compliance-team-eu-ai-act — Claude Code plugin for AI Act work
Three stdlib scripts classify an AI system under the EU AI Act, plan the conformity route and build an obligation matrix for your organizational role.
/plugin marketplace add alirezarezvani/claude-skills
/plugin install compliance-team-eu-ai-act@claude-code-skillsWhat compliance-team-eu-ai-act does
The plugin handles the operational side of Regulation (EU) 2024/1689. ai_system_risk_classifier.py takes system characteristics and returns a tier: prohibited under Article 5, high-risk under Article 6 and Annex III, limited-risk under Article 50, or minimal-risk. conformity_assessment_planner.py picks Module A or Module H for a high-risk system per Article 43 and appends the Annex IV technical documentation checklist. ai_act_obligation_tracker.py maps obligations and deadlines onto the role from Article 25: provider, deployer, importer, distributor or authorized representative.
Four references sit behind the scripts: an Article-by-Article walkthrough of Titles I–XII, the eight Annex III high-risk categories with the Article 6(2) carve-outs, GPAI duties under Articles 51–55 including the systemic-risk threshold, and a mapping to ISO 42001, NIST AI RMF and GDPR. Every script runs on the Python standard library alone. Pass no JSON path and they fall back to embedded samples, which is a quick way to see the output shape before wiring in your own data.
This is not legal advice. Novel questions, such as whether a chatbot counts as a general-purpose AI model, still need outside counsel. ISO 42001 and GDPR live in separate plugins. And remember Article 2: placing a system on the EU market puts you in scope wherever you are established.
What it helps with
- classify an AI system against Article 5, Article 6 and Annex III
- choose between Module A and Module H for conformity assessment
- build the Annex IV technical documentation checklist
- split obligations across provider, deployer, importer and distributor roles
- map AI Act duties onto ISO 42001, NIST AI RMF and GDPR
Example prompts for the agent
- classify our customer support system under the EU AI Act
- draft a conformity assessment plan for a high-risk system
- what are our obligations and deadlines as a deployer under Article 25
How to install compliance-team-eu-ai-act
Plugin marketplace
/plugin marketplace add alirezarezvani/claude-skills/plugin install compliance-team-eu-ai-act@claude-code-skillsStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about compliance-team-eu-ai-act
How do I install compliance-team-eu-ai-act in Claude?
Add the marketplace with /plugin marketplace add alirezarezvani/claude-skills, then install the plugin: /plugin install compliance-team-eu-ai-act@claude-code-skills.
What does compliance-team-eu-ai-act do?
It ships three stdlib tools: a risk classifier for AI systems, a conformity assessment planner for high-risk ones, and an obligation tracker keyed to your organizational role. Four references cite the regulation's articles.
Does it need any external libraries?
No. All three scripts run on the Python standard library. Without a JSON path they execute against embedded samples.
Source text
compliance-team-eu-ai-act · MIT
# compliance-team-eu-ai-act Standalone plugin for **Regulation (EU) 2024/1689 — the EU Artificial Intelligence Act** compliance. **Dual-published**: also bundled inside `ra-qm-skills` (`../skills/eu-ai-act-specialist/`). Manually mirrored to `ra-qm-team/skills/eu-ai-act-specialist/`. See `./skills/eu-ai-act-specialist/SKILL.md` for the full skill documentation. ## What this is The EU AI Act is the world's first comprehensive horizontal regulation of AI systems. Adopted in 2024 (Regulation (EU) 2024/1689; OJEU L of 12 July 2024), it entered into force on 1 August 2024 and applies in phases: | Date | Obligation | |---|---| | **2 Feb 2025** | Article 5 (prohibited AI practices) + Article 4 (AI literacy) in force | | **2 Aug 2025** | GPAI obligations (Articles 51–55) + governance + penalties in force | | **2 Aug 2026** | Article 50 transparency obligations (AI interaction disclosure, deepfake and synthetic-content labelling) | | **2 Dec 2027** | High-risk obligations for stand-alone Annex III systems (deferred from 2 Aug 2026 by Regulation (EU) 2026/1744, the AI Omnibus) | | **2 Aug 2028** | High-risk obligations for products already covered by sectoral law, such as medical devices under MDR/IVDR (Annex I; deferred from 2 Aug 2027) | The Act is binding and directly applicable across all 27 EU Member States. Penalties reach EUR 35M or 7% of worldwide annual turnover (whichever is higher) for Article 5 violations. ## What this plugin provides Three deterministic stdlib tools that operate at the Article level: 1. **`ai_system_risk_classifier.py`** — input: AI system characteristics → output: tier (prohibited / high-risk / limited-risk / minimal-risk) with citing Article and Annex 2. **`conformity_assessment_planner.py`** — input: high-risk AI system + classification → output: Module A (internal control) vs Module H (full QMS) routing per Article 43 + Annex IV technical documentation checklist 3. **`ai_act_obligation_tracker.py`** — input: organization role per Article 25 (provider / deployer / importer / distributor / authorized representative) → output: obligation matrix with deadlines Four references each citing 5+ authoritative sources (the Regulation itself + EDPB + European Commission guidelines + ENISA + IAPP tracker + national supervisory authority guidance). ## What this is NOT - **NOT executive AI strategy.** For board-level AI decisions (build-vs-buy, model selection, US/EU strategy), see `c-level-advisor/chief-ai-officer-advisor/`. - **NOT ISO 42001 compliance.** That's a voluntary management-system standard; this is a binding regulation. They complement each other. See `compliance-team-iso42001`. - **NOT GDPR compliance.** For personal-data processing (which AI systems frequently trigger), see `ra-qm-team/skills/gdpr-dsgvo-expert/`. The two regulations interact (Recital 10, Article 10). - **NOT a legal substitute.** This skill produces Article-cited compliance artifacts. For binding legal advice, especially on novel cases (e.g., is a chatbot a "general-purpose AI model"?), engage outside counsel. ## Critical scope reminders - **Extraterritorial reach** (Article 2): The Act applies to providers placing AI systems on the EU market regardless of where they are established. US/UK/Asian companies serving EU users are in scope. - **Definition of "AI system"** (Article 3(1) + Commission Guidelines Feb 2025): broader than ML; includes rule-based systems if they exhibit "varying levels of autonomy" and "adaptiveness." Most modern enterprise software does NOT qualify; foundation models, predictive models, and decision-support systems frequently do. - **GPAI separate track** (Articles 51–55): General-purpose AI models (e.g., large foundation models) have a parallel regime with stricter rules for "systemic risk" GPAI (10²⁵ FLOPs training threshold). Article 55 obligations include systemic-risk evaluations, adversarial testing, and incident reporting. ## Quick start ```bash # Classify an AI system per the Act python skills/eu-ai-act-specialist/scripts/ai_system_risk_classifier.py # Plan conformity assessment for a high-risk system python skills/eu-ai-act-specialist/scripts/conformity_assessment_planner.py # Track obligations per organizational role python skills/eu-ai-act-specialist/scripts/ai_act_obligation_tracker.py ``` All three tools run with embedded samples if no JSON path is provided. ## License MIT.
© Alireza Rezvani · MIT · alirezarezvani/claude-skills
