doc-auditor — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited doc-auditor (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Hostile documentation review. Assumes every claim in every doc is wrong until verified against the codebase. Two-directional: docs→code (does code match the claim?) and code→docs (is this code undocumented?). Every finding includes the document source, the false or missing claim, and a concrete fix.
When NOT to use: For architecture-specific documentation drift, run arch-auditor instead — it validates structural claims against the detected architecture.
README.md at any directory leveldocs/ directories (excluding docs/audit/ — its own output)| Type | Description |
|---|---|
| Stale | References a function, class, module, file, or parameter that no longer exists or has been renamed |
| Incorrect | Documented behavior contradicts implementation — wrong parameters, wrong return type, wrong description of what the code does |
| Missing | Public API, exported function, architectural boundary, or module with no documentation |
| Outdated architecture | Describes a structural pattern, layer, or component that has changed or been removed |
| Contradiction | Two documentation sources make conflicting claims about the same thing |
| Broken link | Internal link to a file, heading, or section that does not resolve |
| Severity | Meaning |
|---|---|
| Critical | Actively misleads — incorrect behavior description that would cause wrong usage of the code |
| High | Missing documentation for a public API or architectural boundary |
| Medium | Stale reference, outdated architecture description, contradiction |
| Low | Broken internal link, minor inaccuracy, test name that misrepresents behavior |
| Fix | Applied automatically after asking |
|---|---|
| Remove stale references | Yes |
| Update renamed identifiers | Yes |
| Fix broken internal links | Yes |
| Generate missing docs from code | Yes — inferred from signatures, types, and context |
| Update incorrect behavior descriptions | Yes — rewritten from actual implementation |
| Update architecture descriptions | Yes — rewritten from docs/audit/arch-profile.md if present |
| Resolve contradictions | Yes — picks the version consistent with code; notes the resolution |
1. Collect all documentation sources (see Documentation Scanned above)
2. Create docs/audit/ if it does not exist
3. If docs/audit/doc-findings.md exists:
Re-validate each OPEN finding:
- Issue resolved → move to Fixed (record date)
- Finding was wrong → move to Invalid (record reason)
- Still present → leave as Open
4. Audit all docs against codebase (both directions)
5. Add new findings (assign next available ID — never reuse IDs)
6. Present findings summary
7. Ask: "Apply fixes now? (y/n)"
If yes: apply fixes, re-validate, update findings file
8. Write docs/audit/doc-findings.md
9. Ask: "Commit findings to git? (y/n)" — never commit silently# Documentation Audit Findings
Generated: YYYY-MM-DD
Last validated: YYYY-MM-DD
## Summary
- Total: N | Open: N | Fixed: N | Invalid: N
## Open Findings
### Critical
#### [ID] Short title
Category: <stale|incorrect|missing|outdated-architecture|contradiction|broken-link>
Source: <path to the document containing the claim>
Claim: <the exact claim made in the document>
Evidence: <what the code actually does — file path and relevant detail>
Fix: <minimal correction>
### High
[same structure]
### Medium
[same structure]
### Low
[same structure]
## Fixed
### Pass N — YYYY-MM-DD
#### [ID] Short title
Fixed: YYYY-MM-DD
Notes: <what changed>
## Invalid
### Pass N — YYYY-MM-DD
#### [ID] Short title
Notes: <why this finding was wrong>## Fixed h2; all invalid findings go under one ## Invalid h2. Sub-divide each by ### Pass N — YYYY-MM-DD h3 headers. Never create ## Fixed — pass N h2 variants. Never skip header levels (h2 → h4 with no h3 is a structural gap).~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.