agentleFS
TsukiNi22TsukiNi22/ai-utils/skills/legal/legal-check/SKILL.md

legal-check

Skill1 starsChanged 4 days agoMITView on GitHub

Deep legal compliance check of a project or a website, global (any project handling data) or web scope: GDPR / RGPD (bases, consent, notices, rights incl. erasure, records, DPAs, DPIA, retention, security, breaches, transfers), cookies (ePrivacy, CNIL), legal notice (LCEN), terms (CGU / CGV), consumer law, marketing emails, accessibility (RGAA, EAA), language (loi Toubon), the licenses and credits of every dependency / font / asset / model, mandatory mentions, DSA, AI Act, CRA, NIS2, Data Act, UK / Swiss / US laws - each finding with its article, the evidence in the project and the fix, dates checked on the primary texts, delivered as a report. Use whenever the user asks if a project / site is legal or compliant, a RGPD / GDPR / cookies / CGU / mentions légales / RGAA / accessibility / privacy / right to erasure / AI Act / licenses check, "est-ce légal", "audit légal", "conformité".

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Third parties / recipients: every SDK, script, API key, env var, infra file (docker-compose, Terraform, CI secrets), DNS / CDN, email …

    line 27
  2. Very specific:

    Dates and statuses marked ⏱: check them against references/sources.md (EUR-Lex, Légifrance, CNIL, the Legislative Observatory...) with a …

    line 41
  3. Distinctive:

    Look past the checklist: anything else in this project that a regulator, a user or a rights holder could object to (a dark pattern, a …

    line 45

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Legal check (compliance audit)

Find everything legal that can cause a problem in a project, prove it from the project itself, and say how to fix it. Read only: nothing is changed (the fixes: legal-setup). Not legal advice: said once in the report; anything binding (a contract, a sanction risk, a sensitive processing) is flagged "to verify with a lawyer / DPO".

Request given with the skill: the text after /legal-check if any (a scope, a project, a site URL).

1. Scope (skill interactive-questions, one round, only what the project does not tell)

  • Scope: global (any project handling data: references/global.md) · web (a site / web app: global.md + references/web.md) · both. A project with a web front: both, recommended.
  • The facts the checklist needs and the code cannot give: the countries of the users (EU / France, UK, Switzerland, US states, others), B2C or B2B, the organisation (company / individual / public body, its size: micro, < 250 staff, turnover > €250 M), the role (controller / processor), sector (health, finance, minors, public, AI), a live URL to test.
  • Unknown answers become to verify items, never guesses.

2. Inventory (read the project, the live site when given)

  • Personal data: models / schemas / migrations, forms, API payloads, logs, analytics events, files uploaded, emails, backups, caches, search indexes, AI prompts. A table: data · where stored · why · who receives it · kept how long.
  • Third parties / recipients: every SDK, script, API key, env var, infra file (docker-compose, Terraform, CI secrets), DNS / CDN, email provider, payment, error tracking, AI API, fonts, embeds - with their country.
  • Web: load the site with an empty browser profile (or curl -I + the HTML when no browser): cookies / storage set before any consent, third-party requests, the banner and its buttons, the footer links (legal notice, privacy, cookies, terms, accessibility), the forms and their notices, HTTPS / headers.
  • Documents present: privacy policy, legal notice, terms, cookie policy, accessibility statement, records, DPAs, retention policy, breach procedure, security.txt / SECURITY.md, LICENSE / NOTICE.

3. Check (every applicable item of the references, none skipped)

  • Go through references/global.md (and web.md for the web scope) in order; each item gets a status: ok · missing · partial · n/a (+ why) · to verify.
  • Each non-ok item: the legal basis (article), the evidence (file:line, URL, cookie name, screenshot / request seen), the risk (who can sanction, the order of the fines: GDPR up to 4 % / €20 M, LCEN, DGCCRF, CNIL decisions), the fix (what to add / change / remove) and its effort (S / M / L).
  • Dates and statuses marked ⏱: check them against references/sources.md (EUR-Lex, Légifrance, CNIL, the Legislative Observatory...) with a web search / fetch before writing them; a blog alone never settles a date; conflicting sources: say both and mark to verify. Update sources.md when a status changed (said in the report).
  • Licenses of the dependencies and of the content: use audit-deps / license and include their result.
  • Look past the checklist: anything else in this project that a regulator, a user or a rights holder could object to (a dark pattern, a misleading claim, scraping, data sold, staff monitoring, a sensitive field) is added.

4. Report (skill report, format asked: Markdown and / or PDF)

  • Summary: the scope, the laws that apply and why, the counts by severity, the 5 most urgent fixes.
  • Findings table sorted by severity: blocking (illegal now: trackers before consent, no legal notice, no lawful basis, data never deleted, plaintext passwords...) · high · medium · low · then to verify and ok.
  • The inventory tables (data, recipients, documents present / missing), the n/a items with their reason, the sources checked with their date, the "not legal advice" line.
  • Saved in audit/ at the repository root (audit/<YYYY-MM-DD>-legal.md / .pdf, the naming of report).
  • Last: offer legal-setup for the fixes (through the popup: all / the blocking ones / a choice / later).

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.