>partial</td>
          <td>Track</td>
          <td>Track</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>none</td>
          <td>yes</td>
          <td>total</td>
          <td>Track</td>
          <td>Track</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>poc</td>
          <td>no</td>
          <td>partial</td>
          <td>Track</td>
          <td>Track</td>
          <td>Track*</td>
      </tr>
      <tr>
          <td>poc</td>
          <td>no</td>
          <td>total</td>
          <td>Track</td>
          <td>Track*</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>poc</td>
          <td>yes</td>
          <td>partial</td>
          <td>Track</td>
          <td>Track</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>poc</td>
          <td>yes</td>
          <td>total</td>
          <td>Track</td>
          <td>Track*</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>active</td>
          <td>no</td>
          <td>partial</td>
          <td>Track</td>
          <td>Track</td>
          <td>Attend</td>
      </tr>
      <tr>
          <td>active</td>
          <td>no</td>
          <td>total</td>
          <td>Track</td>
          <td>Attend</td>
          <td>Act</td>
      </tr>
      <tr>
          <td>active</td>
          <td>yes</td>
          <td>partial</td>
          <td>Attend</td>
          <td>Attend</td>
          <td>Act</td>
      </tr>
      <tr>
          <td>active</td>
          <td>yes</td>
          <td>total</td>
          <td>Attend</td>
          <td>Act</td>
          <td>Act</td>
      </tr>
  </tbody>
</table>
<p>Applying it to the records above:</p>
<ul>
<li><strong>CVE-2024-4947 (active / no / total):</strong> Track if the affected software is of minimal mission relevance to you, Attend at medium, Act at high.</li>
<li><strong>CVE-2024-25522 (poc / yes / total):</strong> Track at low, Track* at medium, Attend at high. Despite the 9.4 CVSS score, the tree does not send it to Act at any level.</li>
<li><strong>CVE-2024-34974 (none / yes / partial):</strong> Track at low and medium, Attend at high.</li>
<li><strong>CVE-2021-44228 (active / yes / total):</strong> Attend at low, Act at medium and high.</li>
</ul>
<p>Two cautions on this table. First, it uses a data file from the CERT/CC SSVC repository&rsquo;s main branch, which the Vulnrichment README links as the tree CISA currently uses; a newer tree version would change both the version string in records and possibly the table. Second, &ldquo;Track&rdquo; in the tree is not &ldquo;never patch&rdquo;: CISA&rsquo;s description says remediation still follows standard update timelines.</p>
<h2 id="how-to-use-this-without-over-trusting-it">How to use this without over-trusting it</h2>
<ol>
<li><strong>Read the timestamp first.</strong> A May 2024 <code>none</code> for Exploitation says nothing about October 2026. Cross-check the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV feed</a> and the vendor advisory.</li>
<li><strong>Treat a missing block as missing data.</strong> CISA&rsquo;s README describes Vulnrichment as covering new and recent CVEs that CISA assesses. No <code>CISA-ADP</code> entry means no CISA SSVC result, not low risk.</li>
<li><strong>Decide your Mission and Well-being input per asset class, once.</strong> Write down which systems are Minimal, Support or Essential for your organization, so triage does not reopen the question for every CVE. That classification is yours; CISA&rsquo;s data cannot supply it.</li>
<li><strong>Keep your own exposure facts separate.</strong> The tree does not know whether the vulnerable service is reachable from the internet in your environment. An internet-facing instance can justify a faster timeline than the table&rsquo;s output.</li>
<li><strong>Do not confuse SSVC outcomes with legal deadlines.</strong> Our <a href="/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/">BOD 26-04 explainer</a> covers the KEV deadlines that bind federal agencies. SSVC&rsquo;s Track, Attend and Act are decision aids, not deadlines.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<ul>
<li>We ran read-only queries against public records and read documentation; we did not test any automation at scale or measure how complete Vulnrichment coverage is.</li>
<li>The records in the tables are examples chosen to show different profiles, not a statistical sample.</li>
<li>Values in live records can change; the numbers above are as of October 14, 2026.</li>
<li>We did not read CISA&rsquo;s separate SSVC Guide PDF, so the tree logic is taken from the published JSON file rather than from CISA&rsquo;s prose explanation.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Vulnrichment adds SSVC decision points via CISA&rsquo;s ADP container; never overwrites CNA data; further CWE/CVSS analysis rules; CNA data takes precedence; CPE additions stopped December 10, 2024; version string meaning</td>
          <td><a href="https://github.com/cisagov/vulnrichment/blob/develop/README.md">CISA Vulnrichment README</a></td>
          <td>Verified: read October 14, 2026</td>
      </tr>
      <tr>
          <td>Location and shape of SSVC and KEV data in CVE records; commands and outputs</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2024-4947">CVE Services API</a> (example record), run by us</td>
          <td>Verified: live output October 14, 2026; values may change</td>
      </tr>
      <tr>
          <td>Exploitation values and definitions; version difference; not predictive</td>
          <td><a href="https://certcc.github.io/SSVC/reference/decision_points/exploitation/">SSVC project: Exploitation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Automatable definition, kill chain steps, default to Yes</td>
          <td><a href="https://certcc.github.io/SSVC/reference/decision_points/automatable/">SSVC project: Automatable</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Technical Impact definition and values</td>
          <td><a href="https://certcc.github.io/SSVC/reference/decision_points/technical_impact/">SSVC project: Technical Impact</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA tree uses five decision points and four outcomes; outcome descriptions and timelines</td>
          <td><a href="https://www.cisa.gov/stakeholder-specific-vulnerability-categorization-ssvc">CISA SSVC page</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>36-row decision table and Mission and Well-being definitions</td>
          <td><a href="https://github.com/CERTCC/SSVC/blob/main/docs/ssvc-calc/CISA-Coordinator.json">CISA-Coordinator.json, CERT/CC SSVC repository (main branch)</a></td>
          <td>Qualified: main branch file read October 14, 2026; table computed by us from it; tree may change</td>
      </tr>
      <tr>
          <td>Why Mission and Well-being is absent from public records</td>
          <td>Our reading</td>
          <td>Qualified: editorial inference, labelled as such</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>VEX Explained: What a "Not Affected" Statement Means and How to Read One</title><link>https://firewallsync.com/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/</link><guid>https://firewallsync.com/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/</guid><pubDate>Tue, 13 Oct 2026 06:05:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>AppSec &amp; APIs</category><description>A VEX document is a supplier's signed-off claim about whether a specific product is affected by a specific vulnerability. Here is what the four statuses and five justifications mean in CISA's guidance, how OpenVEX and CSAF express them, and why a VEX claim is something to evaluate, not obey.</description><enclosure url="https://firewallsync.com/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/cover_hu12796011277069913728.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/cover_hu12796011277069913728.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/vex-explained-what-a-not-affected-statement-means-and-how-to-read-one/cover_hu12796011277069913728.jpg" alt="A black magnifying glass, a black spiral notebook with a blank white card, a pencil and an eraser on a white desk" width="1200" height="630"></p><p><strong>VEX</strong>, short for Vulnerability Exploitability eXchange, is a way for a supplier (or another party) to say, in a structured document, whether a specific product is affected by a specific vulnerability. A VEX statement about a library you find in your scanner output answers the question the scanner cannot: &ldquo;this component is present, but can the flaw actually be reached?&rdquo; This article explains the vocabulary CISA defined, shows how two formats (OpenVEX and CSAF) express it, and sets out how to evaluate a &ldquo;not affected&rdquo; claim before you let it close a ticket.</p>
<p>We read CISA&rsquo;s June 2022 status-justification guidance, the OpenVEX specification, the CSAF 2.0 standard and parts of Trivy&rsquo;s documentation on October 13, 2026. We did not write or consume a VEX document in a live pipeline for this article, so it is research-based explanation, not hands-on testing.</p>
<h2 id="the-short-answer">The short answer</h2>
<ul>
<li>A VEX statement ties <strong>one product</strong> to <strong>one vulnerability</strong> and gives a <strong>status</strong>.</li>
<li>CISA&rsquo;s guidance says the status must be one of <strong>NOT AFFECTED, AFFECTED, FIXED or UNDER INVESTIGATION</strong>.</li>
<li>A NOT AFFECTED status can carry a <strong>justification</strong> from a fixed list of five, so a tool can act on it.</li>
<li>It is a claim by whoever wrote the document. CISA says the consumer &ldquo;may choose to accept it or not.&rdquo;</li>
</ul>
<h2 id="what-a-vex-document-must-contain">What a VEX document must contain</h2>
<p>CISA&rsquo;s guidance states that VEX documents must contain four things: VEX metadata, product details, vulnerability details and product status. CISA also published a separate &ldquo;Minimum Requirements for VEX&rdquo; document (the CISA page is dated April 21, 2023). We could not download that PDF (CISA&rsquo;s server returned &ldquo;Access Denied&rdquo; to our request), so we rely on the OpenVEX specification, which says it aims to meet those minimum requirements, and on the CISA status-justification guidance quoted above.</p>
<h2 id="the-four-statuses">The four statuses</h2>
<table>
  <thead>
      <tr>
          <th>Status (CISA name)</th>
          <th>Meaning</th>
          <th>OpenVEX value</th>
          <th>CSAF 2.0 field under <code>product_status</code></th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>NOT AFFECTED</td>
          <td>No remediation is required for this vulnerability in this product</td>
          <td><code>not_affected</code></td>
          <td><code>known_not_affected</code></td>
      </tr>
      <tr>
          <td>AFFECTED</td>
          <td>The product is affected; remediation or mitigation is recommended</td>
          <td><code>affected</code></td>
          <td><code>known_affected</code></td>
      </tr>
      <tr>
          <td>FIXED</td>
          <td>These product versions contain a fix</td>
          <td><code>fixed</code></td>
          <td><code>fixed</code></td>
      </tr>
      <tr>
          <td>UNDER INVESTIGATION</td>
          <td>Impact is not yet known</td>
          <td><code>under_investigation</code></td>
          <td><code>under_investigation</code></td>
      </tr>
  </tbody>
</table>
<p>Sources: <a href="https://www.cisa.gov/sites/default/files/2023-01/VEX_Status_Justification_Jun22.pdf">CISA, VEX Status Justifications (June 2022)</a>; <a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification</a>; <a href="https://docs.oasis-open.org/csaf/csaf/v2.0/os/csaf-v2.0-os.html">CSAF 2.0 standard</a>. The column pairing is our mapping of equivalent concepts; each source defines its own values. CSAF defines further categories (for example <code>first_affected</code> and <code>last_affected</code>) that have no direct CISA counterpart.</p>
<p>The OpenVEX specification uses a hypothetical timeline to show why the statuses exist. A vendor learns of a new CVE and publishes <code>under_investigation</code>; later it finds the product affected and publishes <code>affected</code> with mitigation advice; when it ships a patch it publishes <code>fixed</code>. Without that sequence, the spec notes, a scanner user &ldquo;will simply get a third party alert with no details on how the status is evolving,&rdquo; and the alert becomes a false positive once the product is patched.</p>
<h2 id="the-five-not-affected-justifications">The five &ldquo;not affected&rdquo; justifications</h2>
<p>A justification is the machine-readable reason behind a NOT AFFECTED claim. CISA lists five. The definitions below paraphrase CISA&rsquo;s June 2022 document; the examples in the last column are CISA&rsquo;s.</p>
<table>
  <thead>
      <tr>
          <th>Justification</th>
          <th>What the supplier is asserting</th>
          <th>CISA example</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>component_not_present</code></td>
          <td>The vulnerable component is not in the product</td>
          <td>A Python and Elixir product with no Java code cannot be affected by Log4j</td>
      </tr>
      <tr>
          <td><code>vulnerable_code_not_present</code></td>
          <td>The component is present, but the specific vulnerable code is not</td>
          <td>The compiler stripped the relevant function, or the vulnerable code was added in a later version than the one shipped</td>
      </tr>
      <tr>
          <td><code>vulnerable_code_not_in_execute_path</code></td>
          <td>The vulnerable code is present but can never run in this product</td>
          <td>An application that uses OpenSSL only for random-number generation is not exposed to Heartbleed, which was never called from that function</td>
      </tr>
      <tr>
          <td><code>vulnerable_code_cannot_be_controlled_by_adversary</code></td>
          <td>The code runs, but an attacker cannot influence it enough to trigger the flaw without another vulnerability</td>
          <td>A hard-coded variable denies user input to the vulnerable function</td>
      </tr>
      <tr>
          <td><code>inline_mitigations_already_exist</code></td>
          <td>Built-in protections that an attacker or user cannot bypass or disable prevent exploitation</td>
          <td>Input is sanitized further upstream within the product</td>
      </tr>
  </tbody>
</table>
<p>Two caveats from CISA&rsquo;s own text are easy to miss:</p>
<ol>
<li>For <code>vulnerable_code_not_in_execute_path</code>, CISA warns that &ldquo;a single path of execution should not be assumed since the attacker may be able to divert the path of execution.&rdquo;</li>
<li>For <code>inline_mitigations_already_exist</code>, CISA notes that mitigations are effective &ldquo;only in the scenario where the threat model of the software allows it,&rdquo; and a mitigation that works in one product may not in another. The OpenVEX specification adds that this justification &ldquo;could be difficult to prove conclusively.&rdquo;</li>
</ol>
<p>CISA also distinguishes the last two. With <code>vulnerable_code_cannot_be_controlled_by_adversary</code>, nothing the attacker does can reach the vulnerable code in a way that triggers it. With <code>inline_mitigations_already_exist</code>, the attacker can influence execution, but another control blocks the attack. Both, CISA says, &ldquo;rely on an expert analysis of the structure of the code.&rdquo;</p>
<h2 id="how-the-formats-express-it">How the formats express it</h2>
<h3 id="openvex">OpenVEX</h3>
<p>OpenVEX is a JSON format. Per its specification, a document requires <code>@context</code>, <code>@id</code>, <code>author</code>, <code>timestamp</code>, <code>version</code> and <code>statements</code>. Each statement needs a vulnerability, a status and at least one product (these can be inherited from the document in some cases, and the specification says a document with incomplete statements is not valid). For <code>not_affected</code>, the statement must include either a justification or an <code>impact_statement</code>, which is free text. The specification says free-text impact statements are &ldquo;highly discouraged for automated systems.&rdquo; For <code>affected</code>, it must include an <code>action_statement</code> describing how to remediate or mitigate.</p>
<p>A minimal not-affected statement looks like this. Everything below is an invented placeholder: the CVE number, the product identifier and the author are not real.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-json" data-lang="json"><span style="display:flex;"><span>{
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@context&#34;</span>: <span style="color:#e6db74">&#34;https://openvex.dev/ns/v0.2.0&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;https://example.com/vex/2026-0001&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;author&#34;</span>: <span style="color:#e6db74">&#34;Example Product Security Team&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;timestamp&#34;</span>: <span style="color:#e6db74">&#34;2026-10-13T06:00:00Z&#34;</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;version&#34;</span>: <span style="color:#ae81ff">1</span>,
</span></span><span style="display:flex;"><span>  <span style="color:#f92672">&#34;statements&#34;</span>: [
</span></span><span style="display:flex;"><span>    {
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;vulnerability&#34;</span>: { <span style="color:#f92672">&#34;name&#34;</span>: <span style="color:#e6db74">&#34;CVE-2026-00000&#34;</span> },
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;products&#34;</span>: [{ <span style="color:#f92672">&#34;@id&#34;</span>: <span style="color:#e6db74">&#34;pkg:oci/example-app@1.4.2&#34;</span> }],
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;status&#34;</span>: <span style="color:#e6db74">&#34;not_affected&#34;</span>,
</span></span><span style="display:flex;"><span>      <span style="color:#f92672">&#34;justification&#34;</span>: <span style="color:#e6db74">&#34;vulnerable_code_not_in_execute_path&#34;</span>
</span></span><span style="display:flex;"><span>    }
</span></span><span style="display:flex;"><span>  ]
</span></span><span style="display:flex;"><span>}
</span></span></code></pre></div><p>The <code>version</code> field, the specification says, must be incremented when any content of the document changes, which is how a supplier moves from <code>under_investigation</code> to <code>fixed</code> over time.</p>
<h3 id="csaf-20">CSAF 2.0</h3>
<p>CSAF (Common Security Advisory Framework) is an OASIS Standard, approved November 18, 2022, with an errata release dated January 26, 2024 that changes only an aggregator schema. Its <strong>Profile 5: VEX</strong> requires a product tree, a list of vulnerabilities, and at least one of the <code>fixed</code>, <code>known_affected</code>, <code>known_not_affected</code> or <code>under_investigation</code> product-status lists. It also sets two content rules:</p>
<ul>
<li>For each product in <code>known_not_affected</code>, an impact statement must exist, either as a machine-readable <strong>flag</strong> or as a human-readable <code>impact</code> entry in <code>threats</code>.</li>
<li>For each product in <code>known_affected</code>, product-specific action information must be given in <code>remediations</code>; the categories <code>no_fix_planned</code> and <code>none_available</code> are permitted.</li>
</ul>
<p>CSAF&rsquo;s flag labels are the same five justification values as CISA&rsquo;s list.</p>
<h2 id="reading-a-vex-statement-critically">Reading a VEX statement critically</h2>
<p>A scanner that supports VEX can hide findings based on these documents. That is the point, and also the risk. These are our practical checks, not requirements from CISA, OpenVEX or OASIS:</p>
<ol>
<li><strong>Match the product exactly.</strong> Check that the product identifier and version in the statement are the ones you deploy, not a sibling image or a different build.</li>
<li><strong>Check who signed it.</strong> OpenVEX says the author&rsquo;s identity &ldquo;SHOULD be cryptographically associated with the signature of the VEX document.&rdquo; An unsigned file fetched from an arbitrary location is a weaker basis than a signed statement from the supplier.</li>
<li><strong>Prefer machine-readable justifications.</strong> A bare <code>not_affected</code> with free text gives your tooling, and a future auditor, little to evaluate.</li>
<li><strong>Weigh the justification type.</strong> <code>component_not_present</code> is something you can verify against your own SBOM. <code>inline_mitigations_already_exist</code> relies on the supplier&rsquo;s threat model, which is hard to check from outside.</li>
<li><strong>Treat known-exploited vulnerabilities with extra care.</strong> If a vulnerability is on CISA&rsquo;s Known Exploited Vulnerabilities list, a wrong &ldquo;not affected&rdquo; is expensive. Our article on <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS, EPSS and KEV</a> covers how to use that signal.</li>
<li><strong>Keep the audit trail.</strong> Record which VEX document suppressed which finding, and re-evaluate when the document&rsquo;s version changes.</li>
</ol>
<p>CISA&rsquo;s guidance also notes that VEX statements cover <em>known</em> vulnerabilities and that attackers often chain flaws; new vulnerabilities may emerge that do affect a product. A &ldquo;not affected&rdquo; for one CVE says nothing about the next.</p>
<h2 id="using-vex-with-scanners">Using VEX with scanners</h2>
<p>Tool support varies. As one example, Trivy&rsquo;s documentation says VEX can be used &ldquo;to filter vulnerabilities based on their status,&rdquo; and shows the <code>--vex</code> option with four sources: a VEX repository, a local file, a VEX attestation stored in an OCI registry, or VEX files referenced in an SBOM. That page was marked &ldquo;EXPERIMENTAL&rdquo; when we read it and warned that the feature &ldquo;might change without preserving backwards compatibility.&rdquo; We did not test Trivy or any other scanner&rsquo;s VEX handling, and we did not check which formats each scanner accepts, so confirm support for your chosen format and version before relying on it.</p>
<h2 id="where-vex-fits">Where VEX fits</h2>
<ul>
<li><strong>With an SBOM:</strong> the SBOM lists components; VEX says which vulnerabilities in them matter. See our article on the <a href="/posts/sbom-minimum-elements-2026-what-cisa-and-partners-changed-from-the-2021-ntia-baseline/">2026 SBOM minimum elements</a>.</li>
