eu-ai-act-transparency-assessor-oliver-schmidt-prietz
Assesses which of the Art. 50(1)-(5) transparency obligations of the EU AI Act apply to a given AI system's provider or deployer, grounded in the final Code of Practice on Transparency of AI-Generated Content (June 2026) and the Commission's draft Art. 50 Guidelines. Covers AI-chatbot disclosure, deepfake and synthetic-content marking/watermarking, emotion-recognition and biometric-categorisation notices, the machine-readable marking duty, the obviousness exceptions, and the implementation timeline. Outputs a formal mini-report plus a per-obligation compliance checklist with gap flags. For bre
npx skills add lawve-ai/awesome-legal-skills --skill eu-ai-act-transparency-assessor-oliver-schmidt-prietz --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Identifies which Article 50 transparency duties (Regulation (EU) 2024/1689) apply to a system’s provider or deployer, decides what must be implemented and by when, and produces a formal mini-report plus a per-obligation compliance checklist.
How it works
Follows a six-phase workflow:
- Intake: determine the system’s mode and gather context; offers paths for Quick triage, Full assessment, or Implementation plan.
- Role determination: assigns duties to Provider or Deployer based on whether the system is built/marketed (provider), deployed/used (deployer), or both.
- Trigger determination: for each duty (50(1)–50(5)), applies trigger tests and obviousness/exceptions, then outputs a trigger-summary table.
- Implementation deep-dive: for each triggered duty, loads references and prescribes concrete actions across items like 50(2) marking and 50(4) labelling; cites relevant reference materials and implementation checklists.
- Dated roadmap: anchors tasks to timelines (e.g., 2 Aug 2026, 2 Dec 2026, etc.) with live-check guidance.
- Output: presents a Bottom line (6.0) and, in Full mode, the formal artifacts; references report templates for downstream artifacts.
When to use it
Use to determine which Art. 50 duties apply to a system and to produce a structured compliance plan, either as Quick triage (yes/no on triggered duties and deadlines) or as Full assessment with a formal mini-report and per-obligation checklist.
What it can touch
Tools: claude-code. The workflow references external materials (e.g., references/implementation-checklists.md, references/report-template-art50.md) and may invoke web checks per the workflow, but the actual active tool usage is controlled by the phase flow within the skill.
Caveats
The guidance is based on EU law and draft/soft-law materials (e.g., Code of Practice and Guidelines) and is not legal advice; final interpretation rests with competent authorities. Dates and grace periods (e.g., 2 Aug 2026 applicability and 2 Dec 2026 legacy marking) are subject to official publication and status assessments.
# EU AI Act — Article 50 Transparency Assessor Identify which **Article 50 transparency duties** (Regulation (EU) 2024/1689) apply to a system, decide **what must be implemented and by when**, and produce a formal mini-report plus a per-obligation compliance checklist. Works standalone, or ingests a prior classifier `ASSESSMENT CONTEXT` block. ## Disclaimer (show at session start, do not block) > **Important:** This skill provides structured Art. 50 transparency guidance based on the EU AI Act > (Regulation (EU) 2024/1689), the final **Code of Practice on Transparency of AI-Generated Content** > (10 Jun 2026), and the Commission's **draft** Art. 50 Guidelines (8 May 2026). It is **not legal > advice**; final decisions need qualified counsel, and only the **CJEU** can authoritatively interpret > Art. 50. > • **Penalty band:** non-compliance is **Tier 2 — up to EUR 15,000,000 or 3% of worldwide annual > turnover** (Art. 99(4)(g); €750k for EU bodies). *Not* the €35M / 7% band (that is Art. 5 prohibited practices). > • **Dates:** Art. 50 applies from **2 August 2026** (Chapter IV general application — *not* the 2 Aug > 2025 tranche). The 50(2) **legacy-system marking grace to 2 Decemb
- Disclaimer (show at session start, do not block)
- Start here: pick a mode (ask this first)
- Uncertainty markers (use these in every output)
- When to Search the Web (run quietly; report as one line)
- Workflow: Ask Questions ONE AT A TIME
- Phase 1: Intake
- Phase 2: Role determination
- Phase 3: Trigger determination (one sub-section per duty)
- Phase 4: Implementation deep-dive (per triggered duty)
- Phase 5: Dated roadmap
- Phase 6: Output (lead light, then the formal artifacts)
- Related skills
- Critical Reminders
What does the eu-ai-act-transparency-assessor-oliver-schmidt-prietz skill do?
Assesses which of the Art. 50(1)-(5) transparency obligations of the EU AI Act apply to a given AI system's provider or deployer, grounded in the final Code of Practice on Transparency of AI-Generated Content (June 2026) and the Commission's draft Art. 50 Guidelines. Covers AI-chatbot disclosure, deepfake and synthetic-content marking/watermarking, emotion-recognition and biometric-categorisation notices, the machine-readable marking duty, the obviousness exceptions, and the implementation timeline. Outputs a formal mini-report plus a per-obligation compliance checklist with gap flags. For bre
How do I install it?
Run `npx skills add lawve-ai/awesome-legal-skills --skill eu-ai-act-transparency-assessor-oliver-schmidt-prietz --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From lawve-ai/awesome-legal-skills, a repository with 618 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.