CVSS v4 Supplemental Metrics in CVE Records: Who Sets Them

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.

A lit red no-entry signal with a white horizontal bar, glowing against a dark background at night
Photo by Krzysztof Hepner on Unsplash

Short answer: CVSS v4.0 added six optional Supplemental 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 3.4% of records with a CNA cvssV4_0 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’s opinion at publication time.

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 vulnerability prioritization workflow. For reading the rest of a v4.0 vector, see how to read a CVSS v4.0 vector string.

The six Supplemental metrics

The CVSS v4.0 specification introduces “a new, optional metric group called the Supplemental metric group” and states: “No metric will have any impact on the final calculated CVSS score (e.g. CVSS-BTE).” The values “convey additional extrinsic characteristics of the vulnerability itself.”

MetricVectorAllowed valuesCVE JSON propertyJSON enum
SafetySX, N, PSafetyNEGLIGIBLE, PRESENT
AutomatableAUX, N, YAutomatableNO, YES
RecoveryRX, A, U, IRecoveryAUTOMATIC, USER, IRRECOVERABLE
Value DensityVX, D, CvalueDensityDIFFUSE, CONCENTRATED
Vulnerability Response EffortREX, L, M, HvulnerabilityResponseEffortLOW, MODERATE, HIGH
Provider UrgencyUX, Clear, Green, Amber, RedproviderUrgencyCLEAR, GREEN, AMBER, RED

Every JSON enum also allows NOT_DEFINED, which is the schema default (cve-schema cvss-v4.0.json). In the vector, X means Not Defined, and the spec tells consumers to “treat unspecified Threat, Environmental and Supplemental as Not Defined.”

What each one means, in the spec’s words:

  • Safety: whether consequences meet the IEC 61508 consequence categories. P (Present) versus N (Negligible).
  • Automatable: “Can an attacker automate exploitation events for this vulnerability across multiple targets?” Y means “Attackers can reliably automate all 4 steps of the kill chain.”
  • Recovery: whether the system recovers services automatically (A), needs user intervention (U), or is irrecoverable by the user (I) after an attack.
  • Value Density: what one exploitation event gains the attacker. D (Diffuse): “The vulnerable system has limited resources.” C (Concentrated): “The vulnerable system is rich in resources.”
  • Vulnerability Response Effort: how hard it is for the consumer to respond, from L (“low/trivial”) to H (“significant and/or difficult”).
  • Provider Urgency: an optional “pass-through” metric for the provider’s own assessment, from Clear (“no urgency (Informational)”) to Red (“the highest urgency”).

Two schema details trip up parsers:

  1. Property-name casing is inconsistent. Safety, Automatable and Recovery 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.
  2. Provider Urgency uses words in the vector. The vector segment is U:Amber, not U:A, and the JSON value is AMBER. A regex for single-letter values silently drops it.

Who actually sets them

We streamed every record in the cvelistV5 midnight release of November 14, 2026 (2026-11-14_all_CVEs_at_midnight.zip.zip), kept records in state PUBLISHED, and counted CNA cvssV4_0 entries where either the JSON property is set to something other than NOT_DEFINED or the vector has a non-X segment. Counts below are from that run; the method is described so you can repeat it.

MeasureCount
Published records with a CNA cvssV4_0 entry46,973 (47,055 entries)
Records with at least one Supplemental metric1,609 (3.4%)
CNA entries with at least one Supplemental metric1,661
Entries with all six set257
CNAs publishing v4.0 / of those, using Supplemental at least once326 / 149

By metric, across the 1,661 entries:

MetricEntries setValue split
Automatable1,210YES 732, NO 478
Provider Urgency1,152AMBER 617, GREEN 212, RED 208, CLEAR 115
Response Effort1,123MODERATE 726, LOW 330, HIGH 67
Recovery1,067USER 679, AUTOMATIC 345, IRRECOVERABLE 43
Value Density891DIFFUSE 448, CONCENTRATED 443
Safety475NEGLIGIBLE 314, PRESENT 161

Who sets them matters more than the totals:

CNACNA v4.0 entriesWith SupplementalTypical pattern
VulDB14,1150none
VulnCheck11,6102none
GitHub (GitHub_M)5,84131occasional
icscert (CISA ICS)9470none
Juniper426258AU, R, RE, often V
Palo Alto Networks263258AU, R, V, RE, U

Other frequent users in the snapshot include bcorg (mostly U only), Apache, OpenText, GovTech CSG and Google Cloud (mostly U only). By CVE ID year, 880 of the 1,661 entries are 2026 IDs, 456 are 2025 and 301 are 2024.