<li><strong>With exploitation signals:</strong> VEX says whether a product is affected, not whether a flaw is being exploited. For that, check the <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">KEV feed</a> and <a href="/posts/cisa-vulnrichment-how-to-read-ssvc-values-in-a-cve-record-and-turn-them-into-a-patch-decision/">CISA&rsquo;s Vulnrichment SSVC values</a>.</li>
<li><strong>With provenance:</strong> signing and provenance tell you who built an artifact and how, not whether a given CVE affects it. See <a href="/posts/slsa-build-levels-and-provenance-attestations-what-they-prove-and-what-they-dont/">SLSA build levels and provenance attestations</a>.</li>
<li><strong>With image scanning:</strong> VEX is one way to reduce noise from scanners; see <a href="/posts/container-image-scanning-what-ci-pipelines-still-miss/">what container image scanning still misses</a>.</li>
</ul>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We could not read CISA&rsquo;s &ldquo;Minimum Requirements for VEX&rdquo; PDF (access denied). The CISA status-justification document is from June 2022 and its own text says it is not exhaustive and that CISA may add justifications later; we did not look for later revisions. We did not review the CycloneDX VEX specification, or the details of any scanner&rsquo;s VEX support beyond the Trivy page above. The OpenVEX specification page is titled v0.2.0 but also contains an older revision-table entry using a different version number, so check the version your tools implement.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 13, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>VEX documents must contain metadata, product details, vulnerability details and product status; four statuses; five not-affected justifications; statements are the author&rsquo;s assertions the consumer may accept or not</td>
          <td><a href="https://www.cisa.gov/sites/default/files/2023-01/VEX_Status_Justification_Jun22.pdf">CISA, VEX Status Justifications, June 2022</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Definitions and examples for each justification; the caveats on execution path and inline mitigations</td>
          <td>Same document, sections 3.3 to 3.7</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA published &ldquo;Minimum Requirements for VEX,&rdquo; page dated April 21, 2023</td>
          <td><a href="https://www.cisa.gov/resources-tools/resources/minimum-requirements-vulnerability-exploitability-exchange-vex">CISA resource page</a></td>
          <td>Qualified: PDF itself could not be downloaded</td>
      </tr>
      <tr>
          <td>OpenVEX required fields; justification or impact statement required for <code>not_affected</code>; free text discouraged for automation; <code>action_statement</code> for <code>affected</code>; author signature recommendation; version incrementing</td>
          <td><a href="https://github.com/openvex/spec/blob/main/OPENVEX-SPEC.md">OpenVEX specification</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>OpenVEX says it aims to meet CISA&rsquo;s April 2023 minimum requirements</td>
          <td>Same specification, overview</td>
          <td>Qualified: the OpenVEX project&rsquo;s own claim</td>
      </tr>
      <tr>
          <td>CSAF 2.0 is an OASIS Standard dated November 18, 2022; errata dated January 26, 2024</td>
          <td><a href="https://docs.oasis-open.org/csaf/csaf/v2.0/csaf-v2.0.html">CSAF 2.0 errata 01</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CSAF Profile 5 requirements; <code>product_status</code> categories; flag labels</td>
          <td><a href="https://docs.oasis-open.org/csaf/csaf/v2.0/os/csaf-v2.0-os.html">CSAF 2.0 standard</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Trivy <code>--vex</code> options; page marked experimental</td>
          <td><a href="https://trivy.dev/latest/docs/supply-chain/vex/">Trivy VEX documentation</a></td>
          <td>Qualified: read October 13, 2026; may change</td>
      </tr>
      <tr>
          <td>Status mapping table; reading checklist</td>
          <td>Our analysis</td>
          <td>Qualified: editorial judgment, labelled as such</td>
      </tr>
  </tbody>
</table>
<h2 id="correction">Correction</h2>
<p><em>October 13, 2026:</em> An earlier version of this article said OpenVEX meets CISA&rsquo;s minimum requirements for VEX. The OpenVEX specification says it <em>aims to meet</em> them. The wording has been corrected in three places.</p>
]]></content:encoded></item><item><title>CISA BOD 26-04 Explained: KEV Deadlines, Forensic Triage and What Private Teams Can Borrow</title><link>https://firewallsync.com/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/</link><guid>https://firewallsync.com/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/</guid><pubDate>Mon, 12 Oct 2026 06:15:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>CISA's Binding Operational Directive 26-04, issued June 10, 2026, replaces fixed patch deadlines for U.S. federal agencies with risk-based timelines and a three-day forensic triage requirement. A batch of five years-old flaws added to the KEV catalog on October 8, 2026 shows it in action.</description><enclosure url="https://firewallsync.com/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/cover_hu14614838678526543170.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/cover_hu14614838678526543170.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cisa-bod-26-04-explained-kev-deadlines-forensic-triage-and-what-private-teams-can-borrow/cover_hu14614838678526543170.jpg" alt="A small wooden hourglass with blue sand standing at a tilt on pebbles" width="1200" height="630"></p><p>If you work for a U.S. federal civilian agency, <strong>Binding Operational Directive 26-04</strong> (BOD 26-04) sets how fast you must fix known exploited vulnerabilities. If you do not, it still matters, because CISA&rsquo;s Known Exploited Vulnerabilities (KEV) catalog entries now carry its deadlines and its forensic-triage language, and many private teams use KEV as a patching signal. This article explains what the directive says, what we could and could not verify, and which parts transfer to organizations it does not bind.</p>
<p>We read CISA&rsquo;s directive page, CISA&rsquo;s implementation guidance page, the KEV catalog and National Vulnerability Database (NVD) records on October 12, 2026. This is research-based analysis; we did not test any system.</p>
<h2 id="what-bod-26-04-is-and-who-it-covers">What BOD 26-04 is and who it covers</h2>
<table>
  <thead>
      <tr>
          <th>Item</th>
          <th>What CISA&rsquo;s page says</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Title</td>
          <td>Prioritizing Security Updates Based on Risk</td>
      </tr>
      <tr>
          <td>Page date</td>
          <td>June 10, 2026</td>
      </tr>
      <tr>
          <td>Replaces</td>
          <td>Revokes BOD 22-01 and BOD 19-02 (vulnerability remediation guidance that had been split across them)</td>
      </tr>
      <tr>
          <td>Applies to</td>
          <td>Assets in any federal information system of Federal Civilian Executive Branch (FCEB) agencies</td>
      </tr>
      <tr>
          <td>Does not apply to</td>
          <td>National security systems and certain Department of War and Intelligence Community systems</td>
      </tr>
      <tr>
          <td>Contractors</td>
          <td>Generally excluded unless a governing procurement contract directs otherwise</td>
      </tr>
      <tr>
          <td>Cloud</td>
          <td>Agencies must inventory third-party hosted systems; FedRAMP offerings are handled through the FedRAMP Program Management Office</td>
      </tr>
  </tbody>
</table>
<p>Source: <a href="https://www.cisa.gov/news-events/directives/bod-26-04-prioritizing-security-updates-based-risk">CISA, BOD 26-04</a>.</p>
<h2 id="the-four-inputs-that-set-the-deadline">The four inputs that set the deadline</h2>
<p>Instead of one deadline for every KEV entry, the directive uses a table (Table 1, Appendix A) with four yes/no style decision points, measured in calendar days:</p>
<ol>
<li><strong>KEV status:</strong> is the vulnerability in the KEV catalog?</li>
<li><strong>Publicly exposed:</strong> is the asset reachable by unauthenticated or untrusted parties over public networks, wherever it physically sits? If several methods are used and any finds the asset exposed, the value is &ldquo;yes.&rdquo;</li>
<li><strong>Automatable:</strong> can exploitation be scripted reliably? CISA&rsquo;s heuristic: if a public proof-of-concept achieves remote code execution and reliably runs against a vulnerable system, it is automatable. CISA publishes results through its Vulnrichment Program.</li>
<li><strong>Technical impact:</strong> <em>total</em> if the attacker can run arbitrary software, trigger all actions of the component, get a full-privilege account, or both CVSS confidentiality and integrity impacts are High; otherwise <em>partial</em>. Denial of service generally counts as partial.</li>
</ol>
<p>The clock starts at whichever comes first: CISA adding the CVE to KEV, or the agency identifying the vulnerability on an asset and updating its Continuous Diagnostics and Mitigation (CDM) dashboard. Timelines are dynamic: removing an asset from public exposure can lengthen its deadline, and a KEV addition shortens it.</p>
<h3 id="what-the-deadlines-actually-are">What the deadlines actually are</h3>
<p>This is where we must be careful. CISA&rsquo;s text states that where the table says &ldquo;&amp; forensic triage,&rdquo; the agency must remediate or mitigate <strong>within three days</strong> and also perform forensic triage. The page also describes &ldquo;fix on system upgrade&rdquo; for lower-risk cases. The individual day counts for every row are in an image of Table 1 that our reading tool could not extract, so we have not verified them against CISA&rsquo;s own table.</p>
<p>For orientation only, Tenable&rsquo;s <a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">secondary summary</a> describes tiers of 3 days with triage, 3 days without triage, 14 days, 60 days and fix-on-upgrade. Treat that as a pointer, and read Appendix A on CISA&rsquo;s site before relying on a specific number.</p>
<h2 id="phase-dates">Phase dates</h2>
<p>The directive phases its requirements. Phase I applies immediately; Phase II within 60 days of issuance; Phase III within 180 days.</p>
<table>
  <thead>
      <tr>
          <th>Phase</th>
          <th>Requirement (summary)</th>
          <th>Date if counted from June 10, 2026</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>I</td>
          <td>Update vulnerability management policies; monitor KEV and remediate to KEV timelines; report KEV status via CDM or manually every two weeks</td>
          <td>Immediately</td>
      </tr>
      <tr>
          <td>II</td>
          <td>Update processes to cover vulnerabilities in the CVE database as well as KEV</td>
          <td>August 9, 2026</td>
      </tr>
      <tr>
          <td>III</td>
          <td>Remediate within Table 1 timelines; continuously tag externally reachable assets (manual reporters every seven days)</td>
          <td>December 7, 2026</td>
      </tr>
  </tbody>
</table>
<p>The dates in the last column are our arithmetic from the page date. The directive measures from issuance, and we found no separate issuance date, so treat them as approximate. On that count, Phase III had not yet begun as of October 12, 2026.</p>
<h2 id="case-study-five-old-cves-added-on-october-8-2026">Case study: five old CVEs added on October 8, 2026</h2>
<p>The KEV catalog lists five entries added on 2026-10-08, each with a due date of 2026-10-11. Each entry&rsquo;s required action refers to BOD 26-04 and CISA&rsquo;s &ldquo;Forensics Triage Requirements,&rdquo; but the catalog&rsquo;s forensic-triage field is not the same for all five: it is marked &ldquo;Yes&rdquo; for four and &ldquo;No&rdquo; for the BIND denial-of-service flaw, CVE-2015-5477.</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Product</th>
          <th>Published (NVD)</th>
          <th>What NVD says</th>
          <th>CVSS v3.1 base score (CISA enrichment)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2015-3306</td>
          <td>ProFTPD 1.3.5</td>
          <td>May 18, 2015</td>
          <td>mod_copy lets remote attackers read and write arbitrary files via site cpfr and cpto commands</td>
          <td>10.0</td>
      </tr>
      <tr>
          <td>CVE-2021-3199</td>
          <td>ONLYOFFICE Document Server before 5.6.3</td>
          <td>Jan 26, 2021</td>
          <td>Directory traversal with remote code execution in /upload when JWT is used</td>
          <td>9.8</td>
      </tr>
      <tr>
          <td>CVE-2023-22894</td>
          <td>Strapi through 4.5.5</td>
          <td>Apr 19, 2023</td>
          <td>Admin-panel users can infer sensitive user details, including password hashes and reset tokens for a super admin, via the query filter</td>
          <td>7.2 (NIST&rsquo;s own NVD score: 4.9)</td>
      </tr>
      <tr>
          <td>CVE-2016-3081</td>
          <td>Apache Struts 2.3.19 to 2.3.28 (specified ranges)</td>
          <td>Apr 26, 2016</td>
          <td>Remote code execution via method: prefix when Dynamic Method Invocation is enabled</td>
          <td>8.1</td>
      </tr>
      <tr>
          <td>CVE-2015-5477</td>
          <td>ISC BIND 9.x before 9.9.7-P2 and 9.10.x before 9.10.2-P3</td>
          <td>Jul 29, 2015</td>
          <td>Denial of service via TKEY queries (assertion failure and daemon exit)</td>
          <td>7.5</td>
      </tr>
  </tbody>
</table>
<p>Sources: <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a> and the NVD records (<a href="https://nvd.nist.gov/vuln/detail/CVE-2015-3306">CVE-2015-3306</a>, <a href="https://nvd.nist.gov/vuln/detail/CVE-2021-3199">CVE-2021-3199</a>, <a href="https://nvd.nist.gov/vuln/detail/CVE-2023-22894">CVE-2023-22894</a>, <a href="https://nvd.nist.gov/vuln/detail/CVE-2016-3081">CVE-2016-3081</a>, <a href="https://nvd.nist.gov/vuln/detail/CVE-2015-5477">CVE-2015-5477</a>). The scores are the CVSS v3.1 scores CISA added to each CVE record (its &ldquo;ADP&rdquo; enrichment), as updated on October 9, 2026; NIST&rsquo;s own NVD analysis has scored only the Strapi entry, at 4.9, and shows no v3.1 score for the BIND entry. Scores may change.</p>
<p>What this shows, and what it does not:</p>
<ul>
<li><strong>Age is not safety.</strong> These flaws date from 2015 to 2023, yet were added with a three-day due date. The three-day interval is what the catalog and NVD show; we cannot tell from public data which row of Table 1 applies to any agency&rsquo;s asset, since that depends on exposure.</li>
<li><strong>CVSS and KEV answer different questions.</strong> The Strapi entry has an NVD base score of 4.9 yet is in KEV. Our guide to <a href="/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/">CVSS, EPSS and KEV</a> explains why exploitation evidence outranks severity.</li>
<li><strong>We do not know how or by whom these were exploited.</strong> The catalog entries we read do not say, and we have not assumed it.</li>
<li><strong>Dates differ by product version.</strong> If you run any of these products, check your exact version against the NVD affected ranges and your vendor&rsquo;s fixes; backported distribution patches can change the answer.</li>
</ul>
<h2 id="forensic-triage-what-three-days-includes">Forensic triage: what &ldquo;three days&rdquo; includes</h2>
<p>CISA&rsquo;s <a href="https://www.cisa.gov/news-events/directives/bod-26-04-implementation-guidance-prioritizing-security-updates-based-risk">implementation guidance</a> presents baseline steps as best practice, and says the directive requires an adequate triage but not these specific timings. If a KEV entry has its own triage steps, those take precedence.</p>
<table>
  <thead>
      <tr>
          <th>Window</th>
          <th>Step</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>First 2 hours after KEV addition</td>
          <td>Scope: activate the team, define affected systems, set up out-of-band communication</td>
      </tr>
      <tr>
          <td>2 to 24 hours</td>
          <td>Preserve evidence, volatile data first; patch and stabilize after collection</td>
      </tr>
      <tr>
          <td>6 to 24 hours</td>
          <td>Contain and isolate affected and suspect systems</td>
      </tr>
      <tr>
          <td>24 to 48 hours</td>
          <td>Analyze: unauthorized access, persistence, lateral movement, data staging</td>
      </tr>
      <tr>
          <td>48 to 72 hours</td>
          <td>Report and decide: no evidence, suspected, or confirmed compromise (confirmed goes to CISA&rsquo;s reporting system)</td>
      </tr>
  </tbody>
</table>
<p>The ordering matters: evidence is collected before remediation where possible, because patching can destroy the traces you need.</p>
<h2 id="what-non-federal-teams-can-borrow">What non-federal teams can borrow</h2>
<p>The directive does not bind you, so these are our suggestions, not requirements.</p>
<ol>
<li><strong>Make exposure a first-class input.</strong> Tag assets by environment, exposure and type, as the directive&rsquo;s Phase III tagging requires, so &ldquo;is it internet-reachable?&rdquo; is a query rather than a meeting.</li>
<li><strong>Treat KEV as a trigger for compromise checks, not only patching.</strong> An exploited flaw that was exposed before you patched may already have been used. Compare with how we handled <a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">recent zero-day roundups</a> and the <a href="/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/">Cisco SD-WAN log checks</a>.</li>
<li><strong>Rehearse the 72 hours.</strong> Decide who runs scoping, where evidence goes and how you communicate if email is suspect. Our piece on <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">why vulnerability backlogs never shrink</a> covers the process side.</li>
<li><strong>Inventory old software.</strong> The October 8 batch is a reminder that forgotten FTP daemons, unpatched Struts apps and DNS servers can matter. If you cannot list them, you cannot fix them.</li>
<li><strong>Know your response to a zero-day</strong> before one lands: see <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">what to do when a zero-day is announced</a>.</li>
</ol>
<h2 id="limits-of-this-article">Limits of this article</h2>
<p>We could not read Table 1 as text and have not verified individual day counts beyond the three-day forensic triage requirement. We did not find a separate issuance date, so the phase dates are approximate. We did not verify whether any of the five products is exploited in a particular environment, nor the exploitation details behind the KEV additions. We did not review agencies&rsquo; compliance with the directive.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 12, 2026.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>BOD 26-04 page dated June 10, 2026; revokes BOD 22-01 and BOD 19-02; scope and exclusions; phases (immediate, 60, 180 days); four decision points; clock start; exposure, automatable and technical impact definitions</td>
          <td><a href="https://www.cisa.gov/news-events/directives/bod-26-04-prioritizing-security-updates-based-risk">CISA, BOD 26-04</a></td>
          <td>Qualified: no separate issuance date found; Table 1 day counts not readable as text</td>
      </tr>
      <tr>
          <td>Three-day remediation plus forensic triage where the table says so</td>
          <td>Same page</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Tier descriptions (3, 14, 60 days, fix-on-upgrade)</td>
          <td><a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">Tenable analysis</a></td>
          <td>Qualified: secondary source, not checked against Table 1</td>
      </tr>
      <tr>
          <td>Triage steps and timings; directive requires adequate triage, not these timings</td>
          <td><a href="https://www.cisa.gov/news-events/directives/bod-26-04-implementation-guidance-prioritizing-security-updates-based-risk">CISA implementation guidance</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Five CVEs added to KEV 2026-10-08, due 2026-10-11</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog">CISA KEV catalog</a> and the five NVD records linked above</td>
          <td>Verified as of October 12, 2026</td>
      </tr>
      <tr>
          <td>Publication dates, descriptions, affected versions</td>
          <td>NVD records linked above</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS v3.1 scores (CISA enrichment) and the NIST score of 4.9 for Strapi</td>
          <td><a href="https://github.com/CVEProject/cvelistV5">CVE records in the CVE Program&rsquo;s cvelistV5 repository</a>, updated October 9, 2026; NVD page for CVE-2023-22894</td>
          <td>Qualified: scores may change</td>
      </tr>
      <tr>
          <td>Forensic triage field &ldquo;Yes&rdquo; for four entries, &ldquo;No&rdquo; for CVE-2015-5477</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data file</a>, catalog version 2026.10.08</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Phase II and III dates (August 9 and December 7, 2026)</td>
          <td>Our calculation from June 10, 2026</td>
          <td>Qualified: assumes issuance date equals page date</td>
      </tr>
      <tr>
          <td>Suggestions for non-federal teams</td>
          <td>Our analysis</td>
          <td>Qualified: editorial judgment</td>
      </tr>
  </tbody>
</table>
<h2 id="correction">Correction</h2>
<p><em>October 12, 2026:</em> An earlier version of this article described the CVSS v3.1 scores in the KEV table as &ldquo;listed in NVD&rdquo;. They are CISA&rsquo;s enrichment (ADP) scores in the CVE records, not NIST&rsquo;s own NVD scores. It also said every entry added on October 8 was marked for forensic triage; CVE-2015-5477 (BIND) is not. Both points have been corrected.</p>
]]></content:encoded></item><item><title>CVSS vs EPSS vs KEV: Which Signal Should Decide What You Patch First?</title><link>https://firewallsync.com/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/</link><guid>https://firewallsync.com/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/</guid><pubDate>Wed, 07 Oct 2026 08:30:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>CVSS says how bad a flaw could be, EPSS estimates how likely it is to be attacked, and KEV records that it already has been. Seven recently exploited flaws show why the order you check them in matters.</description><enclosure url="https://firewallsync.com/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/cover_hu9987397115794223768.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/cover_hu9987397115794223768.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cvss-vs-epss-vs-kev-which-signal-should-decide-what-you-patch-first/cover_hu9987397115794223768.jpg" alt="Colorful sticky notes pinned to a board" width="1200" height="630"></p><p>Check the CISA <a href="/posts/cisa-kev-json-feed-what-each-field-means-and-how-to-match-it-against-a-scanner-export/">Known Exploited Vulnerabilities (KEV) catalog</a> first, then EPSS, then CVSS, and apply your own knowledge of exposure at every step. KEV tells you a flaw has been exploited. EPSS estimates how likely exploitation is when you have no such evidence. CVSS tells you how bad exploitation would be. Using any one of them alone produces a predictable failure, and seven flaws from the past two weeks show each failure clearly.</p>
