claude-md-audit
Checks AI project documentation and memory files for accuracy, drift, and broken paths.
Install
mkdir -p .claude/skills/claude-md-audit && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/19008" && unzip -o skill.zip -d .claude/skills/claude-md-audit && rm skill.zipInstalls to .claude/skills/claude-md-audit
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Audit CLAUDE.md and the memory store (MEMORY.md + fact files) for drift — stale path citations, cross-file contradictions, duplicated facts, and bloat. Use periodically (monthly, or after a big restructure), not on every change. Read-only by default; reports findings by severity with file:line + suggested fix, and routes anything not fixed this session to a follow-up note. Auto-activates on `/claude-md-audit`.Key capabilities
- →Extract and confirm path citations in CLAUDE.md and MEMORY.md
- →Identify cross-file contradictions in claims
- →Detect duplication and bloat in facts
- →Check for broken index links and orphan memories
- →Report findings by severity with suggested fixes
How it works
The skill performs four checks: stale references, cross-file contradictions, duplication/bloat, and structural hygiene. It extracts paths, compares claims, and identifies issues, then reports them by severity.
Inputs & outputs
When to use claude-md-audit
- →Clean up stale project documentation
- →Audit AI memory integrity
- →Sync instruction files after repo restructuring
About claude-md-audit
Performs periodic audits on CLAUDE.md and memory files to ensure file paths and instructions remain accurate. Detects drift, contradictions, and duplication in the project's knowledge base.
Audit CLAUDE.md and the memory store (MEMORY.md + fact files) for drift — stale path citations, cross-file contradictions, duplicated facts, and bloat. Use periodically (monthly, or after a big restructure), not on every change. Read-only by default; reports findings by severity with file:line + sug
When not to use it
- →On every change to the repository
- →When not performing a periodic audit (monthly or after a big restructure)
Limitations
- →Read-only by default, offers fixes but does not rewrite without approval
- →Does not auto-fix dangling `[[refs]]` as they may be deliberate forward-references
- →Not intended for use on every change
How it compares
This skill provides a periodic, read-only audit of documentation and memory files to catch drift and maintain consistency, unlike real-time checks on every commit.
Compared to similar skills
claude-md-audit side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| claude-md-audit (this skill) | 0 | 3mo | Review | Intermediate |
| wrap | 0 | 9mo | Review | Intermediate |
| session-wrap | 1 | 9mo | Review | Intermediate |
| windsurf-custom-prompts | 1 | 5mo | No flags | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
wrap
team-attention
세션 종료 시 자동 정리를 수행합니다. CLAUDE.md 업데이트 제안, 반복 패턴 자동화 기회 탐지, 배운 것 정리, 미완성 작업 정리. 사용자가 /wrap 또는 세션 마무리 요청 시 사용.
session-wrap
team-attention
This skill should be used when the user asks to "wrap up session", "end session", "session wrap", "/wrap", "document learnings", "what should I commit", or wants to analyze completed work before ending a coding session.
windsurf-custom-prompts
jeremylongshore
Create and manage custom prompt libraries for Cascade. Activate when users mention "custom prompts", "prompt library", "prompt templates", "cascade prompts", or "prompt management". Handles prompt library creation and organization. Use when working with windsurf custom prompts functionality. Trigger with phrases like "windsurf custom prompts", "windsurf prompts", "windsurf".
analyse-issue
monarch-initiative
Analyze MONDO GitHub issues for validity, suggest improvements, and generate structured reports with duplication checks and identifier validation
gsd-validate-phase
lza6
Retroactively audit and fill Nyquist validation gaps for a completed phase
file-justification
jleechanorg
**Purpose**: Before creating or modifying ANY file, you MUST document the justification to prevent unnecessary file creation and ensure changes are well-reasoned.