Verdicts and justifications

Four fields on an entry record your judgement: verdict, severity, disposition, and justification. Of these, only verdict and disposition, together with resolution, decide the entry’s state.

Synopsis

An entry with the verdict affected has a severity and, optionally, a disposition:

    verdict: affected
    severity: medium
    disposition: will fix

An entry with the verdict not affected has a justification:

    verdict: not affected
    justification: vulnerable code not in execute path

An entry without verdict is under investigation.

Field combinations

verdict severity disposition justification

Not set

Ignored

Not allowed

Ignored

affected

Required

Optional

Ignored

not affected

Ignored

Not allowed

Required

The CLI ignores justification without a verdict or with verdict: affected, but the JSON schema rejects it.

Verdict values

Value Meaning

Not set

You’re still investigating whether the vulnerability affects your software.

affected

The vulnerability affects your software.

not affected

The vulnerability doesn’t affect your software.

The CLI rejects under investigation as a value of verdict.

There is no fixed verdict. An entry records the fix in resolution. See Resolution.

Severity values

Value Meaning

critical

Immediate action required. Exploitable with severe impact.

high

Action required. Significant impact if exploited.

medium

Action recommended. Moderate impact.

low

Low priority. Minimal impact or difficult to exploit.

Disposition values

Value Meaning

Not set

You haven’t stated what you plan to do.

will fix

You plan to fix the vulnerability.

wont fix

You accept the risk and don’t plan to fix it.

Justification values

The values match the justifications of OpenVEX documents, which tell your customers' tools whether a release is affected by a vulnerability.

Value Meaning

component not present

The vulnerable component isn’t part of the software you ship.

vulnerable code not present

The component is present, but the vulnerable code isn’t.

vulnerable code not in execute path

The vulnerable code is present, but your software never runs it.

vulnerable code cannot be controlled by adversary

Your software runs the vulnerable code, but an attacker can’t trigger the vulnerability.

inline mitigations already exist

Existing controls, such as input validation or a web application firewall, prevent the attack.

States

Vulnlog derives a state for each entry from its verdict, disposition, and resolution. You don’t write the state into the file. The impact report shows it, and its --state filter selects by it.

State Condition

under investigation

No verdict.

open

affected, no resolution, and disposition is will fix or not set.

accepted

affected, no resolution, and disposition: wont fix.

resolved

affected or not affected, with a resolution.

not applicable

not affected, no resolution.

With --as-of, a resolution in a later release doesn’t count. As of 4.2.0, Log4Shell is open, because 4.3.0 fixes it. For the option, see Filter options. For why states work this way, see Vulnerability states.

Rules

vulnlog validate checks these rules.

Rule Level Message

verdict is affected, not affected, or not set.

Error

Invalid verdict: fixed

An entry with the verdict affected has a valid severity.

Error

Invalid severity: null

An entry with the verdict not affected has a valid justification.

Error

Invalid justification: null

disposition appears only with verdict: affected.

Error

Disposition requires verdict 'affected'.

disposition is will fix or wont fix.

Error

Invalid disposition: maybe

An entry with severity critical doesn’t have disposition: wont fix.

Info

A critical severity vulnerability is marked 'wont fix'.

Examples

The lodash entry is accepted. It’s affected with severity low, and its disposition is wont fix:

  - id: CVE-2021-23337
    description: Command injection through the template function in lodash
    releases:
      - 4.2.0
      - 4.3.0
    packages: ["pkg:npm/lodash@4.17.20"]
    reports:
      - reporter: npm-audit
        at: 2026-03-10
        suppress: {}
    tags: [frontend]
    analysis: >-
      The template function is only used in the admin interface, with templates that administrators write. Administrators can
      already change the shop configuration, so the injection gives them no additional access. The admin interface is removed
      in Acme Shop 5.0.
    analyzed_at: 2026-03-12
    verdict: affected
    severity: low
    disposition: wont fix

The XZ Utils backdoor doesn’t affect Acme Shop, because Debian 12 doesn’t ship the backdoored versions:

  - id: CVE-2024-3094
    name: XZ Utils backdoor
    description: Backdoor in the liblzma library of XZ Utils
    releases: [4.3.0]
    packages: ["pkg:deb/debian/xz-utils@5.4.1"]
    reports:
      - reporter: other
        at: 2026-07-02
        source: Customer inquiry
    tags: [container]
    analysis: >-
      The container image is based on Debian 12, which ships xz-utils 5.4.1. The backdoor is only in the upstream releases 5.6.0
      and 5.6.1.
    analyzed_at: 2026-07-02
    verdict: not affected
    justification: vulnerable code not present

The Log4Shell entry is affected and has a resolution in 4.3.0, so its state is resolved:

  - id: CVE-2021-44228
    name: Log4Shell
    description: Remote code execution through JNDI lookups in Apache Log4j
    aliases: [GHSA-jfh8-c2jp-5v3q]
    releases: [4.2.0]
    packages: ["pkg:maven/org.apache.logging.log4j/log4j-core@2.14.1"]
    reports:
      - reporter: grype
        at: 2026-03-10
      - reporter: other
        at: 2026-03-11
        source: Customer support request
    tags:
      - backend
      - container
    analysis: >-
      The server logs the User-Agent header with log4j-core 2.14.1. A crafted header triggers a JNDI lookup and runs attacker-controlled
      code.
    analyzed_at: 2026-03-11
    verdict: affected
    severity: critical
    disposition: will fix
    resolution:
      in: 4.3.0
      at: 2026-06-15
      note: Updated log4j-core from 2.14.1 to 2.17.1