<p>This article explains what each signal measures, tests them against those seven flaws, and gives an order of operations you can implement.</p>
<h2 id="what-each-one-measures">What each one measures</h2>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVSS</th>
          <th>EPSS</th>
          <th>KEV</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Full name</td>
          <td>Common Vulnerability Scoring System</td>
          <td>Exploit Prediction Scoring System</td>
          <td>Known Exploited Vulnerabilities catalog</td>
      </tr>
      <tr>
          <td>Published by</td>
          <td>FIRST (Forum of Incident Response and Security Teams) maintains the standard; vendors and analysts assign scores</td>
          <td>FIRST&rsquo;s EPSS special interest group; scores generated by Empirical Security</td>
          <td>CISA (US Cybersecurity and Infrastructure Security Agency)</td>
      </tr>
      <tr>
          <td>Question it answers</td>
          <td>How severe is this flaw?</td>
          <td>How likely is exploitation activity in the next 30 days?</td>
          <td>Is there reliable evidence this flaw has been exploited?</td>
      </tr>
      <tr>
          <td>Output</td>
          <td>Score from 0.0 to 10.0 plus a vector string</td>
          <td>Probability from 0 to 1, plus a percentile rank</td>
          <td>A yes or no listing, with a date added and a due date</td>
      </tr>
      <tr>
          <td>Changes over time</td>
          <td>Base score normally does not</td>
          <td>Recalculated daily</td>
          <td>Entries are added as evidence emerges</td>
      </tr>
      <tr>
          <td>Knows about your environment</td>
          <td>No, unless you supply Environmental metrics</td>
          <td>No</td>
          <td>No</td>
      </tr>
  </tbody>
</table>
<h3 id="cvss-severity-not-risk">CVSS: severity, not risk</h3>
<p>The CVSS v4.0 specification describes the base score as reflecting a vulnerability&rsquo;s severity according to intrinsic characteristics that stay constant over time. It covers things like whether the flaw is reachable over a network, whether the attacker needs an account, and what happens to confidentiality, integrity and availability.</p>
<p>The qualitative labels map to score ranges: Low is 0.1 to 3.9, Medium 4.0 to 6.9, High 7.0 to 8.9 and Critical 9.0 to 10.0.</p>
<p>CVSS v4.0 has three other metric groups that most published scores leave empty:</p>
<ul>
<li><strong>Threat</strong> metrics, mainly Exploit Maturity, which can be set to Attacked, Proof-of-Concept or Unreported.</li>
<li><strong>Environmental</strong> metrics, which adjust for your own controls and how much the affected system matters to you.</li>
<li><strong>Supplemental</strong> metrics, which add context without changing the score.</li>
</ul>
<p>The specification names scores by which groups were used: CVSS-B is base only, CVSS-BT adds Threat, CVSS-BE adds Environmental and CVSS-BTE uses all three. It expects vendors and analysts to supply base metrics and the consuming organization to supply Threat and Environmental ones. In practice, the number in an advisory or scanner report is almost always CVSS-B, which is the general-case severity and nothing more.</p>
<p>Two scoring versions are in circulation. Among the flaws discussed below, Cisco and Fortinet published CVSS 3.1 scores and Citrix published CVSS 4.0 scores. The two versions are calculated differently, so treat a 9.8 on one and a 9.5 on the other as &ldquo;both Critical&rdquo;, not as a ranking.</p>
<h3 id="epss-likelihood-with-a-time-window">EPSS: likelihood, with a time window</h3>
<p>FIRST&rsquo;s FAQ defines EPSS as &ldquo;a data-driven model that estimates the probability a vulnerability will be exploited in the wild within the next 30 days.&rdquo; Three details matter when you read a score:</p>
<ul>
<li><strong>It is a probability.</strong> A score of 0.02 means an estimated 2% chance of observed exploitation activity in the next 30 days.</li>
<li><strong>The percentile is a ranking, not a probability.</strong> A percentile of 0.82 means the score is higher than 82% of all scored CVEs. Because most CVEs have very low scores, a high percentile can sit on top of a small probability. When we queried the API on October 7, 2026, it reported about 383,000 scored CVEs.</li>
<li><strong>&ldquo;Exploitation activity&rdquo; means attempts.</strong> FIRST&rsquo;s model description defines it as evidence that exploitation was attempted, not that it succeeded, and says the data comes from sources such as honeypots, intrusion detection sensors and host-based detection.</li>
</ul>
<p>Scores are republished every day, typically just after 13:30 UTC according to the FAQ.</p>
<p>FIRST is direct about the limits. The FAQ says &ldquo;EPSS is not a complete risk score&rdquo;, that it does not measure how much damage exploitation would cause or whether a flaw affects your environment, and that &ldquo;multiplying EPSS by a CVSS score does not compute probability × severity and is never a good idea.&rdquo;</p>
<h3 id="kev-evidence-not-prediction">KEV: evidence, not prediction</h3>
<p>CISA adds a vulnerability to the catalog when three conditions are met: it has a CVE ID, &ldquo;there is reliable evidence that the vulnerability has been actively exploited in the wild&rdquo;, and &ldquo;there is a clear remediation action for the vulnerability, such as a vendor-provided update.&rdquo;</p>
<p>CISA&rsquo;s definition of exploitation is specific. Scanning, security research on an exploit and the existence of a proof of concept do not count. Attempted exploitation, including against a honeypot, does.</p>
<p>As of catalog version 2026.10.04, the catalog held 1,734 entries, of which 250 were added in 2026. It flags 361 entries, about 21%, as known to be used in ransomware campaigns.</p>
<p>What a listing does not tell you is how widespread exploitation is, or whether you are a likely target. And the criteria imply two gaps, which is our reading rather than CISA&rsquo;s wording: a flaw can be exploited before CISA has reliable evidence of it, and a flaw with no clear remediation does not meet the third criterion. Absence from KEV means &ldquo;no listing&rdquo;, not &ldquo;not exploited&rdquo;.</p>
<h3 id="a-fourth-signal-you-already-have-cisas-ssvc-values">A fourth signal you already have: CISA&rsquo;s SSVC values</h3>
<p>SSVC (Stakeholder-Specific Vulnerability Categorization) is a decision-tree method that CISA says was created in 2019. CISA&rsquo;s version uses five decision points, including exploitation status, technical impact and whether exploitation is automatable, and ends in one of four outcomes: Track, Track*, Attend or Act.</p>
<p>The useful part for day-to-day triage is that CISA publishes three of those values inside many public CVE records: Exploitation (none, proof of concept or active), Automatable (yes or no) and Technical Impact (partial or total). Automatable asks whether an attacker can reliably automate creating exploitation events for the vulnerability, in the wording of the SSVC project&rsquo;s documentation. A flaw marked active, automatable and total impact is the kind that gets sprayed across the internet.</p>
<h2 id="seven-exploited-flaws-three-signals">Seven exploited flaws, three signals</h2>
<p>We covered seven flaws that CISA added to KEV between September 27 and October 4, 2026. The table puts their published CVSS base scores next to their EPSS scores for October 6, 2026 and CISA&rsquo;s Automatable value.</p>
<table>
  <thead>
      <tr>
          <th>CVE</th>
          <th>Product</th>
          <th>CVSS base score (version)</th>
          <th>EPSS on Oct 6, 2026</th>
          <th>EPSS percentile</th>
          <th>KEV added</th>
          <th>Automatable (CISA)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-88771</td>
          <td>Citrix NetScaler</td>
          <td>9.5 (4.0)</td>
          <td>1.1%</td>
          <td>64th</td>
          <td>Sept 27</td>
          <td>Yes</td>
      </tr>
      <tr>
          <td>CVE-2026-88772</td>
          <td>Citrix NetScaler</td>
          <td>9.5 (4.0)</td>
          <td>1.3%</td>
          <td>70th</td>
          <td>Sept 27</td>
          <td>No</td>
      </tr>
      <tr>
          <td>CVE-2026-76504</td>
          <td>Cisco Catalyst SD-WAN Manager</td>
          <td>9.8 (3.1)</td>
          <td>1.8%</td>
          <td>78th</td>
          <td>Sept 30</td>
          <td>Yes</td>
      </tr>
      <tr>
          <td>CVE-2026-104286</td>
          <td>Fortinet FortiMail</td>
          <td>9.8 (3.1)</td>
          <td>2.2%</td>
          <td>82nd</td>
          <td>Oct 1</td>
          <td>Yes</td>
      </tr>
      <tr>
          <td>CVE-2026-102489</td>
          <td>Zammad</td>
          <td>8.7 (4.0)</td>
          <td>1.4%</td>
          <td>72nd</td>
          <td>Oct 2</td>
          <td>Yes</td>
      </tr>
      <tr>
          <td>CVE-2026-102490</td>
          <td>Zammad</td>
          <td>8.5 (4.0)</td>
          <td>0.6%</td>
          <td>48th</td>
          <td>Oct 2</td>
          <td>No</td>
      </tr>
      <tr>
          <td>CVE-2026-88779</td>
          <td>Citrix NetScaler</td>
          <td>8.7 (4.0)</td>
          <td>0.6%</td>
          <td>47th</td>
          <td>Oct 4</td>
          <td>Yes</td>
      </tr>
  </tbody>
</table>
<p>EPSS values are rounded from the API&rsquo;s figures, which run from 0.00592 to 0.02201. The Zammad records also carry a second score of 9.4 for the two flaws chained together.</p>
<p>What each signal would have done on its own:</p>
<ul>
<li><strong>EPSS alone would have missed all seven.</strong> Every score is at or below 2.2%. A team that acts on EPSS of 10% or higher, a commonly used cut-off, would have flagged none of them. Even a team working by percentile and taking the top 10% of CVEs would have flagged none, since the highest sits at the 82nd percentile.</li>
<li><strong>CVSS &ldquo;Critical only&rdquo; would have missed three.</strong> Four of the seven score 9.0 or higher. The two Zammad flaws and the NetScaler denial-of-service flaw score 8.5 to 8.7, which is High. A rule of 7.0 and above catches all seven, but that rule also puts every High and Critical flaw in your inventory into the same urgent queue, which is the backlog problem in the first place.</li>
<li><strong>KEV caught all seven, by construction.</strong> We picked these because they were listed. That is the point of the comparison, not a flaw in it: for a flaw with exploitation evidence, the evidence is the signal.</li>
</ul>
<p>This is not a criticism of EPSS, and seven flaws are not a statistical sample. EPSS is doing what FIRST says it does. Its FAQ states that the model generates scores &ldquo;without knowing if the vulnerability was recently exploited&rdquo; and that &ldquo;if direct evidence of active exploitation exists, that evidence should supersede EPSS.&rdquo; These were newly disclosed flaws: CISA added all seven to KEV within two days of their CVE records being published. That is the situation where a model built on broad observation has the least to work with, which is our interpretation, not a statement from FIRST.</p>
<p>The mistake would be on the reader&rsquo;s side: seeing 0.6% next to a CVE and lowering its priority without checking KEV.</p>
<h2 id="an-order-of-operations">An order of operations</h2>
<p>This sequence is our recommendation. It uses each signal for the question it can answer.</p>
<ol>
<li><strong>Is it in KEV, or does the vendor say it is being exploited?</strong> If yes and the affected system is reachable from untrusted networks, act now, and plan a compromise check as well as a patch. If the system is internal only, schedule it within days. Vendor statements count: Fortinet&rsquo;s advisory for CVE-2026-104286 said the flaw was being exploited on the day it was published.</li>
<li><strong>If not, what does EPSS say?</strong> A score above your chosen threshold moves the item up. Our <a href="/posts/epss-data-feed-columns-percentile-vs-probability-and-how-to-join-it-to-a-scanner-export/">EPSS data feed guide</a> shows how to add the daily scores to a scanner export. Pick the threshold from your capacity: lower it until the resulting queue is the size your team can clear.</li>
<li><strong>If EPSS is low, is it Critical and exposed?</strong> A CVSS base score of 9.0 or higher on an internet-facing system still deserves the next patch cycle, because a low EPSS score on a fresh CVE can change quickly.</li>
<li><strong>Everything else follows the routine schedule.</strong></li>
</ol>
<p>At every step, the exposure question is yours to answer. None of the three sources knows whether your FortiMail web interface faces the internet.</p>
<p>Here is that logic as a small Python function. It is an illustration of the ordering, not a finished tool. We ran it against the KEV data file and the seven flaws above; the thresholds are examples.</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-python" data-lang="python"><span style="display:flex;"><span><span style="color:#f92672">import</span> json
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>EPSS_WATCH <span style="color:#f92672">=</span> <span style="color:#ae81ff">0.10</span>  <span style="color:#75715e"># probability, not percentile; tune to your own capacity</span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">load_kev</span>(path: str) <span style="color:#f92672">-&gt;</span> set[str]:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">with</span> open(path) <span style="color:#66d9ef">as</span> f:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> {v[<span style="color:#e6db74">&#34;cveID&#34;</span>] <span style="color:#66d9ef">for</span> v <span style="color:#f92672">in</span> json<span style="color:#f92672">.</span>load(f)[<span style="color:#e6db74">&#34;vulnerabilities&#34;</span>]}
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span>
</span></span><span style="display:flex;"><span><span style="color:#66d9ef">def</span> <span style="color:#a6e22e">triage</span>(cve: str, cvss: float, epss: float, internet_facing: bool, kev: set[str]) <span style="color:#f92672">-&gt;</span> str:
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> cve <span style="color:#f92672">in</span> kev:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;1 - act now&#34;</span> <span style="color:#66d9ef">if</span> internet_facing <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;2 - this week&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> epss <span style="color:#f92672">&gt;=</span> EPSS_WATCH:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;2 - this week&#34;</span> <span style="color:#66d9ef">if</span> internet_facing <span style="color:#66d9ef">else</span> <span style="color:#e6db74">&#34;3 - next cycle&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">if</span> cvss <span style="color:#f92672">&gt;=</span> <span style="color:#ae81ff">9.0</span> <span style="color:#f92672">and</span> internet_facing:
</span></span><span style="display:flex;"><span>        <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;3 - next cycle&#34;</span>
</span></span><span style="display:flex;"><span>    <span style="color:#66d9ef">return</span> <span style="color:#e6db74">&#34;4 - routine&#34;</span>
</span></span></code></pre></div><p><code>load_kev</code> reads the JSON file CISA publishes at <a href="https://github.com/cisagov/kev-data">github.com/cisagov/kev-data</a>. With the KEV check in place, all seven flaws returned &ldquo;1 - act now&rdquo; for an internet-facing system. With an empty KEV set, simulating a team that skips that check, four returned &ldquo;3 - next cycle&rdquo; and three returned &ldquo;4 - routine&rdquo;. Same flaws, same scores, one missing lookup.</p>
<p>Daily EPSS scores are available from FIRST&rsquo;s API at <code>https://api.first.org/data/v1/epss?cve=CVE-2026-104286</code>, which accepts a comma-separated list of CVE IDs.</p>
<h2 id="common-mistakes">Common mistakes</h2>
<ul>
<li><strong>Sorting the backlog by CVSS and working from the top.</strong> This treats a Critical flaw on an isolated test server as more urgent than a High flaw being exploited on your gateway.</li>
<li><strong>Reading a low EPSS score as &ldquo;safe&rdquo;.</strong> It is a 30-day estimate from a model. Check KEV and the vendor advisory before lowering priority.</li>
<li><strong>Confusing EPSS percentile with probability.</strong> &ldquo;82nd percentile&rdquo; sounds urgent. The probability behind it in the table above is 2.2%.</li>
<li><strong>Treating KEV as complete.</strong> It lists what CISA has evidence for and can point to a fix for.</li>
<li><strong>Multiplying the scores together.</strong> FIRST says not to. The result has no defined meaning.</li>
<li><strong>Recording &ldquo;patched&rdquo; instead of the fixed version.</strong> Citrix issued two bulletins for NetScaler within a week, and the builds that fixed the first were still affected by the second, as we covered in <a href="/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/">the NetScaler CVE-2026-88779 article</a>.</li>
<li><strong>Trusting one data source for affected versions.</strong> In the <a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">FortiMail case</a>, the CVE record and the vendor advisory listed different version ranges.</li>
</ul>
<p>For the organizational side of this, including why backlogs grow when everything is ranked by count, see <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">the vulnerability backlog that never shrinks</a>. For what to do once something lands in the top tier, see <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">how to respond to a zero-day</a>.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 7, 2026. EPSS scores change daily and the KEV catalog grows, so the figures below describe those dates only. The seven flaws were chosen because we had already reported on them; they are examples, not a representative sample.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVSS base score reflects severity from intrinsic characteristics; metric groups; CVSS-B, -BT, -BE, -BTE naming; severity ranges; Exploit Maturity values</td>
          <td><a href="https://www.first.org/cvss/v4.0/specification-document">CVSS v4.0 specification</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>EPSS definition, 30-day window, not a complete risk score, do not multiply by CVSS, exploitation evidence supersedes EPSS, daily publication around 13:30 UTC, scores generated by Empirical Security</td>
          <td><a href="https://www.first.org/epss/faq">FIRST EPSS FAQ</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>&ldquo;Exploitation activity&rdquo; means attempted exploitation; data from honeypots, IDS/IPS sensors and host-based detection</td>
          <td><a href="https://www.first.org/epss/model">FIRST EPSS model page</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV criteria; scanning, research and proof of concept do not count as exploitation; use KEV as an input to prioritization</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities">CISA, Known Exploited Vulnerabilities</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>KEV held 1,734 entries, 250 added in 2026, 361 flagged for known ransomware use</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data file</a>, catalog version 2026.10.04; our count</td>
          <td>Verified for that version</td>
      </tr>
      <tr>
          <td>SSVC created in 2019; CISA decision points and four outcomes</td>
          <td><a href="https://www.cisa.gov/stakeholder-specific-vulnerability-categorization-ssvc">CISA, SSVC</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Definition of Automatable</td>
          <td><a href="https://certcc.github.io/SSVC/reference/decision_points/automatable/">SSVC project documentation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS base scores, KEV dates and Automatable values for the seven CVEs</td>
          <td>Public CVE records (for example <a href="https://www.cve.org/CVERecord?id=CVE-2026-104286">CVE-2026-104286</a>) and the KEV data file, retrieved October 7, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>EPSS scores and percentiles for the seven CVEs</td>
          <td><a href="https://api.first.org/data/v1/epss?cve=CVE-2026-88771,CVE-2026-88772,CVE-2026-88779,CVE-2026-76504&amp;date=2026-10-06">FIRST EPSS API</a>, scores dated October 6, 2026</td>
          <td>Verified for that date</td>
      </tr>
      <tr>
          <td>About 383,000 CVEs scored by EPSS</td>
          <td>FIRST EPSS API total, queried October 7, 2026</td>
          <td>Qualified: read from one API response</td>
      </tr>
      <tr>
          <td>10% as a commonly used EPSS cut-off</td>
          <td>Our characterization</td>
          <td>Qualified: an example threshold, not a FIRST recommendation</td>
      </tr>
      <tr>
          <td>Triage function results</td>
          <td>Run in our build environment on Python 3.13 against KEV catalog 2026.10.04</td>
          <td>Verified; thresholds are illustrative</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>ZITADEL Account-Takeover Flaws CVE-2026-105207, -105209 and -105215: Fixed Versions and What to Check</title><link>https://firewallsync.com/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/</link><guid>https://firewallsync.com/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/</guid><pubDate>Tue, 06 Oct 2026 06:35:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Identity &amp; Access</category><description>Three critical ZITADEL flaws let attackers take over accounts through external identity provider linking, passkey enrollment and forged login callbacks. Fixes shipped between July and September 2026, but the CVE records only appeared on October 4, and ZITADEL 3.x has no fix for the worst one.</description><enclosure url="https://firewallsync.com/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/cover_hu14719302590193262893.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/cover_hu14719302590193262893.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/zitadel-account-takeover-cve-2026-105207-cve-2026-105209-cve-2026-105215-fixed-versions/cover_hu14719302590193262893.jpg" alt="Account settings screen showing a verification prompt" width="1200" height="630"></p><p>If you run ZITADEL, the open-source identity platform, check the version number today. Three critical flaws, each of which lets an attacker take over someone else&rsquo;s account, were fixed in releases between mid-July and early September 2026, but their CVE records were published on October 4, 2026. Anything older than 4.17.3 is affected by at least one of them, and ZITADEL 3.x has no fix for the most serious one.</p>
