agentleFS
bamlabbamlab/flashlight/.claude/skills/gdpr-audit/SKILL.md

gdpr-audit

Skill1.6k starsChanged 3 days agoMITView on GitHub

Audit the codebase for GDPR and secret-handling problems — personal data in logs, in third-party payloads (Sentry, Slack, Discord, AWS), in fixtures, committed secrets, retention. Use when asked for a GDPR/RGPD/privacy audit, before a release touching user data, or by the factory's weekly run.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Run on the current repo. In bamlab/flashlight-cloud, include flashlight-open-source/ (the public submodule).

    line 12
  2. Very specific:

    Secrets in git: Run gitleaks git . --no-banner --redact --report-format json --report-path /tmp/gitleaks.json and gitleaks dir .

    line 18
  3. Very specific:

    Personal data in logs: Search logger and console calls whose arguments include email, user, name, ip, whole request/response bodies, or …

    line 19
  4. Very specific:

    Fixtures and test data: Look for real emails, real device dumps, signed URLs (X-Amz-Signature), tokens, phone numbers in __tests__, …

    line 21
  5. Distinctive:

    Don't fix anything in the same run unless asked: an audit reports.

    line 37

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “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

GDPR audit

Find places where personal data or secrets leak, and report them privately. Read .claude/rules/privacy-and-security.md first: it defines personal data and where findings may be written.

Scope

Run on the current repo. In bamlab/flashlight-cloud, include flashlight-open-source/ (the public submodule).

Checks

Work through each check. For each hit, record file:line, what personal data or secret is involved, where it goes, and a severity.

  1. Secrets in git. Run gitleaks git . --no-banner --redact --report-format json --report-path /tmp/gitleaks.json and gitleaks dir . --no-banner --redact. Read results with the repo's .gitleaks.toml in mind. Transcrypt-encrypted files (git check-attr filter <file> says crypt) are encrypted in git: a hit on them only means your local checkout can decrypt them — not a leak. Never print secret values.
  2. Personal data in logs. Search logger and console calls whose arguments include email, user, name, ip, whole request/response bodies, or whole entity objects (Logger.*(, console.*(, morgan, morganBody). Include ops scripts.
  3. Third-party payloads. For each outbound integration (Sentry init/captureException/setUser/beforeSend, Slack, Discord, email, AWS Device Farm specs, analytics events), list which personal fields are sent and whether they're needed.
  4. Fixtures and test data. Look for real emails, real device dumps, signed URLs (X-Amz-Signature), tokens, phone numbers in __tests__, fixtures, *.json, *.txt.
  5. Copies of production data. Scripts or runbooks that copy prod data elsewhere (staging, local disk) without anonymisation.
  6. Retention and deletion. Is there a way to delete a user's data (runs, uploads, videos, auth user)? Are uploads/videos ever expired?
  7. Consent and analytics. Analytics is cookieless (Plausible) and consent tooling is loaded where needed; no personal data in analytics event params.

Severity

  • critical — a live secret exposed, or personal data publicly readable.
  • high — personal data sent to a third party or log store without need, or no way to delete user data.
  • medium — personal data in internal logs/notifications, or real personal data in fixtures.
  • low — hygiene (missing scrubbing on a path that currently carries no personal data).

Output

  1. A table: severity · check · file:line · finding · suggested fix.
  2. Route every finding with /security-findings — never into a public issue.
  3. Don't fix anything in the same run unless asked: an audit reports.

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.