- Home
- Skills
- Legal & Compliance
- EU AI Act Readiness Kit: Is Your AI Product High-Risk?
More screenshots
Works with the AI Agents you already use
EU AI Act Readiness Kit: Is Your AI Product High-Risk?
The EU AI Act is now in force step by step, and the 2026 Digital Omnibus moved the high-risk rules to 2 December 2027 (Annex III) and 2 August 2028 (AI in regulated products).
$29
EU AI Act Readiness Kit: Is Your AI Product High-Risk?
Example session with this skill installed
We're an EU start-up and we sell the resume screening tool in this repo to employers. Check it against the EU AI Act: what risk category is it, what do we have to do and by when, and draft the technical documentation.
- Read your context and instructions
- Compiled the eu ai act
Real output: a fresh AI agent ran this skill on the open-source project sliday/resume-job-matcher (an employer-side resume screening CLI). Nobody was available to answer its questions, so it lists its assumptions.
This is a readiness check, not legal advice.
Result: HIGH-RISK AI system. It analyses and filters job applications and evaluates candidates (pre-filter, pass/fail gate, scores, ranking, rejection e-mails), which is Annex III point 4(a) (Art 6(2)); it profiles candidates, so the Article 6(3) exemption cannot apply. It also has transparency duties under Article 50 (it generates e-mail text).
Already applies:
- AI literacy measures (Art 4) - since 2 Feb 2025
- Machine-readable marking of the generated e-mails (Art 50(2)) - since 2 Aug 2026 (2 Dec 2026 if already on the market before 2 Aug 2026)
Coming next:
- All high-risk provider duties (Art 9-17, 43 internal control, 47-49 declaration/CE/registration, 72-73) - from 2 Dec 2027 (376 days)
- Your employer customers (deployers) must then use it as instructed, assign human oversight, inform workers' representatives and candidates, and keep logs (Art 26).
Do this first:
- Put a human decision in the loop: a recruiter must confirm every gate rejection and every e-mail before it goes out (Art 14). Fully automated rejections may also raise GDPR questions; get privacy advice.
- Test for discrimination: location, language and work-authorisation criteria can act as proxies for nationality or origin; "soft skills" and "education" ratings need checking too (Art 10, Art 15).
- Pin the models: the default
openrouter/autoand env overrides can change the model without a code change, and the README reports 20+ point score differences across models. - Add persistent decision logs (Art 12): today there are only console output and e-mail files.
- Continue the Annex IV draft (21 of 26 items have content, 9 still need the team) and start risk management and the quality management system.
Files in out/resume-job-matcher/: report.md, result.json, technical-documentation-draft.md, transparency-notices.md, ai-literacy-plan.md, scan.md, scan.json, profile.json
Assumptions and open questions: (no user was available)
- You are an EU start-up and the provider; employers and agencies deploy it; already offered in the EU, first release date unknown.
- Whether systems already on the market before 2 Dec 2027 get any transition (Art 111(2)) was not assessed by the kit.
- Company size unknown (SME relief may apply).
Please have a qualified EU AI Act lawyer confirm this before relying on it.
Connects securely to your tools. The creator never sees your data.
About this skill
Building an AI product for European customers? The EU AI Act is now in force step by step, and the 2026 Digital Omnibus moved the high-risk rules to 2 December 2027 (Annex III) and 2 August 2028 (AI in regulated products). Transparency duties already apply since 2 August 2026, and two new bans start on 2 December 2026.
This skill turns your AI coding agent (Claude Code, Codex, Cursor, Gemini CLI) into a careful EU AI Act readiness assistant:
- Scans your codebase for AI use: model libraries, calls to OpenAI, Anthropic, Gemini, Mistral and others, face/voice biometrics, emotion recognition, chat interfaces, generated media (with file:line evidence).
- Fills a structured questionnaire from the code, your docs and your answers (role, intended purpose, company size).
- Checks all 10 prohibited practices, the high-risk routes (Annex I products, Annex III use cases), the Article 6(3) exemption and the profiling rule, transparency duties (Art 50) and general-purpose AI rules.
- Writes a report: category with the articles behind it, every obligation with who / what / date / days left, penalties.
- Drafts documents: Annex IV technical documentation (prefilled from your code where possible), Article 6(3) assessment record, transparency notices, safeguards record for image/video generators, AI literacy plan.
Tested honestly: A fresh AI agent that had never seen them followed the frozen skill on 166 practical examples from the European Commission's own draft high-risk guidelines (May 2026, AI Act Service Desk): 161 / 166 correct (97%), 64 / 64 high-risk systems recognised, 5 / 102 not-high-risk systems wrongly flagged (all errors on the cautious side). A buyer test on 6 real open-source AI products matched my pre-written expected answer on 5 and partly on 1 (fixed in v1.0.1). 44 automated tests. Full results: https://m-shahmeer-kz.github.io/eu-ai-act-readiness-kit/
Not legal advice. Every report says so and recommends that a qualified lawyer confirms the result.
Limitations
- Readiness aid, not legal advice; it can be wrong and the guidance is still being finalised.
- The high-risk test uses the Commission's draft guidelines (May 2026); final guidelines may change some borderline cases.
- Does not cover GDPR, copyright, product liability, consumer or sector-specific law.
- The technical documentation draft is a starting point: the team must fill risk decisions, test results and the declaration of conformity.
- Results depend on the answers given; when the user is not available the agent states its assumptions.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Legal & Compliance.
Not legal advice: a readiness aid that can be wrong; have a lawyer confirm the result. The high-risk test follows the Commission's draft guidelines (May 2026), which may still change. Law as amended by Regulation (EU) 2026/1744, checked 21 Nov 2026. Does not cover GDPR, copyright, product liability or sector law. The Annex IV draft needs the team's own risk decisions, test results and declaration of conformity. Requires Python 3.10+.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
- Software
Python 3.10+
Permissions
File Scopes
Runs one local Python script (standard library only). It reads the project files you point it at to look for AI use, and writes its report and document drafts to a .aiact/ folder. No network access, no API keys, nothing is uploaded.