<p>This article lays out which versions are affected by each flaw, what each requires from an attacker, and which questions to ask of your own logs. It is based on ZITADEL&rsquo;s GitHub security advisories and the CVE records; we did not install ZITADEL or test any of the flaws.</p>
<h2 id="the-three-flaws">The three flaws</h2>
<p>All three are authorization or verification failures around how an account gets tied to a login method. The details come from ZITADEL&rsquo;s advisories, and the dates and CVSS 4.0 scores from the CVE records, which were published by VulnCheck, acting as CVE Numbering Authority (the organization authorized to issue the CVE IDs).</p>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-105207</th>
          <th>CVE-2026-105209</th>
          <th>CVE-2026-105215</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Advisory</td>
          <td>GHSA-g8gj-gq47-xgf4</td>
          <td>GHSA-pq2q-2c6r-75c4</td>
          <td>GHSA-738m-7888-jfv8</td>
      </tr>
      <tr>
          <td>Flaw</td>
          <td>External identity provider (IdP) links are created without verifying a primary factor or the caller&rsquo;s permission</td>
          <td>Passkey or passwordless enrollment codes are issued after checking only the organization named in a request header, not the target user&rsquo;s organization</td>
          <td>Login V1&rsquo;s &ldquo;external account not found&rdquo; registration endpoint trusts client-supplied identity fields without a completed IdP callback</td>
      </tr>
      <tr>
          <td>Result</td>
          <td>Account takeover</td>
          <td>Account takeover across organizations</td>
          <td>Account pre-hijacking</td>
      </tr>
      <tr>
          <td>What the attacker needs</td>
          <td>The victim&rsquo;s login name; no account, no interaction</td>
          <td>User-write permission in any one organization on the same instance</td>
          <td>The victim&rsquo;s external user ID, known or predictable; the victim must later sign in with that IdP</td>
      </tr>
      <tr>
          <td>Affected versions</td>
          <td>4.0.0 to 4.17.2; 3.0.0 to 3.4.15</td>
          <td>4.0.0 to 4.17.0; 3.0.0 to 3.4.14</td>
          <td>4.0.0 to 4.16.1; 3.0.0 to 3.4.13</td>
      </tr>
      <tr>
          <td>Fixed in</td>
          <td>4.17.3 and later; no 3.x fix</td>
          <td>4.17.1 and later; 3.4.15 and later</td>
          <td>4.16.2 and later; 3.4.14 and later</td>
      </tr>
      <tr>
          <td>Login UI affected</td>
          <td>Login V2 and the User Service V2 API; Login V1 unaffected</td>
          <td>Not described as specific to one login UI; concerns issuing enrollment codes</td>
          <td>Login V1 only; Login V2 unaffected</td>
      </tr>
      <tr>
          <td>Advisory score</td>
          <td>9.8 Critical</td>
          <td>9.6 Critical</td>
          <td>9.1 Critical</td>
      </tr>
      <tr>
          <td>CVSS 4.0 score in CVE record</td>
          <td>9.3</td>
          <td>9.3</td>
          <td>9.3</td>
      </tr>
      <tr>
          <td>CVE record published</td>
          <td>October 4, 2026</td>
          <td>October 4, 2026</td>
          <td>October 4, 2026</td>
      </tr>
  </tbody>
</table>
<p>An IdP (identity provider) is an external service such as a company directory or social login that vouches for a user&rsquo;s identity. A passkey is a phishing-resistant sign-in credential bound to a device, covered in our comparison of <a href="/posts/passkeys-vs-passwords-what-actually-changes-for-security-teams/">passkeys and passwords</a>.</p>
<h3 id="what-each-one-means-in-practice">What each one means in practice</h3>
<p><strong>CVE-2026-105207 is the broadest.</strong> ZITADEL&rsquo;s advisory says an attacker who knows only a victim&rsquo;s login name can bind their own external identity to the victim&rsquo;s account through the Login V2 identify-only session, then sign in as the victim. The same weakness is reachable through the API endpoint that adds IdP links, so changing the login screen does not close it. The advisory says it needs no authentication, no privileges and no user interaction. If your instance allows external IdP sign-in at all, treat this as the priority.</p>
<p><strong>CVE-2026-105209 matters most for multi-tenant setups.</strong> If you use separate ZITADEL organizations for separate customers or business units on one instance, an administrator in one organization could obtain an enrollment code for a user in another and register their own authenticator. The attacker needs a legitimate foothold first, so for a single-organization company the realistic threat is an insider or a compromised administrator account. For a service that gives each customer an organization, a customer administrator is a possible attacker.</p>
<p><strong>CVE-2026-105215 is a pre-hijack.</strong> The attacker creates an account in advance, bound to the victim&rsquo;s external identity. The victim&rsquo;s first genuine sign-in through that IdP then lands in the attacker-prepared account. It needs the victim&rsquo;s external user ID, which the advisory says may be known or predictable, and only Login V1 is affected.</p>
<h2 id="when-the-fixes-actually-shipped">When the fixes actually shipped</h2>
<p>The CVE records were published on October 4, 2026; two of them (CVE-2026-105209 and CVE-2026-105215) were updated on October 5. The fixes are older. The &ldquo;public&rdquo; dates in the CVE records are July 29, August 14 and September 4, 2026 for CVE-2026-105215, CVE-2026-105209 and CVE-2026-105207, respectively. ZITADEL&rsquo;s releases page lists 3.4.15 and 4.17.1 as released August 14, 2026 and 4.17.3 as released September 4, 2026, which matches. It lists 4.16.2 and 3.4.14 as released July 29, 2026, which also matches.</p>
<p>Two consequences follow. First, an organization that kept to a routine upgrade cadence may already be on a fixed build without having seen any alert. Second, an organization that skipped those releases has been exposed for weeks to flaws whose fixes were public, and the CVE publication on October 4 makes the technical detail easier for others to find. Whether anyone used them in that window is something we found no reporting on.</p>
<h3 id="where-the-sources-disagree-or-are-unclear">Where the sources disagree or are unclear</h3>
<ul>
<li><strong>Version ranges for CVE-2026-105207.</strong> ZITADEL&rsquo;s advisory lists 3.0.0 through 3.4.15 and 4.0.0 through 4.17.2 as affected, with 4.17.3 as the fix. The CVE record&rsquo;s text agrees, but its structured data also contains a second block that lists versions up to and including 4.19.4 as affected. That contradicts the advisory, which names 4.17.3 as the fixed release, and would flag a patched 4.19.4 server in any scanner that reads that field. We treat the advisory as authoritative and note the record&rsquo;s data as inconsistent; we have not seen a correction.</li>
<li><strong>Severity numbers.</strong> A widely circulated summary lists the three flaws as 9.8, 9.6 and 9.1. Those are the CVSS 3.1 scores. On CVSS 4.0, which the CVE records also carry, all three score 9.3. The two are different scales and should not be compared against each other.</li>
<li><strong>What &ldquo;latest&rdquo; is.</strong> The newest release we found on October 6 was 4.19.4, dated October 1, 2026. The release notes for 4.19.2 (September 28, 2026) say it addresses three further vulnerabilities: two rated High, both account takeover issues (advisories GHSA-x4c7-fpcx-w9q6 and GHSA-jh92-5mrj-p2w2), and one rated Low (GHSA-w4gv-rcwj-w6r5). We did not review those advisories in detail, and they are separate from the three CVEs covered here.</li>
</ul>
<h2 id="what-to-do-in-order">What to do, in order</h2>
<p>This sequence is our recommendation, built on ZITADEL&rsquo;s advisories.</p>
<h3 id="1-find-every-instance-and-its-version">1. Find every instance and its version</h3>
<p>Include development, staging and any instance that serves a single customer. Record whether it runs 3.x or 4.x, and whether Login V1, Login V2 or both are enabled, since that decides which flaws apply.</p>
<h3 id="2-upgrade-to-4173-or-later-preferably-the-newest-4x">2. Upgrade to 4.17.3 or later, preferably the newest 4.x</h3>
<p>Fixing all three requires at least 4.17.3. Given the further fixes in 4.19.2, aim for the newest 4.x release. One upgrade requirement to plan for: the 4.19.2 release notes say Login V2 now requires session cookie signing, that the <code>ZITADEL_SESSION_COOKIE_SECRET</code> environment variable (at least 32 characters) must be set before upgrading, and that users have to sign in again afterward. If you are on 3.x, plan the move to 4.x now: the advisory for CVE-2026-105207 says the 3.x line reached end-of-life on August 31, 2026 and will not be patched, so staying on 3.x leaves that flaw open. Test the upgrade first, because a major-version move can change login behavior.</p>
<h3 id="3-if-you-must-wait-apply-the-partial-workarounds-and-limit-exposure">3. If you must wait, apply the partial workarounds and limit exposure</h3>
<p>ZITADEL lists two partial workarounds. Disabling external IdP account linking reduces the CVE-2026-105207 route through Login V2 but not the API route. Disabling manual account creation on external IdPs reduces CVE-2026-105215 but turns off legitimate self-service registration. For CVE-2026-105209 the advisory says no configuration workaround fully addresses it. Our own suggestion, not ZITADEL&rsquo;s: where an instance is only used by staff, restricting it to a VPN or an allowlist of source addresses reduces the number of people who can reach the vulnerable endpoints, though it does not help against CVE-2026-105209, where the attacker is already an authenticated administrator.</p>
<h3 id="4-look-for-takeovers-that-already-happened">4. Look for takeovers that already happened</h3>
<p>Upgrading removes the flaw, not an attacker who has already taken an account. We did not verify ZITADEL&rsquo;s log or event names, so use your own audit logs and match these questions to what they record:</p>
<ul>
<li><strong>IdP links.</strong> Have any existing accounts gained an external IdP link that the owner did not create, especially since August 2026? A takeover through CVE-2026-105207 leaves a new link on the victim&rsquo;s account.</li>
<li><strong>Passkey and passwordless enrollment.</strong> Were enrollment codes issued by an administrator in organization A for a user in organization B? Did any account register a new authenticator it did not request?</li>
<li><strong>Accounts created through external sign-in that no real person used.</strong> Pre-hijack accounts are created through the registration endpoint, with an external identity attached before the real person arrives.</li>
<li><strong>Sign-ins that follow one of those changes</strong> from an unfamiliar address or device.</li>
</ul>
<p>If you find a suspicious link or authenticator, remove it, revoke the account&rsquo;s active sessions and review what the account could reach. Our guide to <a href="/posts/how-to-detect-and-rotate-leaked-api-keys-before-attackers-use-them/">detecting leaked keys and rotating them</a> covers the follow-up for any secrets those accounts could read.</p>
<h3 id="5-keep-the-logs">5. Keep the logs</h3>
<p>Make sure identity-provider events are retained somewhere an attacker with administrator rights in one tenant cannot erase them. The reasoning is in our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents rather than dashboards</a>.</p>
<h2 id="what-this-says-about-identity-systems-generally">What this says about identity systems generally</h2>
<p>The three flaws share one pattern, which is our reading, not ZITADEL&rsquo;s: the system trusted a claim about who was acting without checking it where the change happened. In CVE-2026-105207 it was a link request, in CVE-2026-105209 a header naming an organization, and in CVE-2026-105215 identity fields supplied by the browser. When you review any identity platform, including a hosted one, ask where an account&rsquo;s credentials can be added or changed, and what is checked at that exact step.</p>
<p>The timing is also a lesson. These are fixed-before-disclosed flaws, and the organizations hurt are those that treat identity software as set-and-forget. Put your identity provider on the same patch watchlist as your edge devices, and rank it using the approach in our <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">vulnerability backlog</a> piece.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 6, 2026. We did not install ZITADEL, test any flaw or review its source code. ZITADEL&rsquo;s advisories and releases page were read through a page-summarizing tool; the CVE records were read directly from the CVE Program API. The affected-version lists, workarounds and release dates below should be checked against the primary pages before you act on them.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-105207: unauthenticated account takeover via IdP linking; affects 4.0.0 to 4.17.2 and 3.0.0 to 3.4.15; fixed in 4.17.3; 3.x end-of-life; Login V1 unaffected</td>
          <td><a href="https://github.com/zitadel/zitadel/security/advisories/GHSA-g8gj-gq47-xgf4">GHSA-g8gj-gq47-xgf4</a>; <a href="https://cveawg.mitre.org/api/cve/CVE-2026-105207">CVE record</a></td>
          <td>Qualified: CVE record&rsquo;s structured data contains an inconsistent second range</td>
      </tr>
      <tr>
          <td>CVE-2026-105209: cross-organization takeover via passkey enrollment; affects 4.0.0 to 4.17.0 and 3.0.0 to 3.4.14; fixed in 4.17.1 and 3.4.15; no complete workaround</td>
          <td><a href="https://github.com/zitadel/zitadel/security/advisories/GHSA-pq2q-2c6r-75c4">GHSA-pq2q-2c6r-75c4</a>; <a href="https://cveawg.mitre.org/api/cve/CVE-2026-105209">CVE record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-105215: pre-hijacking in Login V1; affects 4.0.0 to 4.16.1 and 3.0.0 to 3.4.13; fixed in 4.16.2 and 3.4.14; Login V2 unaffected</td>
          <td><a href="https://github.com/zitadel/zitadel/security/advisories/GHSA-738m-7888-jfv8">GHSA-738m-7888-jfv8</a>; <a href="https://cveawg.mitre.org/api/cve/CVE-2026-105215">CVE record</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Advisory scores 9.8, 9.6, 9.1; CVE records carry CVSS 3.1 scores of 9.8, 9.6, 9.1 and CVSS 4.0 scores of 9.3</td>
          <td>The three advisories and CVE records above</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE records published October 4, 2026 by VulnCheck; public dates July 29, August 14 and September 4, 2026</td>
          <td>CVE records above</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>4.16.2 and 3.4.14 released July 29, 2026; 3.4.15 and 4.17.1 released August 14, 2026; 4.17.3 released September 4, 2026; 4.19.2 (September 28, 2026) addresses three further vulnerabilities and requires a Login V2 session cookie secret; 4.19.4 dated October 1, 2026</td>
          <td><a href="https://github.com/zitadel/zitadel/releases">ZITADEL releases</a>; <a href="https://github.com/zitadel/zitadel/releases/tag/v4.19.2">4.19.2 release notes</a></td>
          <td>Verified: as read October 6, 2026</td>
      </tr>
      <tr>
          <td>Partial workarounds: disable external IdP linking (105207, Login V2 only); disable manual account creation on external IdPs (105215)</td>
          <td>GHSA-g8gj-gq47-xgf4; GHSA-738m-7888-jfv8</td>
          <td>Qualified: vendor advisory; not tested</td>
      </tr>
      <tr>
          <td>Not in CISA KEV (catalog 2026.10.04); CISA enrichment lists exploitation &ldquo;none&rdquo; for CVE-2026-105209 and CVE-2026-105215 as of October 5, 2026</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities.json">CISA KEV data</a>; CVE records</td>
          <td>Qualified: point-in-time; no CISA enrichment seen for CVE-2026-105207</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>NetScaler CVE-2026-88779: Exploited SAML Flaw Needs Newer Builds Than Last Week's Zero-Day Fixes</title><link>https://firewallsync.com/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/</link><guid>https://firewallsync.com/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/</guid><pubDate>Tue, 06 Oct 2026 06:30:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>CVE-2026-88779 can knock SAML-configured NetScaler ADC and Gateway appliances offline and is on CISA's exploited list. The fixed builds are newer than the ones that fixed last week's two zero-days, so an appliance patched on September 27 can still be vulnerable.</description><enclosure url="https://firewallsync.com/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/cover_hu15350512838332202912.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/cover_hu15350512838332202912.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/netscaler-cve-2026-88779-saml-denial-of-service-exploited-patched-last-weeks-zero-days/cover_hu15350512838332202912.jpg" alt="Server rack with network cables attached" width="1200" height="630"></p><p>If you run Citrix NetScaler ADC or NetScaler Gateway with SAML authentication, check your build number today. CVE-2026-88779, a memory overflow that can take a SAML-configured appliance offline, is being exploited, and the fix is in builds newer than the ones Citrix shipped for last week&rsquo;s two zero-days. An appliance upgraded to 14.1-73.37 or 13.1-64.23 on or after September 27 is still inside the affected range.</p>
<p>This article separates what Citrix, CISA and the security researchers have each confirmed, because early reporting mixed denial of service with remote code execution. For the two earlier flaws, see our <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">patch-and-hunt guide for CVE-2026-88771 and CVE-2026-88772</a>.</p>
<h2 id="what-is-confirmed">What is confirmed</h2>
<p>NetScaler, acting as its own CVE Numbering Authority (the organization authorized to assign and describe CVE IDs for its products), published the record for CVE-2026-88779 on October 4, 2026 at 02:35 UTC. The record&rsquo;s public date is 02:19 UTC on October 4, which is the evening of October 3 in US time zones. Sources that say Citrix published its bulletin, CTX697174, on &ldquo;October 3&rdquo; and the CVE record&rsquo;s &ldquo;October 4&rdquo; describe the same event in different time zones.</p>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-88779</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Product</td>
          <td>NetScaler ADC and NetScaler Gateway (on-premises)</td>
      </tr>
      <tr>
          <td>Weakness</td>
          <td>Memory buffer bounds error (CWE-119, as classified by CISA), reported as a memory overflow leading to denial of service</td>
      </tr>
      <tr>
          <td>CVSS 4.0 base score</td>
          <td>8.7 (High); attack over the network, low complexity, no privileges, no user interaction; impact on availability only</td>
      </tr>
      <tr>
          <td>Affected builds</td>
          <td>ADC: before 14.1-73.41, before 13.1-64.28, before 14.1-73.41 FIPS, before 13.1-37.282. Gateway: before 14.1-73.41 and before 13.1-64.28</td>
      </tr>
      <tr>
          <td>Fixed in</td>
          <td>14.1-73.41, 13.1-64.28, 14.1-73.41 FIPS, 13.1-37.282 (FIPS and NDcPP ADC builds)</td>
      </tr>
      <tr>
          <td>Configuration condition</td>
          <td>Appliance configured as a SAML service provider or SAML identity provider, per watchTowr and The Hacker News</td>
      </tr>
      <tr>
          <td>CISA KEV added</td>
          <td>October 4, 2026; due date October 7, 2026</td>
      </tr>
  </tbody>
</table>
<p>CVSS (Common Vulnerability Scoring System) is the standard 0 to 10 severity scale; this score uses version 4.0. KEV is CISA&rsquo;s Known Exploited Vulnerabilities catalog.</p>
<h3 id="why-last-weeks-patch-level-is-not-enough">Why last week&rsquo;s patch level is not enough</h3>
<p>The earlier bulletin, CTX697096 of September 27, 2026, fixed CVE-2026-88771 through CVE-2026-88778 in builds 14.1-73.37 and 13.1-64.23 (and 14.1-73.37 FIPS and 13.1-37.279 for FIPS and NDcPP ADC). The new record uses &ldquo;before&rdquo; ranges that start higher:</p>
<table>
  <thead>
      <tr>
          <th>Branch</th>
          <th>Fixes the September 27 bulletin (CVE-2026-88771 to -88778)</th>
          <th>Fixes CVE-2026-88779 (October 3-4 bulletin)</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>14.1</td>
          <td>14.1-73.37</td>
          <td>14.1-73.41</td>
      </tr>
      <tr>
          <td>13.1</td>
          <td>13.1-64.23</td>
          <td>13.1-64.28</td>
      </tr>
      <tr>
          <td>14.1 FIPS</td>
          <td>14.1-73.37 FIPS</td>
          <td>14.1-73.41 FIPS</td>
      </tr>
      <tr>
          <td>13.1 FIPS and NDcPP (ADC only)</td>
          <td>13.1-37.279</td>
          <td>13.1-37.282</td>
      </tr>
  </tbody>
</table>
<p>An appliance on 14.1-73.37, for example, is earlier than 14.1-73.41, so the record lists it as affected. SecurityWeek reported that appliances already patched for the earlier flaws were hit; that matches the version arithmetic. The practical consequence is that &ldquo;we patched the NetScaler last week&rdquo; is no longer a complete answer, and the target build is the newer one.</p>
<p>Two limits apply. The Gateway records list only the 14.1 and 13.1 non-FIPS builds, which is why the FIPS builds are shown for ADC only. And we did not test any build; this is the reading of the published ranges.</p>
<h2 id="what-attackers-can-do-with-it-and-what-is-unconfirmed">What attackers can do with it, and what is unconfirmed</h2>
<p>The evidence supports one consequence: denial of service. CISA&rsquo;s catalog entry says the flaw &ldquo;could allow for a denial of service&rdquo;. Citrix said, as quoted by The Hacker News, that it observed targeted attacks on unmitigated deployments that can lead to denial of service, and that it had not identified an impact on the integrity of customer data. Help Net Security also quotes Citrix as saying that if the condition is triggered repeatedly, the service may remain unavailable.</p>
