agentleFS
legalithm-orglegalithm-org/legalithm/skills/eu-ai-act-compliance/SKILL.md

eu-ai-act-compliance

Skill1 starsChanged 4 months agoMITView on GitHub

Use when adding, changing or reviewing an AI feature that will be available in the EU, or when someone asks about the EU AI Act, risk tiers, Article 50 transparency, or what obligations apply. Classifies the use case and cites the Regulation rather than guessing.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    Rule, before anything else: classify runs first: Every risk argument you pass to another tool must come from a classify result in this …

    line 22
  2. Very specific:

    1. Classify: Call classify with the role (provider or deployer), the domain, a plain-sentence description of the use case, and the audience.

    line 29
  3. Very specific:

    2. Get the actual duties: Call explain_obligation with the role and the tier as returned by classify.

    line 34
  4. Distinctive:

    Use those named duties. Do not paraphrase the Act from memory and do not write fear copy.

    line 37
  5. Very specific:

    3. Article 50 transparency: If the feature is user-facing AI, generate the disclosure with generate_disclosure rather than hand-writing …

    line 40
  6. Distinctive:

    Never claim "instant compliance", "compliant by default", or "guaranteed".

    line 55
  7. Very specific:

    Every answer carries an asOf date from the corpus.

    line 56
  8. Distinctive:

    When a result is low-confidence, say so and recommend human review.

    line 58

≈ “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

EU AI Act compliance

The EU AI Act (Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744) applies to AI systems placed on the EU market. This skill routes the work to the legalithm MCP tools, which answer from a versioned rule corpus and cite an Article for every claim.

When to reach for this

  • A new AI feature is being added, or an existing one materially changes
  • Someone asks "does the AI Act apply to us?" or "what do we have to do?"
  • A customer questionnaire, tender, or investor asks about AI compliance
  • A chatbot, generated content, deepfake, or emotion-recognition feature ships

How to work

Rule, before anything else: classify runs first. Every risk argument you pass to another tool must come from a classify result in this conversation, never from your own reading of the case. If you have not called classify yet, you do not know the tier, and calling explain_obligation with a guessed tier returns a confident list of the wrong duties. Guessing "high" and being right is not the same as knowing.

1. Classify. Call classify with the role (provider or deployer), the domain, a plain-sentence description of the use case, and the audience. It returns a risk tier — unacceptable, high, limited, or minimal — with a cited rationale and a confidence. The tier determines everything downstream.

2. Get the actual duties. Call explain_obligation with the role and the tier as returned by classify. Each obligation comes back with its Article citation, for example Article 9 risk management or Article 10 data governance. Use those named duties. Do not paraphrase the Act from memory and do not write fear copy.

3. Article 50 transparency. If the feature is user-facing AI, generate the disclosure with generate_disclosure rather than hand-writing legal text. Scenarios are chatbot (Art 50(1)), genai-content (Art 50(2)), deepfake (Art 50(4)) and emotion (Art 50(3)); English and German are supported. Article 50 has applied since 2 August 2026, so this is a live obligation, not a future one.

4. Keep the record current. compliance/legalithm.json is the committed compliance record. Run npx legalithm check before shipping; if it reports drift, re-run init and review the diff.

How to talk about the results

  • Frame output as checked against Regulation (EU) 2024/1689 — not legal advice. The tools give a cited starting point that a human reviews.
  • Never claim "instant compliance", "compliant by default", or "guaranteed".
  • Every answer carries an asOf date from the corpus. If the user is making a decision that depends on a date, surface it rather than hiding it.
  • When a result is low-confidence, say so and recommend human review. A confident wrong answer about a legal obligation is worse than no answer.

What runs where

classify, explain_obligation and generate_disclosure run entirely offline — the rule corpus ships with the server, so no source code, prompt or result leaves the machine. check_record is the only tool that makes a network call, fetching a public Trust Center record by the org slug you give it.

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.