The practical reading: a missing AU or U 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.

Only five ADP cvssV4_0 entries exist in the snapshot (all CISA-ADP), and none set a Supplemental metric. CISA’s exploitation and automatability signals arrive through SSVC instead; see how to read SSVC values in a CVE record.

Trap: the vector and the JSON fields drift apart

The schema stores each Supplemental value twice: once as a segment in vectorString 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:

  • CVE-2026-110803 (Vaadin): JSON sets all six properties (Automatable: YES, providerUrgency: CLEAR and so on) while the vector ends at SA:N with no Supplemental segments.
  • CVE-2026-41292 (NLnet Labs): the vector ends E:U/U:Red, but no providerUrgency property exists.

We found no entry where both sides were set to different defined values; the drift is always one side missing. Tools that read only vectorString or only the JSON properties will each miss part of the data.

A parser that reads both and flags drift

This Python 3 function (run with Python 3.13 against the three records named in this article, downloaded from raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/...) returns defined Supplemental values from a cvssV4_0 object, preferring the vector and falling back to JSON, and reports mismatches:

import json, sys

# vector abbreviation -> (JSON property in cvss-v4.0.json, vector value -> JSON enum)
SUPPLEMENTAL = {
    "S":  ("Safety", {"N": "NEGLIGIBLE", "P": "PRESENT"}),
    "AU": ("Automatable", {"N": "NO", "Y": "YES"}),
    "R":  ("Recovery", {"A": "AUTOMATIC", "U": "USER", "I": "IRRECOVERABLE"}),
    "V":  ("valueDensity", {"D": "DIFFUSE", "C": "CONCENTRATED"}),
    "RE": ("vulnerabilityResponseEffort", {"L": "LOW", "M": "MODERATE", "H": "HIGH"}),
    "U":  ("providerUrgency", {"Clear": "CLEAR", "Green": "GREEN", "Amber": "AMBER", "Red": "RED"}),
}

def supplemental(cvss4):
    """Return {json_field: value} for defined Supplemental metrics, plus conflicts."""
    segs = dict(p.split(":", 1) for p in cvss4.get("vectorString", "").split("/")[1:] if ":" in p)
    out, conflicts = {}, []
    for abbr, (field, mapping) in SUPPLEMENTAL.items():
        from_vec = mapping.get(segs.get(abbr))            # None if absent or X
        from_json = cvss4.get(field)
        from_json = None if from_json in (None, "NOT_DEFINED") else from_json
        if from_vec != from_json:          # differs, or one side missing
            conflicts.append((field, from_vec, from_json))
        value = from_vec or from_json
        if value:
            out[field] = value
    return out, conflicts

if __name__ == "__main__":
    record = json.load(open(sys.argv[1]))
    for m in record["containers"]["cna"].get("metrics", []):
        if "cvssV4_0" in m:
            vals, conflicts = supplemental(m["cvssV4_0"])
            print(record["cveMetadata"]["cveId"], vals)
            for c in conflicts:
                print("  vector/JSON mismatch:", c)

Output for CVE-2024-0012 (two v4.0 entries, one per scenario) and CVE-2026-41292:

CVE-2024-0012 {'Automatable': 'NO', 'Recovery': 'USER', 'valueDensity': 'CONCENTRATED', 'vulnerabilityResponseEffort': 'HIGH', 'providerUrgency': 'RED'}
CVE-2024-0012 {'Automatable': 'NO', 'Recovery': 'USER', 'valueDensity': 'CONCENTRATED', 'vulnerabilityResponseEffort': 'HIGH', 'providerUrgency': 'RED'}
CVE-2026-41292 {'providerUrgency': 'RED'}
  vector/JSON mismatch: ('providerUrgency', 'RED', None)

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.

Automatable: CVSS AU versus SSVC Automatable

CVSS AU and SSVC Automatable are close cousins. CERT/CC’s SSVC decision point asks “Can an attacker reliably automate creating exploitation events for this vulnerability?” and defines yes as “Attackers can reliably automate steps 1-4 of the kill chain” (CERTCC/SSVC automatable.py). The CVSS definition likewise refers to “all 4 steps of the kill chain.”