<p>Several details in early coverage deserve caution:</p>
<ul>
<li><strong>Remote code execution.</strong> Help Net Security says attackers reportedly attempt to download and install web shells. SOC Prime, citing SecurityWeek, describes authentication requests containing shell commands in username fields and a researcher&rsquo;s report of a malware binary downloaded on one honeypot. These are described as attempts. None of the sources we read shows code execution through CVE-2026-88779 itself. SecurityWeek reported that watchTowr concluded the flaw is a denial-of-service issue that can only crash systems. The earlier flaws, CVE-2026-88771 and CVE-2026-88772, are the ones for which web shell deployment has been documented, and some of that traffic may be aimed at them.</li>
<li><strong>Why attackers would crash appliances.</strong> watchTowr suspects, according to SecurityWeek, that crashing appliances may be intended to speed up exploitation of CVE-2026-88771. That is a researcher&rsquo;s hypothesis, not a confirmed attacker goal.</li>
<li><strong>A reboot after a set number of crashes.</strong> One summary we read claims the appliance reboots after a particular number of crashes. We could not confirm that in Citrix&rsquo;s or watchTowr&rsquo;s own text, so we are not repeating the number.</li>
<li><strong>Which flaw caused a given crash.</strong> Administrators told reporters that patched appliances were rebooting and that interim workarounds sometimes failed to stop the crashes, per SecurityWeek. Those are reports from the field, not measurements of a specific flaw.</li>
</ul>
<h3 id="what-cisas-data-says">What CISA&rsquo;s data says</h3>
<p>CISA added the flaw to KEV on October 4, 2026 with a due date of October 7, marked forensic triage as required, and listed ransomware campaign use as &ldquo;Unknown&rdquo;. The due date comes from Binding Operational Directive 26-04 and applies to US Federal Civilian Executive Branch agencies, not to private organizations; CISA encourages everyone to adopt the same risk-based approach.</p>
<p>CISA&rsquo;s own enrichment of the CVE record rates exploitation as &ldquo;active&rdquo;, the flaw as automatable by an attacker (&ldquo;yes&rdquo;), and technical impact as &ldquo;partial&rdquo;. The Exploit Prediction Scoring System (EPSS), a model that estimates the chance of exploitation in the next 30 days, gave the flaw 0.00534 (about 0.5%, 43rd percentile) on October 5, 2026. That number is low despite confirmed exploitation. Use it as a reminder that EPSS lags new disclosures, not as a reason to wait.</p>
<h2 id="what-to-do-in-order">What to do, in order</h2>
<p>This sequence is our recommendation, built on the vendor and researcher guidance cited above.</p>
<h3 id="1-establish-whether-saml-is-configured">1. Establish whether SAML is configured</h3>
<p>watchTowr says the condition is a SAML service provider or identity provider setup. WorkOS suggests searching the configuration file for the two relevant commands; the search itself changes nothing:</p>
<pre tabindex="0"><code>grep -E &#34;add authentication saml(Action|IdPProfile)&#34; /nsconfig/ns.conf
</code></pre><p>Run it from the appliance shell and read the result against the Citrix bulletin. A match means the appliance meets the described condition. No match is encouraging but the safe default is still to move to the fixed build, since every build you would choose in the next maintenance window should be at or above the newer ones.</p>
<h3 id="2-upgrade-to-the-new-fixed-build">2. Upgrade to the new fixed build</h3>
<p>Target 14.1-73.41 or later, or 13.1-64.28 or later (or the FIPS and NDcPP builds in the table above). Confirm exact build strings on Citrix&rsquo;s download page before upgrading FIPS or NDcPP appliances, because our earlier article noted inconsistent formatting of the 13.1 FIPS string in CVE records. Versions 12.1 and 13.0 are end-of-life, and we found no statement on whether they are affected; plan the migration rather than assuming they are not.</p>
<h3 id="3-apply-interim-mitigation-if-you-cannot-upgrade-today">3. Apply interim mitigation if you cannot upgrade today</h3>
<p>watchTowr and Help Net Security both point to Citrix&rsquo;s Global Deny List signatures, delivered through NetScaler Console, as the interim step; WorkOS also mentions responder policies available from Citrix Support. SecurityWeek&rsquo;s reporting that workarounds sometimes failed to stop the crashes means mitigation buys time and is not a substitute for the fixed build.</p>
<h3 id="4-look-for-crashes-then-for-what-came-with-them">4. Look for crashes, then for what came with them</h3>
<p>WorkOS lists crashes in <code>/var/log/ns.log</code> or <code>/var/log/messages</code>, repeated daemon restarts, &ldquo;Pitboss declaring system failure&rdquo; messages and unexplained reboots since late September as signs to look for. Help Net Security reports that Citrix offers an indicator-of-compromise script run through NetScaler Console. We did not run either.</p>
<p>A denial-of-service flaw should not close the investigation. If an appliance rebooted unexpectedly in the past two weeks, check it for the configuration theft and web shell activity described in our <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">CVE-2026-88771 and CVE-2026-88772 article</a>. Preserve logs first; our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents</a> explains why off-box copies matter.</p>
<h3 id="5-decide-what-an-outage-costs-you">5. Decide what an outage costs you</h3>
<p>Many organizations authenticate staff to everything through the same NetScaler SAML configuration. A flaw that can crash it is an availability risk to every application behind it. Whether you can reach critical systems when the gateway is down is worth testing, and our <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">vulnerability backlog</a> piece covers how to rank exposed and exploited items ahead of the rest. The general sequence for flaws attacked before a patch plan existed is in our <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">zero-day response guide</a>.</p>
<h2 id="what-to-take-from-this">What to take from this</h2>
<p>Two security bulletins in under a week for one product means the version you recorded last week is stale. Track the fixed build per advisory, not &ldquo;patched&rdquo; as a status, and recheck the product&rsquo;s bulletin page after any exploited-flaw announcement before closing the ticket.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 6, 2026. We did not test the vulnerability, run any script or access a NetScaler appliance. Citrix&rsquo;s own bulletin pages (CTX697174 and the Citrix community post) returned an access error from our environment, so Citrix&rsquo;s statements are quoted through the CVE record, CISA data and the secondary reports named in the table.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Record published October 4, 2026 02:35 UTC; affects ADC before 14.1-73.41, 13.1-64.28, 14.1-73.41 FIPS and 13.1-37.282, and Gateway before 14.1-73.41 and 13.1-64.28</td>
          <td><a href="https://cveawg.mitre.org/api/cve/CVE-2026-88779">CVE record via the CVE Program API</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS 4.0 base score 8.7 (High), availability impact only; CWE-119</td>
          <td>Same CVE record</td>
          <td>Verified: score assigned by NetScaler; CWE assigned by CISA, not NetScaler</td>
      </tr>
      <tr>
          <td>CISA KEV added October 4, 2026; due October 7; forensic triage required; ransomware use unknown; description says it could allow denial of service</td>
          <td><a href="https://github.com/cisagov/kev-data/blob/main/known_exploited_vulnerabilities.json">CISA KEV data</a>, catalog version 2026.10.04; <a href="https://www.cisa.gov/news-events/alerts/2026/10/04/cisa-adds-one-known-exploited-vulnerability-catalog">CISA alert</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 applies to US federal civilian agencies; CISA encourages all organizations to adopt the approach</td>
          <td>CISA alert above</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA enrichment: exploitation active, automatable yes, technical impact partial</td>
          <td>CVE record, CISA ADP section</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>September 27 bulletin fixed builds were 14.1-73.37 and 13.1-64.23</td>
          <td><a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">FirewallSync article on CVE-2026-88771 and CVE-2026-88772</a>, from NetScaler&rsquo;s CVE records</td>
          <td>Verified: our earlier reporting</td>
      </tr>
      <tr>
          <td>Condition: SAML service provider or identity provider</td>
          <td><a href="https://watchtowr.com/intelligence/citrix-netscaler-denial-of-service-memory-overflow-cve-2026-88779/">watchTowr, October 5, 2026</a>; <a href="https://thehackernews.com/2026/10/new-netscaler-zero-day-exploited-in.html">The Hacker News</a></td>
          <td>Qualified: not stated in NetScaler&rsquo;s CVE record</td>
      </tr>
      <tr>
          <td>Citrix: targeted attacks on unmitigated deployments can lead to denial of service; no impact on customer data integrity identified</td>
          <td>The Hacker News; <a href="https://www.securityweek.com/exploitation-of-citrix-netscaler-zero-day-hits-appliances-patched-days-earlier/">SecurityWeek</a></td>
          <td>Qualified: quoted second-hand; Citrix pages not accessible</td>
      </tr>
      <tr>
          <td>watchTowr concluded denial of service only; suspects use to speed exploitation of CVE-2026-88771</td>
          <td>SecurityWeek</td>
          <td>Qualified: researcher assessment and hypothesis</td>
      </tr>
      <tr>
          <td>Patched appliances reported rebooting; workarounds sometimes failed</td>
          <td>SecurityWeek</td>
          <td>Qualified: administrator reports</td>
      </tr>
      <tr>
          <td>Global Deny List signatures as interim mitigation</td>
          <td>watchTowr post; <a href="https://www.helpnetsecurity.com/2026/10/05/cisa-flags-new-exploited-netscaler-flaw-as-attackers-crash-appliances-cve-2026-88779/">Help Net Security, October 5, 2026</a></td>
          <td>Qualified: Citrix guidance not directly read</td>
      </tr>
      <tr>
          <td>Log indicators and configuration search command</td>
          <td><a href="https://workos.com/blog/netscaler-saml-zero-day-cve-2026-88779">WorkOS</a></td>
          <td>Qualified: third-party guidance, not run</td>
      </tr>
      <tr>
          <td>EPSS 0.00534 (43rd percentile) on October 5, 2026</td>
          <td><a href="https://api.first.org/data/v1/epss?cve=CVE-2026-88779">FIRST EPSS API</a></td>
          <td>Verified: figure changes daily</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Cisco Catalyst SD-WAN Manager CVE-2026-76504: Fixed Releases and Log Checks</title><link>https://firewallsync.com/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/</link><guid>https://firewallsync.com/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/</guid><pubDate>Mon, 05 Oct 2026 09:05:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>Cisco says an unauthenticated authentication bypass in Catalyst SD-WAN Manager is being exploited. There is no workaround, only upgrades, a temporary shield and a short list of log entries worth checking today.</description><enclosure url="https://firewallsync.com/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/cover_hu14211424645131340512.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/cover_hu14211424645131340512.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/cisco-catalyst-sd-wan-manager-cve-2026-76504-authentication-bypass-fixed-releases-and-log-checks/cover_hu14211424645131340512.jpg" alt="Fiber optic cables plugged into a network switch" width="1200" height="630"></p><p>If you run Cisco Catalyst SD-WAN Manager (formerly vManage) on your own infrastructure, you need to do two things today: confirm whether it is reachable from the internet, and look at two log files for requests that mention <code>j_security_check</code>. Cisco says the flaw, CVE-2026-76504, lets an unauthenticated attacker act as the <code>admin</code> user through the manager&rsquo;s API, and that its security team became aware of active exploitation in September 2026. The only remediation is a software upgrade.</p>
<h2 id="what-cisco-has-published">What Cisco has published</h2>
<p>Cisco published advisory <a href="https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-sdwan-webauth-xr8beuuU">cisco-sa-sdwan-webauth-xr8beuuU</a> on September 30, 2026. Version 1.1, dated October 2, 2026, added information about a temporary protection called Live Protect. Cisco rates it Critical.</p>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-76504</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Product</td>
          <td>Cisco Catalyst SD-WAN Manager (the management and control interface for a Cisco SD-WAN network)</td>
      </tr>
      <tr>
          <td>Weakness</td>
          <td>Improper handling of URI encoding (CWE-177) in API session-based authentication</td>
      </tr>
      <tr>
          <td>Access needed</td>
          <td>None. The attacker does not need an account</td>
      </tr>
      <tr>
          <td>What the attacker gains</td>
          <td>Access to the API with the privileges of the <code>admin</code> user</td>
      </tr>
      <tr>
          <td>CVSS 3.1 base score</td>
          <td>9.8 (Critical), as assigned by Cisco</td>
      </tr>
      <tr>
          <td>Affected systems</td>
          <td>SD-WAN Manager &ldquo;regardless of system configuration&rdquo;</td>
      </tr>
      <tr>
          <td>Workaround</td>
          <td>None</td>
      </tr>
      <tr>
          <td>Exploitation</td>
          <td>Cisco PSIRT became aware of active exploitation in September 2026. The advisory gives no more precise date</td>
      </tr>
      <tr>
          <td>How it was found</td>
          <td>During the resolution of a Cisco Technical Assistance Center (TAC) support case</td>
      </tr>
  </tbody>
</table>
<p>CVSS (Common Vulnerability Scoring System) is the standard 0 to 10 severity scale; this score uses version 3.1.</p>
<p>Cisco describes the mechanism this way: improper handling of URI encoding in an HTTP request lets a request bypass an authentication rule meant to restrict access to a specific API endpoint. URI encoding replaces a character with a percent sign and two hexadecimal digits, so <code>j</code> can be written <code>%6a</code>. Cisco&rsquo;s description implies the check looks at the encoded form while the application later treats it as the plain character. That last step is our reading of the mechanism, not a statement in the advisory.</p>
<h2 id="which-releases-are-fixed">Which releases are fixed</h2>
<p>Cisco&rsquo;s advisory gives these first fixed releases:</p>
<table>
  <thead>
      <tr>
          <th>SD-WAN software release train</th>
          <th>First fixed release</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Earlier than 20.9</td>
          <td>Migrate to a fixed release</td>
      </tr>
      <tr>
          <td>20.9</td>
          <td>20.9.10.1</td>
      </tr>
      <tr>
          <td>20.12</td>
          <td>20.12.8.2</td>
      </tr>
      <tr>
          <td>20.15</td>
          <td>20.15.6.1</td>
      </tr>
      <tr>
          <td>20.18</td>
          <td>20.18.4.1</td>
      </tr>
      <tr>
          <td>26.1</td>
          <td>26.1.2.1</td>
      </tr>
      <tr>
          <td>26.2</td>
          <td>26.2.1</td>
      </tr>
  </tbody>
</table>
<p>Two points on reading this table. First, the table lists only these trains. If you run a train that is not in it, the advisory neither marks that train fixed nor says it is unaffected, and we could not find a Cisco statement on it. Treat it as unresolved and ask Cisco TAC rather than assume it is covered. Second, Cisco says it validates only the affected and fixed release information in the advisory, so use the advisory, not a scanner&rsquo;s interpretation of the CVE record, to decide.</p>
<p>For Cisco-hosted deployments, Cisco says the flaw is addressed in SD-WAN Cloud (Cisco Managed) release 20.15.605 with no customer action needed.</p>
<h2 id="the-live-protect-shield-what-it-is-and-is-not">The Live Protect shield: what it is and is not</h2>
<p>Cisco&rsquo;s October 2 update says it released a Live Protect shield for this CVE &ldquo;to provide temporary security coverage to allow time for software upgrade planning&rdquo;. Cisco is explicit about the limits: it &ldquo;offers only temporary partial protection&rdquo; and &ldquo;the only way to remediate this vulnerability is to upgrade to the first fixed software release&rdquo;. It also warns of a side effect: a legitimate user who uses URI encoding might not be able to log in to SD-WAN Manager once the shield is applied.</p>
<p>Our reading: use the shield, if you use it at all, to buy days while you schedule the upgrade, not as the end state. &ldquo;Partial&rdquo; means Cisco is not claiming it blocks every variant. Cisco&rsquo;s documentation for the feature is <a href="https://www.cisco.com/c/en/us/td/docs/routers/sdwan/26x-later/network-monitoring/network-monitoring-guide/live-protect-for-cisco-catalyst-sd-wan.html">Live Protect for Cisco Catalyst SD-WAN</a>. We did not test the shield.</p>
<h2 id="reduce-exposure-first">Reduce exposure first</h2>
<p>Cisco&rsquo;s mitigation for on-premises deployments is to restrict access from unsecured networks, including the internet. If you need internet access to the manager, Cisco says to restrict it to known, trusted hosts on the ports and protocols in its user guides, and to place the control components behind a filtering device such as a firewall. Cisco&rsquo;s advisory says managers &ldquo;exposed to the internet&rdquo; with ports exposed are at risk.</p>
<p>Practical steps that follow from that:</p>
<ul>
<li>Find out whether the manager&rsquo;s web and API ports are reachable from the internet. Check from outside your network, not from the configuration alone.</li>
<li>If they are, restrict them to a management network or a named allowlist before you do anything else. This does not fix the flaw, but it removes the unauthenticated attacker&rsquo;s path while you plan the upgrade.</li>
<li>Remember what the manager controls. SD-WAN Manager is the administrative plane for the WAN, so admin API access is access to the network&rsquo;s configuration. That is our assessment of impact, not a Cisco statement of what an attacker did.</li>
</ul>
<h2 id="check-for-exploitation">Check for exploitation</h2>
<p>Cisco says its indicators may also occur during normal operations, so assess them against your usual traffic. It names two files:</p>
<ol>
<li><strong><code>/var/log/nms/containers/service-proxy/serviceproxy-access.log</code>.</strong> Look for entries related to <code>j_security_check</code> from unknown or unauthorized IP addresses. Cisco&rsquo;s example is a <code>POST /%6a_security_check</code> request.</li>
<li><strong><code>/var/log/nms/vmanage-server.log</code>.</strong> Look for entries related to <code>j_security_check</code> from unknown or unauthorized IP addresses, specifically for users whose names start with <code>viptela-reserved-</code>. Cisco&rsquo;s example reads <code>Request Stored in Map is (/%6a_security_check) for user (viptela-reserved-..)</code>.</li>
</ol>
<p><code>viptela-reserved-</code> accounts are system service accounts that Cisco documents in the authentication section of its SD-WAN configuration guide. A request that authenticates as one of them from an outside address is the pattern Cisco is pointing to.</p>
<p>Cisco states the <code>%6a</code> form is only an example: &ldquo;the vulnerability will allow any one character that is encoded in the request to be used to exploit this.&rdquo; So searching for <code>%6a</code> alone will miss other variants. As an untested starting point (ours, not Cisco&rsquo;s), also list every request to the endpoint whose path contains a percent-encoded character:</p>
<pre tabindex="0"><code>grep -E &#39;POST /[^ ]*%[0-9A-Fa-f]{2}[^ ]*&#39; serviceproxy-access.log
</code></pre><p>Review the results against your normal logins. Expect some noise, because other legitimate requests can contain encoded characters.</p>
<p>Cisco&rsquo;s other guidance:</p>
<ul>
<li>To determine whether a system has been compromised, open a Cisco TAC case as Severity 3 with <code>CVE-2026-76504</code> in the title. Before opening it, run <code>request admin-tech</code> on the manager so the diagnostic file can go to TAC.</li>
<li>Cisco publishes Snort rule 67179 for this advisory, for teams that run Snort-based inspection in front of the manager.</li>
<li>Send manager logs to an external server and retain them long enough for investigations, which is also general hardening advice in the advisory. Our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents rather than dashboards</a> covers why.</li>
</ul>
<h2 id="upgrade-then-decide-what-else-to-rotate">Upgrade, then decide what else to rotate</h2>
<p>Cisco&rsquo;s advice is to upgrade to a fixed release as soon as possible. The advisory does not describe what an attacker with admin API access did in the observed attacks, and we found no public report that does. So the following is general incident response practice, not vendor guidance, and applies if your logs show a suspicious request:</p>
<ul>
<li>Treat the manager&rsquo;s configuration as readable and changeable by an outsider for the period of exposure. Review recent user, template and policy changes.</li>
<li>Rotate the administrator credentials and any API keys or integration credentials the manager holds.</li>
<li>Cisco&rsquo;s hardening list also says to change the default administrator password and to restrict the administrator account by creating named user accounts based on access needs. Do both if you have not.</li>
</ul>
<h2 id="what-the-cisa-entry-means">What the CISA entry means</h2>
<p>CISA added CVE-2026-76504 to its Known Exploited Vulnerabilities (KEV) catalog on September 30, 2026, the day Cisco published. The entry sets a due date of October 3, 2026, requires forensic triage, and lists ransomware campaign use as &ldquo;Unknown&rdquo;. CISA&rsquo;s enrichment data on the CVE record rates exploitation &ldquo;active&rdquo;, automatable &ldquo;yes&rdquo; and technical impact &ldquo;total&rdquo;. The due date comes from Binding Operational Directive 26-04 and applies to US Federal Civilian Executive Branch agencies; it does not bind private companies.</p>
