30,000 staff. 21 Commissioners. Real procedures. Now on your laptop.
A startup lawyer needs a defensible GDPR DPIA for an AI hiring tool.
/eu-privacy:dpia "AI hiring tool" produces the full Art. 39 EUDPR assessment — DPO threshold, architecture, legal basis, security, EDPS determination — with citations the DPO can actually sign.
Your team must know if a draft regulation will die in inter-service consultation.
/eu-simulation:inter-service-consultation "Algorithmic pricing regulation" runs every affected DG and returns their objections with the exact treaty or regulatory basis.
You are drafting and want to know which Commissioners will fight it.
/eu-simulation:mandate-conflict "AI liability regulation" maps every structural conflict across all 21 portfolios, with the legal basis on each side and severity.
You need to know whether your proposal even passes subsidiarity.
/eu-simulation:subsidiarity-stress "Harmonised SME insolvency rules" tests the necessity argument against five different member-state configurations.
This is not generic prompting. It is the European Commission's institutional machinery — mandates, procedures, and adversarial checks — made executable.
The EU Commission is not primarily a regulator. It is a structured argumentation engine: 21 mandate-scoped agents forced to reconcile conflicting treaty obligations through explicit procedures (subsidiarity, impact assessment, inter-service consultation, College deliberation, trilogue).
That machinery is reusable. The same scaffold that produces defensible policy for 450 million people works for any high-stakes decision where multiple legitimate perspectives must survive adversarial scrutiny.
The most powerful commands in this system are the ones that are impossible for a single model:
| Command | What only multi-agent + real mandates can do |
|---|---|
/eu-simulation:mandate-conflict |
Maps every structurally guaranteed clash across 21 portfolios with treaty bases and severity |
/eu-simulation:red-team-college |
Runs the proposal through all 21; returns only SEVERE objections + adoptability verdict |
/eu-simulation:subsidiarity-stress |
Tests the necessity argument against 5 different member-state configurations |
/eu-simulation:timeline |
Produces a realistic OLP timeline with QMV thresholds and trilogue risk points |
See NOI.md for the full argument.
# 1. Add the marketplace (once)
/plugin marketplace add montoyer/eu-agents
# 2. Install what you need
/plugin install eu-simulation@eu-agents
/plugin install eu-legislative@eu-agents
/plugin install eu-privacy@eu-agents
# Or install everything:
/plugin install --all@eu-agents# Run a Commissioner
/eu-simulation:commissioner competition
# Full College deliberation
/eu-simulation:college-deliberation "Should the EU ban algorithmic pricing in retail?"
# Draft a regulation
/eu-legislative:legislative-proposal "Regulation on synthetic biology, legal basis Art. 114 TFEU"
# Compound stress tests
/eu-simulation:mandate-conflict "Regulation on AI liability in critical infrastructure"
/eu-simulation:red-team-college "Carbon border adjustment for agriculture"
/eu-simulation:subsidiarity-stress "Harmonised insolvency rules for SMEs"Every skill output ends with:
DRAFT — For review by an EU official before use. Not an official Commission position.
| Plugin | Purpose |
|---|---|
eu-simulation |
21 Commissioners + College + ISC + trilogue + OLP + compound analysis commands |
eu-legislative |
Impact assessment, legislative drafting, treaty check, consultation, REFIT, policy cycle |
eu-privacy |
Full Art. 39 EUDPR DPIA workflow (DPO → IT-PM → Legal → Security → EDPS) |
eu-competition |
Antitrust (Arts. 101-102), state aid (107-109), mergers, dawn raids, Legal Service |
eu-trade |
Trade defence: anti-dumping, anti-subsidy, safeguards |
eu-grants-enforcement |
Grant management, infringement procedures (Arts. 258-260), public procurement |
eu-institutional-management |
HR, unit management, financial circuits, CDR/AAR/AMP, access to documents |
eu-data-communication |
Eurostat, scoreboards, press releases, lines to take, transparency |
eu-careers |
EPSO grade/step estimation, presentation coaching, offer analysis |
See CLAUDE.md for the complete command reference.
- Single-skill:
/eu-legislative:impact-assessment "brief here" - Commissioner persona:
/eu-simulation:commissioner digital - Multi-agent sessions:
/eu-simulation:inter-service-consultation "proposal" - Compound (unique value):
/eu-simulation:mandate-conflict,red-team-college,subsidiarity-stress,timeline
Most plugins also include a /eu-xxx:cold-start-interview to tailor behaviour to your DG or dossier.
- One agent per mandate. The Competition Commissioner never speaks for Agriculture.
- Real procedures. Workflows follow actual EU rules (Better Regulation, Joint Practical Guide, subsidiarity protocol, etc.).
- Adversarial by design. Commissioners disagree. The system surfaces conflict rather than smoothing it over.
- Grounded output. Every claim that needs verification carries an attribution tag.
- EU policy officers and legislative drafters
- Lawyers doing EU compliance, state aid, or antitrust work
- Consultants and lobbyists who need the institutional position fast
- Researchers modelling policy processes
- Journalists who want the legal basis and cross-portfolio tensions
- Students learning the ordinary legislative procedure interactively
- Anyone who needs structured, multi-perspective reasoning under constraint
- QUICKSTART.md — 5-minute onboarding
- CLAUDE.md — Complete command reference and agent rules
- docs/getting-started.md
- docs/examples/ — Full worked simulations
- NOI.md — Why the EU's reasoning machinery is worth programming
- ARCHITECTURE.md — How the layers fit together
Production scaffolding for serious use. Core Commissioners, DGs, legislative and privacy workflows are stable. Compound simulation commands and additional domains are under active development.
European Union Public Licence v. 1.2 (EUPL-1.2)
See LICENSE and joinup.ec.europa.eu/collection/eupl.
See CONTRIBUTING.md. New skills, Commissioners, DGs, and connectors are welcome.