They can still disagree within a single record. In CVE-2024-0012, the PAN-OS management web interface authentication bypass, Palo Alto Networks’ CVSS vector says AU:N. The CISA-ADP SSVC entry in the same record, timestamped 2024-11-18, says Exploitation: active, Automatable: yes, Technical Impact: total. CISA added the CVE to KEV on 2024-11-18.

The record does not explain the difference, so we will not guess at it. The lesson for tooling is narrower: the CNA’s AU is the CNA’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 KEV catalog, not from CVSS.

How to use them in a patch decision

  • Display, do not score. Show AU, R, RE and U 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.
  • Prefer SSVC and KEV for exploitation-driven urgency. Where CISA-ADP has SSVC values, they reflect exploitation status; CVSS Supplemental values do not. See CVSS vs EPSS vs KEV for how the signals divide the work.
  • Use Response Effort for scheduling, not priority. RE:H (as in CVE-2024-0012) is a planning hint: the fix needs real work, so book the window early.
  • Treat Safety: Present as a routing flag. 161 entries set S:P. If you run OT, medical or vehicle systems, route those to the owners who understand IEC 61508 consequences.
  • Never require a value. A rule like “page on U:Red” fires for 208 entries in the entire CVE corpus and stays silent for every VulDB, VulnCheck and CISA ICS v4.0 score.

Common mistakes

  1. Reading only vectorString or only the JSON properties. Read both; 61 CNA entries differ.
  2. Assuming camelCase for all six JSON names. Three are capitalised.
  3. Parsing U as a single letter. Its values are Clear, Green, Amber, Red.
  4. Treating an absent AU as “not automatable”. It almost always means “not evaluated”.
  5. 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.

Limits of this article

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’s values are accurate. The per-CNA table names the short names used in cveMetadata.assignerShortName.

Sources and verification

Important claimSourceStatus
Six Supplemental metrics, allowed vector values, “No metric will have any impact on the final calculated CVSS score”FIRST CVSS v4.0 Specification DocumentVerified
JSON property names, enums, NOT_DEFINED defaultcve-schema cvss-v4.0.jsonVerified
Corpus counts (3.4%, per-metric and per-CNA figures, 61 mismatches, 257 with all six)cvelistV5 release cve_2026-11-14_0000Z, our countVerified (computed)
CVE-2024-0012 AU:N, U:Red, RE:H; CISA-ADP SSVC Automatable: yesCVE-2024-0012 record in cvelistV5Verified
CVE-2024-0012 added to KEV 2024-11-18; KEV CVE countscisagov/kev-dataVerified
SSVC Automatable definitionCERTCC/SSVC automatable.pyVerified

Frequently asked questions

Do CVSS v4.0 Supplemental metrics change the CVSS score?

No. The CVSS v4.0 specification says no Supplemental metric will have any impact on the final calculated CVSS score. A vector ending in AU:Y/U:Red produces the same CVSS-B or CVSS-BT number as the same vector without those segments.

What are the JSON field names for CVSS v4 Supplemental metrics in a CVE record?

In the cvssV4_0 object defined by cve-schema's cvss-v4.0.json they are Safety, Automatable, Recovery, valueDensity, vulnerabilityResponseEffort and providerUrgency. The enums are upper case (for example YES, NO, AMBER, RED) with NOT_DEFINED as the default, while the vector string uses S, AU, R, V, RE and U with short values such as Y, N and Amber.

Is CVSS Automatable the same as SSVC Automatable?

They ask nearly the same question: whether attackers can reliably automate steps 1-4 of the kill chain (reconnaissance, weaponization, delivery, exploitation). They are set by different parties at different times. In CVE-2024-0012 the CNA, Palo Alto Networks, set AU:N in its CVSS vector while CISA's SSVC entry in the same record says Automatable: yes.

Should I use Provider Urgency to prioritise patches?

Use it as one input from the supplier, not as a ranking key. The specification calls Provider Urgency a pass-through metric for the provider's own assessment. Only 1,152 CNA v4.0 entries across all published CVE records set it, so a policy that depends on it will see no value for almost every CVE.

Vulnerability ManagementCvssCvss V4Cve Json 5Supplemental MetricsAutomatableProvider UrgencySsvcParsing