<h2 id="what-this-adds-up-to">What this adds up to</h2>
<p>This section is editorial analysis.</p>
<p>CISA&rsquo;s catalog added Citrix NetScaler (September 27), this Cisco flaw (September 30), Fortinet FortiMail (October 1) and Zammad (October 2) within six days. They are different products and different bugs, but each is software reachable over the network that was already being exploited when the advisory appeared. We covered the <a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">FortiMail file-write zero-day</a>, the <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">Citrix NetScaler zero-days</a> and the <a href="/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/">Zammad zero-days</a> separately. Three habits shorten the next response:</p>
<ul>
<li><strong>Keep management planes off the public internet.</strong> Cisco&rsquo;s own mitigation is this. If the only thing standing between the internet and admin access is a login check, one encoding bug removes it.</li>
<li><strong>Know your release train before the advisory arrives.</strong> The table above only helps if you can say in a minute which train you run.</li>
<li><strong>Ship manager logs off the box.</strong> Cisco&rsquo;s indicators are log lines. They are only useful if the logs exist and an intruder could not have edited them. For triage order, see <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">what a zero-day is and how to respond to one</a> and <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">the vulnerability backlog that never shrinks</a>.</li>
</ul>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 5, 2026. We did not test the vulnerability, the Live Protect shield or the grep pattern on a Cisco SD-WAN Manager. We could not open CISA&rsquo;s website from our environment, so CISA data was read from CISA&rsquo;s published KEV data file and from CISA&rsquo;s entry embedded in the CVE record.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Unauthenticated remote attacker can gain admin-level API access via URI-encoding handling flaw; affects SD-WAN Manager regardless of configuration</td>
          <td><a href="https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-sdwan-webauth-xr8beuuU">Cisco advisory cisco-sa-sdwan-webauth-xr8beuuU</a>, first published September 30, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS 3.1 base score 9.8; CWE-177</td>
          <td>Cisco advisory; <a href="https://www.cve.org/CVERecord?id=CVE-2026-76504">CVE record</a>, updated October 2, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Cisco PSIRT became aware of active exploitation in September 2026</td>
          <td>Cisco advisory, Exploitation and Public Announcements</td>
          <td>Verified: no more precise date given</td>
      </tr>
      <tr>
          <td>First fixed releases 20.9.10.1, 20.12.8.2, 20.15.6.1, 20.18.4.1, 26.1.2.1, 26.2.1; earlier than 20.9 migrate</td>
          <td>Cisco advisory, Fixed Releases table</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>No fixed release is listed for trains not named in the table</td>
          <td>Cisco advisory, Fixed Releases table, as read October 5, 2026</td>
          <td>Qualified: absence of a listing, not a Cisco statement</td>
      </tr>
      <tr>
          <td>No workaround; mitigation is restricting access from unsecured networks; already deployed for cloud-hosted</td>
          <td>Cisco advisory</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Cloud (Cisco Managed) release 20.15.605 addresses the flaw</td>
          <td>Cisco advisory</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Live Protect shield added October 2, 2026; temporary, partial; may block legitimate URI-encoded logins</td>
          <td>Cisco advisory, version 1.1</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Log files and <code>j_security_check</code> / <code>viptela-reserved-</code> indicators; any single encoded character can be used</td>
          <td>Cisco advisory, Indicators of Compromise</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>TAC case at Severity 3 with CVE ID in title; run <code>request admin-tech</code> first; Snort rule 67179</td>
          <td>Cisco advisory</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Added to CISA KEV September 30, 2026; due October 3; forensic triage required; ransomware use unknown</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data</a>, catalog version 2026.10.04</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CISA rates exploitation active, automatable yes, technical impact total</td>
          <td>CISA ADP entry in the CVE record, dated September 30, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 binds US federal civilian agencies</td>
          <td><a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">Tenable analysis of BOD 26-04</a></td>
          <td>Qualified: secondary source</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Zammad Zero-Days CVE-2026-102489 and CVE-2026-102490: What Is Fixed and What Is Not</title><link>https://firewallsync.com/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/</link><guid>https://firewallsync.com/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/</guid><pubDate>Mon, 05 Oct 2026 09:00:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>Two Zammad flaws chained in the breach of the Dutch Institute for Vulnerability Disclosure are on CISA's exploited list. The sources disagree on which versions are safe, and no fix for the privilege-escalation flaw was visible as of October 5, 2026.</description><enclosure url="https://firewallsync.com/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/cover_hu4738225789966563780.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/cover_hu4738225789966563780.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/zammad-zero-days-cve-2026-102489-and-cve-2026-102490-what-is-fixed-and-what-is-not/cover_hu4738225789966563780.jpg" alt="Person wearing an earbud looking at code on a monitor" width="1200" height="630"></p><p>If you run Zammad, the helpdesk and ticketing system, check your version today. Anything on 6.5 or older is on a branch Zammad no longer patches and is the version range where the remote code execution flaw, CVE-2026-102489, can be used. Moving to Zammad 7.2.0, the version Zammad currently recommends, closes that path. It does not, on the evidence available as of October 5, 2026, settle the second flaw, CVE-2026-102490, which lets someone who already runs code as the <code>zammad</code> user become root.</p>
<p>This article separates what the CVE records, the discoverer, the vendor and CISA each say, because on this incident they do not fully agree.</p>
<h2 id="what-happened">What happened</h2>
<p>The Dutch Institute for Vulnerability Disclosure (DIVD) is a non-profit made up largely of volunteer security researchers. It was breached. In its public timeline for the incident (case DIVD-2026-00014), DIVD dates the first malicious access to September 21, 2026 and says it became aware of the activity on September 22. On September 30 it said the attackers got in through two previously unknown flaws in Zammad, an open-source helpdesk it ran, which it then tracked as CVE-2026-102489 and CVE-2026-102490. On October 1 it said volunteer data had been taken, including email addresses and possibly contact details.</p>
<p>DIVD also says the intrusion was run by an automated AI agent that chose each next step itself, and that the agent moved from the Zammad user to root &ldquo;in seconds&rdquo;. That is DIVD&rsquo;s own assessment from its forensic work with Merlon Security. We have not seen independent forensic evidence, and the trade press (Help Net Security) notes it is unknown whether this was part of a larger attack or a capability test. Treat the AI-agent description as DIVD&rsquo;s account.</p>
<p>DIVD&rsquo;s response gives a useful second detail. It says network segmentation and its incident team&rsquo;s actions stopped the attackers from going deeper into its systems.</p>
<h2 id="the-two-flaws">The two flaws</h2>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-102489</th>
          <th>CVE-2026-102490</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Type</td>
          <td>Session hijack (CISA labels the weakness session fixation, CWE-384)</td>
          <td>Local privilege escalation (improper privilege management, CWE-269)</td>
      </tr>
      <tr>
          <td>What it gives an attacker</td>
          <td>Remote code execution as the <code>zammad</code> user</td>
          <td>Root on the server, starting from the <code>zammad</code> user</td>
      </tr>
      <tr>
          <td>Access needed</td>
          <td>No account; CVSS marks user interaction as &ldquo;passive&rdquo;</td>
          <td>Already running as the local <code>zammad</code> user</td>
      </tr>
      <tr>
          <td>Severity scored by DIVD (CVSS 4.0)</td>
          <td>8.7 High on its own</td>
          <td>8.5 High on its own</td>
      </tr>
      <tr>
          <td>Severity when chained</td>
          <td>9.4 Critical</td>
          <td>9.4 Critical</td>
      </tr>
      <tr>
          <td>Affected versions per CVE record</td>
          <td>6.3.0 up to 6.5.4 exploitable; 7.0.0 to 7.1.3 present but not exploitable</td>
          <td>1.5.0 up to 7.1.0-alpha; text says &ldquo;all versions&rdquo;</td>
      </tr>
      <tr>
          <td>CVE record published</td>
          <td>September 30, 2026</td>
          <td>September 30, 2026</td>
      </tr>
      <tr>
          <td>CISA KEV added</td>
          <td>October 2, 2026</td>
          <td>October 2, 2026</td>
      </tr>
  </tbody>
</table>
<p>CVSS (Common Vulnerability Scoring System) is the standard 0 to 10 severity scale. These scores use version 4.0 and were assigned by DIVD, which also acted as the CVE Numbering Authority, the body that issued these CVE IDs.</p>
<p>Two details in secondary coverage are easy to get wrong. Some reports say both flaws score 9.4. In DIVD&rsquo;s records, 9.4 is the score for the chain; each flaw alone scores lower. And one report describes an attack going &ldquo;from unauthenticated access to root&rdquo;. That is the chain, not a single flaw: the first needs no account, the second needs the foothold the first provides.</p>
<h3 id="where-the-sources-disagree">Where the sources disagree</h3>
<p><strong>Is there a fix?</strong> Help Net Security reported on October 1 that both flaws were &ldquo;currently without a fix&rdquo;. DIVD&rsquo;s case file lists patch status as &ldquo;Available&rdquo; with the recommendation &ldquo;Upgrade to Zammad version 7&rdquo;. Zammad&rsquo;s statement says CVE-2026-102489 does not affect current versions and that it hardened the code anyway in 7.2.0. These are not the same claim. The consistent reading is that 7.x removes the remote path for the first flaw, and no fix has been announced for the second.</p>
<p><strong>Affected versions.</strong> The CVE record&rsquo;s structured data lists CVE-2026-102489 as affected from 6.3.0 up to, but not including, 6.5.4. DIVD&rsquo;s case file and the record&rsquo;s own description say 6.3.0 to 6.5.4. A scanner that reads only the structured field could skip 6.5.4. Zammad&rsquo;s wording is broader still: &ldquo;6.5 and older&rdquo;. Use Zammad&rsquo;s range if you are deciding whether to act.</p>
<p><strong>How the flaw was disclosed.</strong> DIVD&rsquo;s case file says it reported the vulnerabilities to Zammad on September 24, 2026 and scanned for exposed instances and created a limited disclosure on September 26. Zammad says it first received a report about the session issue in August 2026, and that DIVD had not given it technical details of the second flaw until October 1. Zammad also wrote that it does &ldquo;not consider this a responsible way to handle vulnerabilities&rdquo;. We cannot reconcile the August report with DIVD&rsquo;s September 24 date from the public record, so we state both. For readers the practical point is the same: the CVE IDs were public before the vendor had published fixes or an advisory for the second flaw.</p>
<p><strong>How exploitable is the second flaw?</strong> DIVD&rsquo;s case file says it affects &ldquo;all versions&rdquo;. Zammad said on October 1 that it &ldquo;cannot be exploited remotely on its own. An attacker would already need access to your server.&rdquo; Both can be true: it matters most as the second step of a chain, which is exactly how it was used.</p>
<h2 id="what-cisa-says">What CISA says</h2>
<p>CISA added both CVEs to its Known Exploited Vulnerabilities (KEV) catalog on October 2, 2026. The entries set a due date of October 5, 2026, mark forensic triage as required, and list ransomware campaign use as &ldquo;Unknown&rdquo;. CISA&rsquo;s required action is to apply mitigations per vendor instructions or stop using the product if mitigations are unavailable. The due date comes from Binding Operational Directive 26-04 and applies to US Federal Civilian Executive Branch agencies, not to private organizations.</p>
<p>CISA&rsquo;s own enrichment data on the CVE records rates exploitation as &ldquo;active&rdquo; and technical impact as &ldquo;total&rdquo; for both. It rates the first flaw &ldquo;automatable: yes&rdquo; and the second &ldquo;automatable: no&rdquo;, which fits the difference above: one can be launched against many servers over the network, the other needs a foothold.</p>
<p>One data point worth knowing, with its limits. The Exploit Prediction Scoring System (EPSS), a model that estimates the probability a CVE will be exploited in the next 30 days, gave CVE-2026-102489 a score of 0.0140 (about 1.4%, 71st percentile) and CVE-2026-102490 0.0063 (about 0.6%, 48th percentile) on October 4, 2026. Exploitation has already been confirmed for both. A low EPSS number soon after disclosure reflects how little public data the model had, not safety. Do not use it to lower priority for a flaw on CISA&rsquo;s exploited list.</p>
<h2 id="what-to-do-in-order">What to do, in order</h2>
<p>This sequence is our recommendation, built on DIVD&rsquo;s, Zammad&rsquo;s and the Dutch NCSC&rsquo;s published advice as cited. Vendor advice is marked as such.</p>
<h3 id="1-find-every-zammad-instance-and-its-version">1. Find every Zammad instance and its version</h3>
<p>Include test, staging and forgotten copies. DIVD&rsquo;s recommendation covers any version, and its affected range for the root escalation begins at 1.5.0. Record whether each one is reachable from the internet.</p>
<h3 id="2-preserve-logs-before-you-upgrade">2. Preserve logs before you upgrade</h3>
<p>The Dutch National Cyber Security Centre (NCSC-NL) advised copying application and network logs before updating, because they may help you check later for abuse of the second flaw, as reported by Help Net Security. Do this first. Our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents rather than dashboards</a> covers why off-server copies matter.</p>
<h3 id="3-remove-the-remote-entry-point">3. Remove the remote entry point</h3>
<p>Zammad recommends moving to 7.2.0, released September 23, 2026 according to its release page. If you cannot upgrade today, DIVD&rsquo;s alternative is to take the instance offline. Our addition: if the helpdesk must stay up for staff, restrict it to a VPN or an allowlist of source addresses until you have upgraded. If you are on 6.5 or older, treat the upgrade as overdue regardless of this incident, since Zammad says those versions get no security fixes.</p>
<h3 id="4-check-for-compromise-even-after-upgrading">4. Check for compromise, even after upgrading</h3>
<p>DIVD published a log-check script for indicators of compromise, linked from its <a href="https://csirt.divd.nl/cases/DIVD-2026-00015/">case file</a>. We did not run it. Read any script before running it against production logs, and run it on a copy. A clean result lowers the likelihood of this specific activity. It does not prove a server was never accessed, and DIVD itself says it assumes breach until it can prove otherwise.</p>
<p>Upgrading after an intrusion does not remove an attacker who already has root. If the logs show suspicious sessions or code execution, plan a rebuild rather than a patch.</p>
<h3 id="5-assume-the-ticket-system-holds-more-than-tickets">5. Assume the ticket system holds more than tickets</h3>
<p>This part is analysis, not vendor guidance. A helpdesk typically stores customer emails and attachments, internal notes, and credentials for connected channels such as mailboxes and chat. If a server may have been reached, rotate those integration credentials and API tokens, and revoke active sessions. Our <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">zero-day response guide</a> covers the general order of work.</p>
<h3 id="6-watch-for-the-second-fix">6. Watch for the second fix</h3>
<p>Follow Zammad&rsquo;s <a href="https://github.com/zammad/zammad/security/advisories">GitHub security advisories</a> for CVE-2026-102490. When a fixed version is named, that becomes the target, not &ldquo;version 7&rdquo; in general.</p>
<h2 id="the-lesson-that-does-not-depend-on-zammad">The lesson that does not depend on Zammad</h2>
<p>Two points from DIVD&rsquo;s account apply to any internal tool, offered here as our reading of it.</p>
<ul>
<li><strong>Low-severity entry points still matter when a second step exists.</strong> The first flaw lands on a service account with limited rights. The second turns that into root. Defenders who triage each CVE alone can underrate a chain. When two advisories name the same product on the same day, read them together.</li>
<li><strong>Segmentation is what limited the damage.</strong> DIVD credits it for stopping lateral movement. A ticketing server that can reach everything is a larger risk than the ticketing software itself.</li>
</ul>
<p>The same pattern of an edge or internal service exploited before a full patch plan existed appeared in the <a href="/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/">FortiMail zero-day</a> and the <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">Citrix NetScaler zero-days</a> in recent days, and the <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">vulnerability backlog</a> is where &ldquo;exposed and exploited&rdquo; should jump the queue.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 5, 2026. We did not test either vulnerability, run DIVD&rsquo;s script or install Zammad. We could not open Zammad&rsquo;s GitHub security advisories page or CISA&rsquo;s website from our environment, so CISA data was read from CISA&rsquo;s published KEV data file and from CISA&rsquo;s entries embedded in the CVE records.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>DIVD dates first malicious access to September 21, 2026; report to Zammad September 24; limited disclosure September 26</td>
          <td><a href="https://csirt.divd.nl/cases/DIVD-2026-00014/">DIVD case DIVD-2026-00014</a> and <a href="https://csirt.divd.nl/cases/DIVD-2026-00015/">DIVD-2026-00015</a></td>
          <td>Verified: DIVD&rsquo;s own timeline</td>
      </tr>
      <tr>
          <td>Attackers used CVE-2026-102489 and CVE-2026-102490; volunteer email addresses and possibly contact details taken</td>
          <td>DIVD case DIVD-2026-00014, statements of September 30 and October 1, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Attack was run by an AI agent</td>
          <td>DIVD case DIVD-2026-00014; <a href="https://www.helpnetsecurity.com/2026/10/01/divd-agentic-ai-attack-breach/">Help Net Security, October 1, 2026</a></td>
          <td>Qualified: DIVD&rsquo;s assessment; no independent forensic confirmation seen</td>
      </tr>
      <tr>
          <td>CVE-2026-102489: 6.3.0 to 6.5.4 exploitable; 7.0.0 to 7.1.3 present but not exploitable</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-102489">CVE record</a>, updated October 3, 2026; DIVD-2026-00015</td>
          <td>Qualified: structured data says &ldquo;before 6.5.4&rdquo;, text says &ldquo;to 6.5.4&rdquo;</td>
      </tr>
      <tr>
          <td>CVE-2026-102490: local zammad user to root; affects 1.5.0 up to 7.1.0-alpha</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-102490">CVE record</a>, updated October 3, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Scores: 8.7 and 8.5 individually, 9.4 chained (CVSS 4.0, assigned by DIVD)</td>
          <td>CVE records for both CVEs</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Zammad: 7.0 and later not affected by CVE-2026-102489; hardened in 7.2.0; 6.5 and older receive no security fixes</td>
          <td><a href="https://community.zammad.org/t/take-care-local-privilege-escalation-cve-2026-102490-is-reported-as-being-actively-exploited/21297/2">Zammad community statement, October 1, 2026</a></td>
          <td>Verified as Zammad&rsquo;s statement</td>
      </tr>
      <tr>
          <td>Zammad: CVE-2026-102490 cannot be exploited remotely on its own</td>
          <td>Zammad community thread, post of October 1, 2026</td>
          <td>Verified as Zammad&rsquo;s statement</td>
      </tr>
      <tr>
          <td>Zammad 7.2 released September 23, 2026</td>
          <td><a href="https://zammad.com/en/product/releases/7-2">Zammad 7.2 release page</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>No fix named for CVE-2026-102490</td>
          <td>Absence in the sources above as of October 5, 2026</td>
          <td>Qualified: GitHub advisories page could not be opened</td>
      </tr>
      <tr>
          <td>Both CVEs added to CISA KEV October 2, 2026; due October 5; forensic triage required; ransomware use unknown</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV data</a>, catalog version 2026.10.04</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 binds US federal civilian agencies</td>
          <td><a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">Tenable analysis of BOD 26-04</a></td>
          <td>Qualified: secondary source</td>
      </tr>
      <tr>
          <td>EPSS 0.01396 and 0.00629 on October 4, 2026; EPSS predicts 30-day exploitation probability</td>
          <td><a href="https://api.first.org/data/v1/epss?cve=CVE-2026-102489,CVE-2026-102490">FIRST EPSS API</a>; <a href="https://www.first.org/epss/faq">FIRST EPSS FAQ</a></td>
          <td>Verified: figures change daily</td>
      </tr>
      <tr>
          <td>NCSC-NL advised copying logs before updating</td>
          <td>Help Net Security, October 1, 2026</td>
