agentleFS
Sign inStart free
ArangoGutierrezArangoGutierrez/claude-toolkit/.claude/skills/config-audit/SKILL.md

config-audit

Skill4 starsChanged 3 months agoMITView on GitHub

Audit your own .claude config surface (skills/hooks/MCP/settings) for secrets, injection sinks, over-broad permissions, and hook hygiene. Triggered by "audit my config", "scan .claude", or /config-audit

We split this skill into the rules it gives an agent

3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run the scanner on the live config: bash scripts/scan-config.sh "$HOME/.claude" Optionally also scan this repo: bash scripts/scan-config.sh …

    line 20
  2. Very specific:

    Add # config-audit:ignore <category|all> on a line to suppress a known-safe match.

    line 39
  3. Distinctive:

    Read-only by design — never auto-fix.

    line 47

“Common” is measured: kept by at least 6 in 10 of the 4 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Config Audit — Self-Audit of the Agent Surface

Audits the .claude configuration itself (not project code). Read-only scan, then propose fixes for your approval.

Process

  1. Run the scanner on the live config: bash scripts/scan-config.sh "$HOME/.claude" Optionally also scan this repo: bash scripts/scan-config.sh .claude
  2. Parse the TSV output: SEVERITY<TAB>CATEGORY<TAB>FILE:LINE<TAB>MESSAGE.
  3. Present a severity-ranked report (high → low) with file:line and a concrete fix.
  4. For each finding, propose a fix. Apply only after user approval, in a separate turn and a separate commit. The skill never mutates config silently.
  5. On completion, record the run: date +%Y-%m-%d > "$HOME/.claude/audit/.last-config-audit".

Categories

  • secrets — hardcoded tokens/keys/passwords/PEM blocks.
  • injection-sink — untrusted content piped to a shell / eval.
  • broad-perms — sandbox/permission bypass, wildcard Bash grants.
  • hook-hygiene — unhardened shell scripts, executable .bak backups.
  • mcp-hygiene — too many enabled MCP servers (>10) inflating context.

Suppression

Add # config-audit:ignore <category|all> on a line to suppress a known-safe match.

Exit Codes (for CI later)

0 clean · 1 low/medium · 2 high/critical · 3 scan aborted / nothing scanned (fails closed — e.g. missing target dir, mktemp failure, or a findings-buffer write error; never reports clean on abort).

Gotchas

  • Read-only by design — never auto-fix. Propose, then apply on approval.
  • Re-run after fixes to confirm the finding is gone.
  • A finding is not a vulnerability until triaged; review before acting.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free