agentleFS
Edzard9Edzard9/stash/skills/ops/repo-audit/SKILL.md

repo-audit

Skill0 starsChanged 8 days agoMITView on GitHub

Audit your own public repos for leaked secrets and quiet exposures before the world finds them: scan files and full commit history for API keys, tokens, private keys, .env files and hardcoded passwords, then check the leaks that are not code (commit author emails, screenshots in READMEs, hostnames). Use when the user wants to check a repo is safe to share, says "audit repo", "cek repo aman", "scan for secrets", "periksa kebocoran". Do NOT use to audit other people's repos for their secrets.

  • Reads credentials

25 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use handoff and code-review.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    handoff

    Compact the current conversation into a handoff document for another agent to pick up.

    Also in 3 projects that use this

  • GoogleSkill

    code-review

    Adversarial systems, concurrency, memory-model, and architecture code review protocol. Use this skill whenever asked to review code, PRs, diffs, or branches in Filament.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 21 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Search current files and every commit for:

    line 30
  2. Distinctive:

    Use the search tooling available to you.

    line 43
  3. Distinctive:

    README and docs media: Read every screenshot and image in the README and docs as an attacker would: hostnames, usernames, file paths, IP …

    line 56
  4. Distinctive:

    Not a penetration test and not a code review.

    line 85
  5. Distinctive:

    Not permission to rewrite history, force-push, or rotate keys on the agent's own initiative.

    line 88

“Common” is measured: kept by at least 6 in 10 of the 21 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

Repo Audit

Nobody finds their own leaked API key by reading their repo. They find it in a stranger's screenshot, or in a bill. This audit exists because a real pass over 15 public repos found zero leaked keys, and still found two real exposures hiding outside the code: the owner's private email as the commit author on nearly every commit, and a README screenshot showing a server hostname and service list. Secrets are only half the audit. This skill is the whole checklist, in order.

Ground rules

  • Audit only repos the user owns or is authorized to audit.
  • Report values by type and location, never reproduce a found secret in the report, the chat, or any file. "A Stripe-shaped key in config.js, line 12, present since commit a1b2c3d" is the full detail. The key itself never travels.
  • History counts. A secret deleted in the latest commit is still public in every earlier one. The audit covers all of git history, not the working tree alone.

Pass 1: files and history, secret shapes

Search current files and every commit for:

  1. Credential shapes: API key and token patterns (sk-, AKIA, ghp_, xox, Bearer values), password =, secret =, connection strings with embedded credentials.
  2. Key material: files named *.pem, *.key, id_rsa*, anything under .ssh/, and private-key headers inside any file.
  3. Environment files: .env and variants, committed config with real values where a .env.example was intended.
  4. Infrastructure identifiers in code: production IPs, internal hostnames, database hosts, bucket names that are not meant to be public.

Use the search tooling available to you; the point is coverage of all history, not a specific scanner. Note false positives as such (example keys, test fixtures) instead of deleting them from the report.

Pass 2: the quiet exposures

  1. Commit author emails. List the author emails across history. A private address used as author is public data on every clone. The fix is forward-looking: enable the host's "keep my email private" setting and use the provided noreply address for future commits. Rewriting published history is a separate, explicit decision, never a silent side effect of an audit.
  2. README and docs media. Read every screenshot and image in the README and docs as an attacker would: hostnames, usernames, file paths, IP addresses, dashboard contents, browser tabs, notification text. List what each image discloses.
  3. Issue and PR text. Secrets and internal details get pasted into issues during debugging. Scan open and closed issues for the same shapes as Pass 1.
  4. Descriptions and metadata. Repo description, homepage field, and release notes sometimes carry internal URLs. Read them too.

The report

Per repo, in this shape:

Repo: <name>
Files + history: clean / findings (type, file, since which commit)
Author emails: <addresses, flagged if private>
Media disclosures: <image, what it shows>
Issues/PRs: clean / findings
Verdict: safe to share / fix first (list)

Close with the fixes in priority order: rotate anything real that leaked (a found key is a burned key, deletion is not rotation), then the forward-looking settings, then the media replacements.

What this skill is not

  • Not a penetration test and not a code review. It answers one question: what does this repo disclose that its owner did not mean to publish?
  • Not permission to rewrite history, force-push, or rotate keys on the agent's own initiative. Those are the owner's calls, made with the report in hand.

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.