          <td>Qualified: secondary report of NCSC-NL advice</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>FortiMail CVE-2026-104286: Affected Builds, IBE Workaround</title><link>https://firewallsync.com/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/</link><guid>https://firewallsync.com/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/</guid><pubDate>Sun, 04 Oct 2026 06:30:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>Fortinet says an unauthenticated file-write flaw in FortiMail is being exploited, and as of October 4, 2026 its advisory still lists the fixed builds as upcoming. The workaround is one setting. The compromise check takes longer.</description><enclosure url="https://firewallsync.com/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/cover_hu7787959317508176642.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/cover_hu7787959317508176642.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/fortimail-zero-day-cve-2026-104286-no-patch-yet-what-to-do/cover_hu7787959317508176642.jpg" alt="Mail app icon showing a notification badge on a phone screen" width="1200" height="630"></p><p>If you run Fortinet FortiMail and the IBE service is switched on, switch it off today or take the webmail interface off the internet. Fortinet says the flaw tracked as CVE-2026-104286 is being exploited, and as of October 4, 2026 its advisory still describes the fixed builds as &ldquo;upcoming&rdquo;. Then check whether someone has already used it, because the attacks Fortinet describes change the appliance&rsquo;s mail archiving settings.</p>
<p>This article covers what Fortinet has confirmed, where the public records disagree with each other, and the order to work in.</p>
<h2 id="what-fortinet-disclosed">What Fortinet disclosed</h2>
<p>On October 1, 2026, Fortinet published advisory <a href="https://fortiguard.fortinet.com/psirt/FG-IR-26-175">FG-IR-26-175</a>. It describes the flaw as a path traversal weakness (CWE-22) combined with improper neutralization of a NULL byte (CWE-158) that &ldquo;may allow an unauthenticated attacker to write arbitrary files on the underlying system via crafted HTTP or HTTPS requests.&rdquo;</p>
<p>The advisory adds: &ldquo;This has been reported to be exploited in the wild, customers are urged to apply the workaround below.&rdquo;</p>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-104286</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Product</td>
          <td>Fortinet FortiMail (email security gateway)</td>
      </tr>
      <tr>
          <td>Weakness</td>
          <td>Path traversal (CWE-22) and NULL byte handling (CWE-158)</td>
      </tr>
      <tr>
          <td>Access needed</td>
          <td>None. The attacker does not need an account</td>
      </tr>
      <tr>
          <td>What the attacker gains</td>
          <td>The ability to write arbitrary files on the appliance. Fortinet lists the impact as &ldquo;Execute unauthorized code or commands&rdquo;</td>
      </tr>
      <tr>
          <td>CVSS 3.1 base score</td>
          <td>9.8 (Critical), as assigned by Fortinet</td>
      </tr>
      <tr>
          <td>Advisory published</td>
          <td>October 1, 2026</td>
      </tr>
      <tr>
          <td>Found by</td>
          <td>Fortinet&rsquo;s own product security team, per the advisory</td>
      </tr>
  </tbody>
</table>
<p>CVSS (Common Vulnerability Scoring System) is the industry-standard 0 to 10 severity scale; this score uses version 3.1.</p>
<p>The workaround and the indicators both point at one feature: IBE, or identity-based encryption. Fortinet&rsquo;s administration guide describes it as a scheme that &ldquo;uses identities (such as email addresses) to calculate encryption keys.&rdquo; In practice it is the feature that lets FortiMail send an encrypted message to an outside recipient, who then opens it through a web link. That web-facing piece is what the workaround turns off.</p>
<p>Fortinet&rsquo;s advisory does not say whether IBE is enabled by default, and we could not confirm it from the documentation we read. Check your own configuration rather than assuming either way.</p>
<h3 id="affected-and-fixed-builds-according-to-fortinets-advisory">Affected and fixed builds, according to Fortinet&rsquo;s advisory</h3>
<table>
  <thead>
      <tr>
          <th>Branch</th>
          <th>Affected</th>
          <th>Fortinet&rsquo;s stated solution</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>FortiMail 8.0</td>
          <td>8.0.0 through 8.0.1</td>
          <td>Upgrade to upcoming 8.0.2 or above</td>
      </tr>
      <tr>
          <td>FortiMail 7.6</td>
          <td>7.6.0 through 7.6.6</td>
          <td>Upgrade to upcoming 7.6.7 or above</td>
      </tr>
      <tr>
          <td>FortiMail 7.4</td>
          <td>7.4.0 through 7.4.8</td>
          <td>Upgrade to upcoming 7.4.9 or above</td>
      </tr>
      <tr>
          <td>FortiMail 7.2</td>
          <td>7.2.0 through 7.2.9</td>
          <td>Upgrade to branch 7.4 or above</td>
      </tr>
  </tbody>
</table>
<p>We read the advisory on October 4, 2026. Its timeline showed a single entry, &ldquo;2026-10-01: Initial publication&rdquo;, and the solution column still used the word &ldquo;upcoming&rdquo;. We found no Fortinet statement giving a release date. That status can change at any time, so read the advisory itself before you plan around it.</p>
<p>Note what this means for the 7.2 branch: Fortinet lists no fixed 7.2 build. The stated path is a migration to 7.4 or later, which is a larger change than a patch release.</p>
<h2 id="the-cve-record-does-not-match-the-advisory">The CVE record does not match the advisory</h2>
<p>This is the part most likely to cause a wrong decision this week.</p>
<p>The <a href="https://www.cve.org/CVERecord?id=CVE-2026-104286">CVE record</a> is published by Fortinet as the CVE Numbering Authority (the organization authorized to describe CVE IDs for its own products). We retrieved it on October 4, 2026; it was last updated on October 2, 2026. It disagrees with the advisory, and with itself:</p>
<table>
  <thead>
      <tr>
          <th>Field</th>
          <th>8.0 branch</th>
          <th>7.6 branch</th>
          <th>7.4 branch</th>
          <th>7.2 branch</th>
          <th>7.0 branch</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Advisory FG-IR-26-175: affected</td>
          <td>8.0.0 to 8.0.1</td>
          <td>7.6.0 to 7.6.6</td>
          <td>7.4.0 to 7.4.8</td>
          <td>7.2.0 to 7.2.9</td>
          <td>Not listed</td>
      </tr>
      <tr>
          <td>CVE record, text description: affected</td>
          <td>8.0.0 to 8.0.1</td>
          <td>7.6.0 to 7.6.6</td>
          <td>7.4.0 to 7.4.8</td>
          <td>7.2.0 to 7.2.9</td>
          <td>Not listed</td>
      </tr>
      <tr>
          <td>CVE record, structured version data: affected</td>
          <td>8.0.0 only</td>
          <td>7.6.0 to 7.6.5</td>
          <td>7.4.0 to 7.4.6</td>
          <td>7.2.0 to 7.2.9</td>
          <td>7.0.0 to 7.0.9</td>
      </tr>
      <tr>
          <td>Advisory: fixed in</td>
          <td>8.0.2</td>
          <td>7.6.7</td>
          <td>7.4.9</td>
          <td>None (move to 7.4)</td>
          <td>Not listed</td>
      </tr>
      <tr>
          <td>CVE record, solutions field: fixed in</td>
          <td>8.0.1</td>
          <td>7.6.6</td>
          <td>7.4.8</td>
          <td>7.2.10</td>
          <td>Not listed</td>
      </tr>
  </tbody>
</table>
<p>The CVE record&rsquo;s solutions field also mentions FortiRecorder, a different product that the advisory does not mention at all.</p>
<p>Why this matters: vulnerability scanners and asset inventories usually match on the structured version data, not on the advisory text. A tool reading that data could report FortiMail 8.0.1, 7.6.6, 7.4.7 or 7.4.8 as not affected, while Fortinet&rsquo;s advisory lists all four as affected. The record&rsquo;s solutions field would also lead you to treat 7.6.6 as the fix when the advisory says 7.6.7.</p>
<p>We do not know which source Fortinet will correct. Until it does, the cautious reading is the advisory&rsquo;s wider range, plus the 7.0 branch that only the CVE record lists. If your scanner says a FortiMail build on these branches is clean, verify the build number by hand.</p>
<h2 id="what-attackers-have-been-doing">What attackers have been doing</h2>
<p>Fortinet&rsquo;s advisory lists indicators of compromise rather than a narrative, but the indicators tell a story. Three kinds of entry appear in the system event log examples it publishes:</p>
<ul>
<li>A scheduled job running as root: a log line beginning <code>msg=&quot;(root) CMD (/bin/sh -c 'O=/migadmin ...</code>, with <code>ui=cron</code>.</li>
<li>An administrator logout with no interface recorded: <code>msg=&quot;User admin logged out from (null).&quot;</code></li>
<li>A new mail archive account pointing off the appliance: <code>msg=&quot;Added 'archive234' to 'archive account'&quot;</code> with <code>destination[remote]</code>, <code>remote-ip[79.141.169.187]</code> and <code>remote-directory[/uploads]</code>.</li>
</ul>
<p>The third one is the consequence to worry about. An archive account with a remote destination sends copies of mail to another server. Beazley Security, summarizing the same indicators, describes &ldquo;attackers creating a remote mail archive account&rdquo; that &ldquo;sends archived mail&rdquo; to that address. For a mail gateway that sees an organization&rsquo;s inbound and outbound email, that is a continuing leak of message content, not a one-time break-in.</p>
<p>The advisory also lists two entries from the encryption logs: an IBE decryption error ending <code>'Invalid Base64 Encoding at pos 0. Character=0x2a'</code>, and a failed login for an internal user written as <code>*@domain.tld</code>.</p>
<p>Fortinet names two IP addresses: <code>79.141.169.187</code> and <code>45.129.0.192</code>. Attacker addresses change, so treat a match as strong evidence and a non-match as weak evidence.</p>
<p>Two security firms, Truesec and Beazley Security, additionally list files they say were added or modified on compromised appliances. Their lists overlap but are not identical. Beazley Security presents the files as Fortinet&rsquo;s indicators and publishes hashes for each; we did not see a file list in the advisory text we were able to read, so we attribute them to those firms:</p>
<table>
  <thead>
      <tr>
          <th>File</th>
          <th>Reported as</th>
          <th>Listed by</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><code>/data/lib/liblog.so</code></td>
          <td>Added</td>
          <td>Truesec, Beazley Security</td>
      </tr>
      <tr>
          <td><code>/data/bin/webconsole</code></td>
          <td>Added</td>
          <td>Truesec, Beazley Security</td>
      </tr>
      <tr>
          <td><code>/data/bin/mailservice</code></td>
          <td>Added</td>
          <td>Truesec, Beazley Security</td>
      </tr>
      <tr>
          <td><code>/data/etc/ld.so.preload</code></td>
          <td>Added</td>
          <td>Beazley Security</td>
      </tr>
      <tr>
          <td><code>/bin/smit</code></td>
          <td>Modified</td>
          <td>Truesec, Beazley Security</td>
      </tr>
      <tr>
          <td><code>/data/migadmin.tar.gz</code></td>
          <td>Modified</td>
          <td>Beazley Security</td>
      </tr>
      <tr>
          <td><code>/data/etc/httpd.conf</code></td>
          <td>Modified</td>
          <td>Beazley Security</td>
      </tr>
  </tbody>
</table>
<p>No source we read attributes the activity to a named group, gives a count of affected organizations, or says when exploitation began. CISA&rsquo;s catalog entry lists ransomware use as &ldquo;Unknown&rdquo;.</p>
<h2 id="what-to-do-in-order">What to do, in order</h2>
<h3 id="1-find-out-whether-you-are-exposed">1. Find out whether you are exposed</h3>
<p>Treat the appliance as exposed if it runs a build on the 8.0, 7.6, 7.4, 7.2 or 7.0 branch at or below the versions in the tables above and its web interface accepts requests from networks you do not control. Record the exact build number and whether the IBE service is on. Fortinet&rsquo;s advisory does not spell out the exact conditions for exploitation, so do not rule an appliance out on the IBE setting alone until you have confirmed it in the configuration.</p>
<h3 id="2-preserve-logs-before-you-change-the-configuration">2. Preserve logs before you change the configuration</h3>
<p>The indicators Fortinet published are log entries. Export the system event logs and encryption logs off the appliance first, and take a snapshot if FortiMail runs as a virtual machine. A compromised appliance can have its local logs altered, and a later upgrade or rebuild may discard them. Our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents rather than dashboards</a> covers why off-device copies matter.</p>
<h3 id="3-apply-the-workaround">3. Apply the workaround</h3>
<p>Fortinet&rsquo;s primary workaround is to disable IBE feature support. In the GUI, the advisory gives the path as Encryption, then IBE, then set IBE Service to off. From the CLI:</p>
<pre tabindex="0"><code>config system encryption ibe
set status disable
end
</code></pre><p>The trade-off is functional: with the service off, FortiMail stops providing IBE encrypted delivery. If your organization relies on it for outside recipients, tell the people who use it before you switch it off, and decide what they should use in the meantime.</p>
<p>The advisory lists two alternatives:</p>
<ul>
<li>Disable access to the FortiMail webmail interface from the internet, or limit access to a trusted private network.</li>
<li>If a web application firewall sits in front of FortiMail, block POST requests to <code>/ibe</code> that contain <code>../</code>.</li>
</ul>
<p>Our assessment, not Fortinet&rsquo;s: treat the firewall rule as the weakest of the three. A rule that matches the literal string <code>../</code> depends on how the firewall decodes and normalizes a request before matching, and this flaw involves a second trick (the NULL byte) alongside the traversal. Removing the exposed feature or the exposed interface does not depend on matching the attacker&rsquo;s exact input.</p>
<h3 id="4-check-for-compromise">4. Check for compromise</h3>
<p>Search the exported logs for the patterns above:</p>
<ul>
<li>Any archive account you did not create, especially one with a remote destination. <code>archive234</code> is the name in Fortinet&rsquo;s example; an attacker can choose a different one, so review every archive account, not only that name. Beazley Security gives the GUI location as Email Archiving, then Archive Account.</li>
<li>Cron entries running shell commands as root.</li>
<li>Administrator logouts recorded as coming from <code>(null)</code>.</li>
<li>IBE decryption errors mentioning <code>Invalid Base64 Encoding</code>.</li>
<li>Connections to or from <code>79.141.169.187</code> and <code>45.129.0.192</code> in firewall and proxy logs, not only on the appliance.</li>
</ul>
<p>If you have shell-level visibility through Fortinet support, compare against the file list above. A clean result lowers the likelihood of this specific activity. It does not prove the appliance is clean.</p>
<h3 id="5-if-you-find-indicators">5. If you find indicators</h3>
<p>Fortinet&rsquo;s advisory gives no recovery procedure, so this is general incident response practice rather than vendor guidance:</p>
<ul>
<li>Isolate the appliance and open a case with Fortinet support.</li>
<li>Remove the unauthorized archive account, and work out from its creation time in the logs how long mail was being copied. That period defines what may need to be reported under your breach notification obligations.</li>
<li>Rotate credentials the appliance holds or has seen: administrator passwords, LDAP bind accounts, API keys and TLS private keys.</li>
<li>Plan a rebuild from known good firmware rather than cleaning files by hand. Several of the reported files, including an <code>ld.so.preload</code> entry, are the kind used to reload attacker code after a restart.</li>
</ul>
<h3 id="6-upgrade-when-the-builds-ship">6. Upgrade when the builds ship</h3>
<p>Watch the advisory for the release of 8.0.2, 7.6.7 and 7.4.9, and upgrade when your branch is available. Keep the workaround in place until then. If you run 7.2, start planning the move to 7.4 now, since that is the only path the advisory gives.</p>
<h2 id="what-the-cisa-deadline-means-for-you">What the CISA deadline means for you</h2>
<p>CISA added CVE-2026-104286 to its Known Exploited Vulnerabilities (KEV) catalog on October 1, 2026, the same day Fortinet published. The catalog entry sets a due date of October 4, 2026 and marks forensic triage as required.</p>
<p>Two points of scope. First, the due date comes from Binding Operational Directive 26-04, which, according to Tenable&rsquo;s analysis of the directive, applies to US Federal Civilian Executive Branch agencies and is not binding on private companies. Second, the required action in the entry is to &ldquo;apply mitigations in accordance with vendor instructions&rdquo; or to &ldquo;discontinue use of the product if mitigations are unavailable.&rdquo; With the fixed builds listed as upcoming, the vendor instruction available today is the workaround.</p>
<p>For everyone outside the US federal government, read the three-day window as a statement of urgency, and note that CISA paired the mitigation with a compromise check. That is the same sequence as the steps above.</p>
<h2 id="the-pattern-worth-fixing-afterward">The pattern worth fixing afterward</h2>
<p>This section is editorial analysis rather than reported fact.</p>
<p>A mail gateway is an unusually valuable thing to compromise. It reads every message, it is trusted by the mail servers behind it, and it is expected to accept connections from anywhere. Three habits reduce the cost of the next advisory like this one:</p>
<ul>
<li><strong>Turn off web-facing features you do not use.</strong> The exposed surface here was one optional feature. An inventory of which optional services are enabled on each edge device turns &ldquo;are we affected?&rdquo; into a lookup.</li>
<li><strong>Alert on configuration changes that redirect data.</strong> A new archive destination, a new forwarding rule or a new administrator account on a mail system should raise an alert whoever makes the change.</li>
<li><strong>Do not let one data source decide exposure.</strong> This week the advisory and the CVE record disagree. A process that checks the vendor advisory for any actively exploited flaw would catch that.</li>
</ul>
<p>We covered the same patch-then-hunt sequence for a different edge device in <a href="/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/">the Citrix NetScaler zero-days</a>, and the general response order in <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">what a zero-day is and how to respond to one</a>. For where this belongs in your queue, see <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">the vulnerability backlog that never shrinks</a>.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 4, 2026. We did not test the workaround or the indicators on a FortiMail appliance; the commands and log patterns are quoted from Fortinet&rsquo;s advisory.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Unauthenticated attacker can write arbitrary files via crafted HTTP or HTTPS requests; exploited in the wild</td>
          <td><a href="https://fortiguard.fortinet.com/psirt/FG-IR-26-175">Fortinet advisory FG-IR-26-175</a>, published October 1, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Affected: 8.0.0 to 8.0.1, 7.6.0 to 7.6.6, 7.4.0 to 7.4.8, 7.2.0 to 7.2.9</td>
          <td>Fortinet advisory FG-IR-26-175</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Fixed builds 8.0.2, 7.6.7 and 7.4.9 described as upcoming</td>
          <td>Fortinet advisory FG-IR-26-175, as read October 4, 2026</td>
          <td>Qualified: status can change after this date</td>
      </tr>
      <tr>
          <td>Workaround: disable IBE; alternatives are restricting webmail access or a firewall rule</td>
          <td>Fortinet advisory FG-IR-26-175</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVSS 3.1 base score 9.8</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-104286">CVE record, Fortinet CNA</a>, updated October 2, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE record lists different affected and fixed versions than the advisory, and adds the 7.0 branch</td>
          <td>CVE record as retrieved October 4, 2026, compared with the advisory</td>
          <td>Verified: describes the record on that date</td>
      </tr>
      <tr>
          <td>Log indicators and IP addresses 79.141.169.187 and 45.129.0.192</td>
          <td>Fortinet advisory FG-IR-26-175</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Added and modified files on compromised appliances</td>
          <td><a href="https://www.truesec.com/hub/blog/cve-2026-104286-fortimail-path-traversal-vulnerability-actively-exploited">Truesec</a>, <a href="https://beazley.security/alerts-advisories/critical-path-traversal-in-fortinet-fortimail-under-active-exploitation">Beazley Security</a></td>
          <td>Qualified: reported by these firms; not seen in the advisory text we read</td>
      </tr>
      <tr>
          <td>Added to CISA KEV October 1, 2026; due October 4, 2026; forensic triage required; ransomware use unknown</td>
          <td><a href="https://github.com/cisagov/kev-data">CISA KEV catalog data</a>, catalog version 2026.10.02</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 binds US federal civilian agencies only</td>
          <td><a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">Tenable&rsquo;s analysis of BOD 26-04</a></td>
          <td>Qualified: secondary source</td>
      </tr>
      <tr>
          <td>What IBE is</td>
          <td><a href="https://docs.fortinet.com/document/fortimail/7.4.2/administration-guide/286643">FortiMail administration guide, Configuring IBE encryption</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
]]></content:encoded></item><item><title>Citrix NetScaler Zero-Days (CVE-2026-88771, CVE-2026-88772): Patch, Then Hunt</title><link>https://firewallsync.com/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/</link><guid>https://firewallsync.com/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/</guid><pubDate>Sat, 03 Oct 2026 08:00:00 +0000</pubDate><dc:creator>FirewallSync Editorial</dc:creator><category>Security Operations</category><description>Two NetScaler ADC and Gateway flaws were exploited before Citrix disclosed them on September 27, 2026. Upgrading closes the hole. It does not tell you whether someone already came through it.</description><enclosure url="https://firewallsync.com/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/cover_hu11562816909962696335.jpg" type="image/jpeg" length="0"/><media:content url="https://firewallsync.com/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/cover_hu11562816909962696335.jpg" medium="image" type="image/jpeg" width="1200" height="630"/><content:encoded><![CDATA[<p><img src="https://firewallsync.com/posts/citrix-netscaler-zero-days-cve-2026-88771-cve-2026-88772-patch-and-hunt/cover_hu11562816909962696335.jpg" alt="Fiber optic cables connected to a network switch in a server rack" width="1200" height="630"></p><p>If you run Citrix NetScaler ADC or NetScaler Gateway and are not yet on one of the fixed builds listed below, upgrade now. If the appliance was reachable from the internet at any point in September, also assume the upgrade is only half the job: both flaws were being exploited before Citrix disclosed them, and none of the vendor reports says an upgrade removes what attackers installed.</p>
