audit-config
Skill1 starsChanged 4 months agoMITView on GitHub
Audit this claude-config repository for drift — unregistered agents, missing Extends links, broken relative links, oversized files, and README translation gaps. Use before releasing a version or after adding a layer, agent, or skill.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- ethereum/execution-specstheir own version, first seen Oct 8, 2026
- ArhamBafna/agent-skillstheir own version, first seen Oct 3, 2026
- stevenprimo-debug/claude-code-audittheir own version, first seen Aug 5, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- ethereum/execution-specs★ 1.2ksame name
- ArhamBafna/agent-skills★ 0same name
- stevenprimo-debug/claude-code-audit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository14 other files it gives its agents.
- .claude/rules/common/agents.mdAGENTS.md
- docs/agents.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- standards-<layer>Skill
- add-agentSkill
- add-standards-layerSkill
- plan-featureSkill
- research-reuseSkill
- review-changesSkill
- shipSkill
- standards-fastapiSkill
- standards-pythonSkill
- standards-typescriptSkill
- tddSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 22
Run every check and report all findings — do not stop at the first one.
- Very specific: line 26
Every file in agents/ must appear in the table in .claude/rules/common/agents.md, and every row in that table must have a file.
- Very specific: line 32
Every topic file in a skills/standards-*/ layer opens with a > Extends [...] line pointing at the layer above it (common/ for a language, …
- Distinctive: line 34
No rule is restated from a layer above.
- Very specific: line 42
Every SKILL.md has name and description.
- Distinctive: line 48
Every relative markdown link resolves to a file that exists.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Audit this repository against its own standards and report what has drifted.
Inventory
ls agents/ skills/ .claude/rules/common/ .claude/rules/_template/ 2>/dev/null
echo "--- roster ---"
grep -c "^| " .claude/rules/common/agents.md
echo "--- line counts ---"
find . -name "*.md" -not -path "./.git/*" -exec wc -l {} + | sort -rn | head -15Checks
Run every check and report all findings — do not stop at the first one.
1. Agent roster sync
Every file in agents/ must appear in the table in .claude/rules/common/agents.md, and every row in
that table must have a file. Nothing enforces this automatically, so it drifts silently. Report
missing rows and orphaned rows separately.
2. Layer discipline
- Every topic file in a
skills/standards-*/layer opens with a> Extends [...]line pointing at the layer above it (common/for a language, the language layer for a framework). - No rule is restated from a layer above. Flag duplicated guidance by topic and quote both copies.
- Process topics (
agents,code-review,development-workflow,git-workflow) appear only incommon/.
3. Frontmatter
- Files in
.claude/rules/common/have no frontmatter — that is what makes them load unconditionally. - Every
SKILL.mdhasnameanddescription. Thenamematches its directory. - Skills with side effects (commits, scaffolding, deploys) set
disable-model-invocation: true. - Skills using
context: forkname anagent, and the agent exists.
4. Links
Every relative markdown link resolves to a file that exists. Check CLAUDE.md, the rules, the skills,
and the docs. Report each broken link with its source file and line.
Two exclusions, both deliberate:
.claude/rules/_template/— its links are written for the destination a copy lands in (skills/standards-<layer>/), so they do not resolve from the template directory by design.- Anything inside a code span or fenced block — those are illustrations of what to write, not links.
5. Size limits
The repository applies its own standard to itself: markdown files ≤800 lines. Flag anything over, and anything approaching it, with a suggested split.
6. Documentation parity
README.md and README.ru.md must cover the same sections in the same order. Report sections present
in one and missing from the other. Same for docs/ and docs/ru/.
7. Claims match reality
Flag any place the documentation describes something that does not exist — a file listed in a directory map that is absent, a command referenced that is not wired, a topic named as shipped that was never written.
Output
Group findings by check, ordered by severity:
- CRITICAL — the config is broken or silently inactive (a rule that never loads, a skill that cannot resolve).
- HIGH — drift that misleads a reader (unregistered agent, docs describing something absent).
- MEDIUM — maintainability (duplication across layers, oversized file).
- LOW — cosmetic.
Give each finding a file path, a line where applicable, and the concrete fix. End with a one-line verdict: clean, or the count at each severity.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