<p>This article covers what is confirmed, where the vendor reports differ, and the order to work in.</p>
<h2 id="what-citrix-disclosed">What Citrix disclosed</h2>
<p>On September 27, 2026, Citrix published security bulletin <a href="https://support.citrix.com/support-home/kbsearch/article?articleNumber=CTX697096">CTX697096</a> covering eight vulnerabilities, CVE-2026-88771 through CVE-2026-88778. Two of them were already under attack. BleepingComputer quotes the bulletin as saying exploits of CVE-2026-88771 and CVE-2026-88772 &ldquo;on unmitigated NetScaler deployments have been observed.&rdquo;</p>
<p>The details below come from the CVE records that NetScaler itself published as the CVE Numbering Authority (the organization authorized to assign and describe CVE IDs for its own products).</p>
<table>
  <thead>
      <tr>
          <th></th>
          <th>CVE-2026-88771</th>
          <th>CVE-2026-88772</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Weakness</td>
          <td>Improper input validation (CWE-20)</td>
          <td>Memory buffer bounds error (CWE-119)</td>
      </tr>
      <tr>
          <td>Impact</td>
          <td>Unauthenticated attacker can execute arbitrary commands</td>
          <td>Remote code execution or denial of service</td>
      </tr>
      <tr>
          <td>CVSS 4.0 base score</td>
          <td>9.5 (Critical)</td>
          <td>9.5 (Critical)</td>
      </tr>
      <tr>
          <td>Attack complexity</td>
          <td>Low</td>
          <td>High</td>
      </tr>
      <tr>
          <td>Privileges required</td>
          <td>None</td>
          <td>None</td>
      </tr>
      <tr>
          <td>Record published</td>
          <td>September 27, 2026</td>
          <td>September 27, 2026</td>
      </tr>
  </tbody>
</table>
<p>CVSS (Common Vulnerability Scoring System) is the industry-standard 0 to 10 severity scale; these scores use version 4.0.</p>
<p>The other six CVEs in the bulletin, CVE-2026-88773 through CVE-2026-88778, carry CVSS 4.0 base scores between 7.0 and 9.3 in their CVE records. As of October 3, 2026, none of the six is in CISA&rsquo;s Known Exploited Vulnerabilities catalog. All eight records list the same fixed builds.</p>
<h3 id="affected-and-fixed-builds">Affected and fixed builds</h3>
<table>
  <thead>
      <tr>
          <th>Product branch</th>
          <th>Affected</th>
          <th>Fixed in</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>NetScaler ADC and Gateway 14.1</td>
          <td>Before 14.1-73.37</td>
          <td>14.1-73.37 and later</td>
      </tr>
      <tr>
          <td>NetScaler ADC and Gateway 13.1</td>
          <td>Before 13.1-64.23</td>
          <td>13.1-64.23 and later</td>
      </tr>
      <tr>
          <td>NetScaler ADC 14.1-FIPS</td>
          <td>Before 14.1-73.37 FIPS</td>
          <td>14.1-73.37 FIPS and later</td>
      </tr>
      <tr>
          <td>NetScaler ADC 13.1-FIPS and 13.1-NDcPP</td>
          <td>Before 13.1-37.279</td>
          <td>13.1-37.279 and later</td>
      </tr>
  </tbody>
</table>
<p>A formatting note: seven of the eight CVE records write the last build as &ldquo;13.1.37.279&rdquo;, while the record for CVE-2026-88773 and Tenable&rsquo;s summary write &ldquo;13.1-37.279&rdquo;, the format NetScaler build numbers normally use. We read these as the same build; confirm the exact build string on the Citrix download page before upgrading a FIPS or NDcPP appliance.</p>
<p>Versions 12.1 and 13.0 are end-of-life and receive no security updates. Tenable noted on September 27 that Citrix had not said whether they are affected. If you still run either, the absence of a statement is not evidence of safety; plan the migration rather than waiting for one.</p>
<h3 id="the-dtls-precondition">The DTLS precondition</h3>
<p>Rapid7 and Tenable both report that CVE-2026-88771 affects default configurations, while CVE-2026-88772 requires DTLS to be enabled. DTLS (Datagram Transport Layer Security) is TLS adapted for UDP, and NetScaler Gateway uses it on UDP port 443 alongside the usual TCP port 443.</p>
<p>That precondition narrows less than it sounds. <a href="https://docs.netscaler.com/en-us/netscaler-gateway/current-release/configure-dtls-virtual-server-using-ssl-virtual-server">NetScaler&rsquo;s own documentation</a> states: &ldquo;By default, the DTLS functionality is set to ON for the existing SSL VPN virtual server.&rdquo; Unless someone turned it off, a Gateway deployment meets the condition.</p>
<h2 id="what-is-known-about-exploitation">What is known about exploitation</h2>
<p>Three security vendors have published observations. They do not contradict each other, but each describes only what its own telemetry or incident response work showed, so the dates differ.</p>
<table>
  <thead>
      <tr>
          <th>Source</th>
          <th>What it reports</th>
          <th>Dates</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>Palo Alto Networks Unit 42</td>
          <td>Requests that fingerprint NetScaler Gateway appliances</td>
          <td>From August 21, 2026</td>
      </tr>
      <tr>
          <td>Unit 42</td>
          <td>Repeated requests to files in the appliance&rsquo;s <code>/vpn/scripts/linux/</code> folder</td>
          <td>September 4 to 24, 2026</td>
      </tr>
      <tr>
          <td>Google Threat Intelligence Group and Mandiant</td>
          <td>Exploitation campaign against organizations in North America and Europe</td>
          <td>Since at least early September 2026</td>
      </tr>
      <tr>
          <td>Rapid7</td>
          <td>Earliest exploitation it reports: a command that archived the configuration directory into a web-readable path</td>
          <td>September 20, 2026, 14:28 UTC</td>
      </tr>
      <tr>
          <td>Unit 42</td>
          <td>PHP web shell dropped in three stages against a US-based target</td>
          <td>September 21, 2026</td>
      </tr>
  </tbody>
</table>
<p>The practical reading: the exposure window for an unpatched, internet-facing appliance opened weeks before the bulletin, not on the day of it. Unit 42 does not attribute the activity to a named group.</p>
<p>On scale, Unit 42 says its Cortex Xpanse scanning &ldquo;identified 50,277 exposed instances that could potentially be vulnerable&rdquo; as of September 27, 2026. That is a count of exposed appliances, not of confirmed compromises.</p>
<p>Tenable reported no public proof-of-concept exploit for either flaw as of September 27, 2026. Rapid7 reports that exploitation widened after disclosure. Treat the September 27 statement as a point in time, not a current guarantee.</p>
<h3 id="what-attackers-did-after-getting-in">What attackers did after getting in</h3>
<p>The behavior described by Google and Rapid7 explains why patching alone is not enough:</p>
<ul>
<li><strong>Configuration theft.</strong> Rapid7 observed <code>tar czf /var/netscaler/gui/vpn/c -C /flash nsconfig</code>, which packs the configuration directory into a location the web server will hand out. Rapid7 notes that directory holds encrypted administrator passwords, SSL certificates and private keys, and SSH host keys.</li>
<li><strong>Web shells behind innocent-looking extensions.</strong> Google documented changes to <code>/etc/httpd.conf</code> that make Apache run <code>.deb</code> and <code>.sig</code> files as PHP, plus an <code>AliasMatch</code> rule that maps <code>.ico</code> requests under <code>/vpn/media/</code> onto those files. The shells answer with HTTP 404, so they look like misses in a casual log review.</li>
<li><strong>Root persistence.</strong> Google observed <code>chmod u+s /bin/sh</code>, which sets the SUID bit so later web requests execute with root privileges.</li>
<li><strong>Tunneling into the network.</strong> Google named two tools: WHIPSHOT, a PHP web shell acting as a tunnel front end, and SLAPSHOT, a Python proxy that listens on a loopback port and records it in <code>/tmp/.uxdport</code>.</li>
</ul>
<h2 id="what-to-do-in-order">What to do, in order</h2>
<h3 id="1-preserve-evidence-before-you-change-anything">1. Preserve evidence before you change anything</h3>
<p>If the appliance is a virtual machine, take a snapshot that includes memory before rebooting or upgrading. Google recommends this explicitly, and <a href="https://support.citrix.com/external/article/CTX694799/steps-to-take-if-netscaler-adc-is-suspec.html">Citrix&rsquo;s guidance for a suspected compromise</a> lists a snapshot, a technical support bundle and a packet engine core dump as the first step. An upgrade reboots the device and discards memory-only evidence.</p>
<h3 id="2-upgrade-to-a-fixed-build">2. Upgrade to a fixed build</h3>
<p>Move to the fixed build for your branch from the table above. Google lists the upgrade as its recommended option and advises opening a Severity 1 support case with Citrix if you cannot locate the right build.</p>
<p>If you cannot upgrade immediately, Google describes compensating controls for CVE-2026-88772 only:</p>
<ul>
<li>Disable DTLS on internet-facing Gateway virtual servers where you do not need it. The documented command is <code>set vpn vserver &lt;name&gt; -dtls off</code>.</li>
<li>Block inbound UDP port 443 on the perimeter firewall or edge router. Google warns that ACLs on the appliance are not sufficient, because the traffic reaches the vulnerable packet engine first.</li>
</ul>
<p>Neither control addresses CVE-2026-88771. They buy time for one of the two bugs; they are not a substitute for the upgrade.</p>
<h3 id="3-hunt-for-compromise">3. Hunt for compromise</h3>
<p>Run these on the appliance shell. All commands are taken from <a href="https://cloud.google.com/blog/topics/threat-intelligence/defending-against-active-exploitation-of-citrix-netscaler-adc-and-gateway-appliances">Google&rsquo;s published hunting guidance</a>. A clean result reduces the likelihood of this specific tooling; it does not prove the device is clean.</p>
<p>Check for tampering in the web server configuration:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>grep -En -i <span style="color:#e6db74">&#34;application/x-httpd-php|php_flag|AliasMatch&#34;</span> /etc/httpd.conf
</span></span></code></pre></div><p>Look for handlers that register non-PHP extensions such as <code>.deb</code> or <code>.sig</code> as PHP, and for <code>AliasMatch</code> rules pointing <code>/vpn/media/</code>, <code>/vpn/theme/</code> or <code>/vpn/images/</code> at script directories.</p>
<p>Search the web directories for PHP where none belongs:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>grep -rlE <span style="color:#e6db74">&#34;&lt;\?php|eval\(|base64_decode\(|shell_exec\(&#34;</span> /var/netscaler/gui/ /netscaler/ns_gui/ /var/vpn/ /netscaler/portal/ 2&gt;/dev/null
</span></span></code></pre></div><p>Check for the tunneling tool&rsquo;s files, the SUID shell and stray Python processes:</p>
<div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>ls -la /tmp/.uxdport* /tmp/.uxdlock
</span></span><span style="display:flex;"><span>ls -l /bin/sh
</span></span><span style="display:flex;"><span>ps aux | grep -E <span style="color:#e6db74">&#34;python.*(\.uxd|uxdport|uxdlock|base64)&#34;</span>
</span></span></code></pre></div><p>Permissions of <code>-rwsr-xr-x</code> on <code>/bin/sh</code> indicate the SUID change. Either <code>/tmp/.uxd*</code> file means the proxy is running or ran recently.</p>
<p>Also check for two artifacts from Rapid7&rsquo;s report: a file at <code>/var/netscaler/gui/vpn/c</code>, which is the output path of the archive command it observed, and a web shell at <code>/var/netscaler/logon/LogonPoint/custom/.ctxs.receiver</code>.</p>
<p>In logs, Google flags two patterns worth alerting on. The first is a DTLS handshake failure with <code>ClientVersion DTLSv1.0</code> and the reason &ldquo;Handshake failure-Internal Error&rdquo;, followed within minutes by a crash of NSPPE (the NetScaler Packet Processing Engine) in <code>/var/log/messages</code>. The second is HTTP 404 responses to <code>/vpn/media/*.ico</code> that carry multi-kilobyte bodies.</p>
<p>Network indicators such as IP addresses and file hashes change quickly. Pull current lists from the <a href="https://cloud.google.com/blog/topics/threat-intelligence/defending-against-active-exploitation-of-citrix-netscaler-adc-and-gateway-appliances">Google</a>, <a href="https://www.rapid7.com/blog/post/etr-zero-day-exploitation-of-citrix-netscaler-adc-and-gateway-cve-2026-88771-and-cve-2026-88772/">Rapid7</a> and <a href="https://unit42.paloaltonetworks.com/netscaler-zero-days-exploited/">Unit 42</a> reports rather than a copy in an article. Google also publishes YARA rules for the web shells and the configuration tampering.</p>
<h3 id="4-if-you-find-anything-rebuild-rather-than-clean">4. If you find anything, rebuild rather than clean</h3>
<p>Citrix&rsquo;s guidance for a suspected compromise is to remove the appliance from the network, erase and reinstall firmware, upgrade to the latest version, and restore a known good backup that predates the compromise. Google adds one step that is easy to miss in a high-availability pair: disable HA synchronization and assess each node on its own, so a tampered configuration is not replicated to the healthy node.</p>
<p>Given the early September exploitation dates above, a &ldquo;known good&rdquo; backup should predate early September 2026 at a minimum, not merely the bulletin. That is our inference from the reported timeline, not vendor guidance.</p>
<h3 id="5-rotate-what-the-appliance-knew">5. Rotate what the appliance knew</h3>
<p>Because the configuration directory was a target, Google advises assuming credentials on a compromised appliance are exposed, and rotating them after the appliance is patched:</p>
<ul>
<li>NetScaler administrator and local account passwords, and SSH keys</li>
<li>TLS certificates and their private keys</li>
<li>LDAP bind and service accounts, RADIUS shared secrets, TACACS credentials</li>
<li>SNMP community strings and NITRO API credentials</li>
</ul>
<p>Invalidate active administrative, Gateway and VPN sessions as well. Citrix&rsquo;s guidance goes further on restored devices: rotate the Key Encryption Keys and replace all SSL certificates.</p>
<p>Then look downstream. Google recommends reviewing StoreFront servers, Delivery Controllers and virtual desktop hosts for unexpected interactive logons, RDP activity and signs of credential dumping, since the appliance sits directly in front of them.</p>
<h2 id="what-the-cisa-deadline-means-for-you">What the CISA deadline means for you</h2>
<p>CISA added both CVEs to its Known Exploited Vulnerabilities (KEV) catalog on September 27, 2026, with a due date of September 30, 2026. The catalog entry also requires forensic triage, noting that &ldquo;customers must conduct forensic triage as directed by BOD 26-04.&rdquo;</p>
<p>Two points of scope. First, according to Tenable&rsquo;s analysis of Binding Operational Directive 26-04, CISA issued the directive on June 10, 2026, it applies to US Federal Civilian Executive Branch agencies, and it is not binding on the private sector, though CISA encourages adoption. Second, the due date has already passed. For an organization outside the federal government, the deadline is useful as a signal: three days is the shortest remediation window in the directive as Tenable describes it, and CISA paired the patch with a mandatory compromise check, which is the same sequence recommended above.</p>
<h2 id="the-pattern-worth-fixing-after-the-incident">The pattern worth fixing after the incident</h2>
<p>This section is editorial analysis rather than reported fact.</p>
<p>Edge appliances are attractive targets for structural reasons: they face the internet by design, they hold credentials for the directory and authentication systems behind them, and most cannot run the endpoint detection agents that cover servers and laptops. Three changes reduce the cost of the next advisory of this kind:</p>
<ul>
<li><strong>Ship appliance logs off the device.</strong> The hunting steps above depend on <code>ns.log</code>, <code>/var/log/messages</code> and the HTTP access and error logs. Logs that live only on the appliance are available to whoever compromises it. Our piece on <a href="/posts/logging-for-incidents-not-for-dashboards/">logging for incidents rather than dashboards</a> covers what to retain.</li>
<li><strong>Default-deny outbound traffic from the appliance.</strong> Google recommends permitting only approved destinations such as DNS, NTP, certificate status checks and the backend applications. A tunneling tool is far less useful on a device that cannot open arbitrary connections.</li>
<li><strong>Keep management interfaces off the internet.</strong> Restrict the NSIP management address, SSH and the NITRO API to administrative networks.</li>
</ul>
<p>For triage of the rest of your queue, an actively exploited flaw on an internet-facing device is the case that should jump every line. We cover that prioritization in <a href="/posts/the-vulnerability-backlog-that-never-shrinks/">the vulnerability backlog that never shrinks</a>, and the general response sequence in <a href="/posts/what-is-a-zero-day-vulnerability-and-how-should-you-actually-respond-to-one/">what a zero-day is and how to respond to one</a>.</p>
<h2 id="sources-and-verification">Sources and verification</h2>
<p>Checked on October 3, 2026. We could not load the Citrix bulletin page directly, so bulletin details are taken from the CVE records NetScaler published and from vendors quoting the bulletin; those rows are marked Qualified.</p>
<table>
  <thead>
      <tr>
          <th>Important claim</th>
          <th>Source</th>
          <th>Verification</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td>CVE-2026-88771 lets an unauthenticated attacker execute arbitrary commands; CVSS 4.0 base score 9.5</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-88771">CVE record, NetScaler CNA</a>, published September 27, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>CVE-2026-88772 can lead to remote code execution or denial of service; CVSS 4.0 base score 9.5</td>
          <td><a href="https://www.cve.org/CVERecord?id=CVE-2026-88772">CVE record, NetScaler CNA</a>, published September 27, 2026</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Fixed builds: 14.1-73.37, 13.1-64.23, 14.1-73.37 FIPS, 13.1-37.279 FIPS/NDcPP</td>
          <td>Same CVE records</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Citrix states exploitation of both CVEs has been observed</td>
          <td>Bulletin <a href="https://support.citrix.com/support-home/kbsearch/article?articleNumber=CTX697096">CTX697096</a>, as quoted by <a href="https://www.bleepingcomputer.com/news/security/citrix-admins-warned-to-shut-down-netscalers-over-2-exploited-zero-days/">BleepingComputer</a></td>
          <td>Qualified: quoted secondhand</td>
      </tr>
      <tr>
          <td>CVE-2026-88772 requires DTLS to be enabled</td>
          <td>Bulletin, as reported by <a href="https://www.rapid7.com/blog/post/etr-zero-day-exploitation-of-citrix-netscaler-adc-and-gateway-cve-2026-88771-and-cve-2026-88772/">Rapid7</a> and <a href="https://www.tenable.com/blog/frequently-asked-questions-about-reported-citrix-netscaler-zero-day-vulnerabilities">Tenable</a></td>
          <td>Qualified: reported secondhand</td>
      </tr>
      <tr>
          <td>DTLS is on by default for an SSL VPN virtual server</td>
          <td><a href="https://docs.netscaler.com/en-us/netscaler-gateway/current-release/configure-dtls-virtual-server-using-ssl-virtual-server">NetScaler Gateway documentation</a></td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>Both CVEs added to CISA KEV on September 27, 2026, due September 30, 2026</td>
          <td><a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-88771">CISA KEV catalog</a>, catalog version 2026.10.02</td>
          <td>Verified</td>
      </tr>
      <tr>
          <td>BOD 26-04 binds US federal civilian agencies only</td>
          <td><a href="https://www.tenable.com/blog/cisa-bod-26-04-FAQ-vulnerability-remediation-impact">Tenable analysis of BOD 26-04</a></td>
          <td>Qualified: secondary source</td>
      </tr>
      <tr>
          <td>Exploitation activity predates disclosure</td>
          <td><a href="https://cloud.google.com/blog/topics/threat-intelligence/defending-against-active-exploitation-of-citrix-netscaler-adc-and-gateway-appliances">Google</a>, <a href="https://www.rapid7.com/blog/post/etr-zero-day-exploitation-of-citrix-netscaler-adc-and-gateway-cve-2026-88771-and-cve-2026-88772/">Rapid7</a>, <a href="https://unit42.paloaltonetworks.com/netscaler-zero-days-exploited/">Unit 42</a></td>
          <td>Qualified: each vendor reports its own visibility; dates differ</td>
      </tr>
      <tr>
          <td>50,277 exposed instances potentially vulnerable</td>
          <td><a href="https://unit42.paloaltonetworks.com/netscaler-zero-days-exploited/">Unit 42</a>, as of September 27, 2026</td>
          <td>Qualified: one vendor&rsquo;s scan, exposure not compromise</td>
      </tr>
      <tr>
          <td>Hunting commands and post-exploitation behavior</td>
          <td><a href="https://cloud.google.com/blog/topics/threat-intelligence/defending-against-active-exploitation-of-citrix-netscaler-adc-and-gateway-appliances">Google Threat Intelligence Group and Mandiant</a>, September 29, 2026</td>
          <td>Verified against the source; not run by us</td>
      </tr>
      <tr>
          <td>Steps for a suspected compromise</td>
          <td><a href="https://support.citrix.com/external/article/CTX694799/steps-to-take-if-netscaler-adc-is-suspec.html">Citrix CTX694799</a></td>
          <td>Verified</td>
      </tr>
  </tbody>
</table>
<p>FirewallSync has not tested these vulnerabilities or run the hunting commands against a compromised appliance. This is research-based analysis of the sources above.</p>
]]></content:encoded></item></channel></rss>