eu-ai-act

EU AI Act Compliance Toolkit CLI — Classify AI systems, track compliance checklists, and generate documentation required under the EU AI Act.

Disclaimer: This tool does not constitute legal advice. Consult qualified legal counsel for compliance decisions.

Installation

npm i -g eu-ai-act

Or use without installing:

npx eu-ai-act --help

Quick Start

# 1. Classify your AI system
eu-ai-act classify

# 2. View your compliance obligations
eu-ai-act checklist high-risk

# 3. Check your progress
eu-ai-act status

# 4. Generate documentation
eu-ai-act generate conformity --system "MyAI" --provider "Acme Corp" --purpose "Customer support chatbot"

# 5. Export a compliance report
eu-ai-act report --output report.md

Commands

eu-ai-act classify

Classify your AI system's risk tier under the EU AI Act. Runs an interactive wizard by default.

Interactive mode:

eu-ai-act classify

Flag mode (for CI/scripting):

eu-ai-act classify --gpai --flops 1e25
eu-ai-act classify --social-scoring
eu-ai-act classify --annex-iii biometrics --json

Flags: | Flag | Description | |------|-------------| | --social-scoring | Article 5(1)(c) — social scoring | | --realtime-biometrics | Article 5(1)(h) — real-time biometric ID | | --subliminal | Article 5(1)(a) — subliminal techniques | | --exploits-vulnerabilities | Article 5(1)(b) — exploits vulnerabilities | | --facial-scraping | Article 5(1)(e) — untargeted facial scraping | | --emotion-workplace | Article 5(1)(f) — emotion in workplace | | --biometric-categorization | Article 5(1)(g) — biometric categorization | | --predictive-policing | Article 5(1)(d) — predictive policing | | --gpai | General-purpose AI model | | --flops <number> | Training compute in FLOPs | | --open-source | Open source GPAI model | | --systemic-risk | Designated systemic risk | | --annex-i | Safety component (Annex I) | | --annex-i-third-party | Requires third-party assessment | | --annex-iii <category> | Annex III high-risk category | | --interacts-persons | Interacts with natural persons | | --synthetic-content | Generates synthetic content | | --emotion-recognition | Performs emotion recognition | | --biometric-categorizing | Biometric categorization | | --json | Output as JSON |

Example output:

  Classification Result
  ────────────────────────────────────────

  🤖 Risk Tier:    GPAI
  📅 Enforcement: 2025-08-02
  📋 Conformity:  none

  Applicable Articles
  51, 52, 53

  Reasoning
  1. ✗ Not prohibited — no prohibited practices detected
  2. ✓ General-purpose AI model detected
  3. → Classified as GPAI (standard)

eu-ai-act checklist <tier>

Display the compliance checklist for a risk tier.

eu-ai-act checklist high-risk
eu-ai-act checklist gpai --filter incomplete
eu-ai-act checklist limited --json
eu-ai-act checklist high-risk --filter 9        # Filter by Article 9

Options: | Flag | Description | |------|-------------| | --filter <value> | Filter: complete, incomplete, or article number | | --json | Output as JSON |

Valid tiers: prohibited, high-risk, gpai, gpai-systemic, limited, minimal

eu-ai-act timeline

Display the EU AI Act enforcement timeline with key compliance dates.

eu-ai-act timeline
eu-ai-act timeline --json

Example output:

  📅 EU AI Act Enforcement Timeline
  ────────────────────────────────────────

  ┌──────────────┬──────────────────────────────────┬────────────┬──────────────┐
  │ Date         │ Milestone                        │ Status     │ Countdown    │
  ├──────────────┼──────────────────────────────────┼────────────┼──────────────┤
  │ 2024-08-01   │ EU AI Act enters into force      │ ✓ Done     │ 606d ago     │
  │ 2025-02-02   │ Prohibited practices apply       │ ✓ Done     │ 421d ago     │
  │ 2025-08-02   │ GPAI rules apply                 │ ✓ Done     │ 240d ago     │
  │ 2026-08-02   │ High-risk requirements apply     │ ○ Future   │ 125d left    │
  └──────────────┴──────────────────────────────────┴────────────┴──────────────┘

eu-ai-act status

Show compliance summary for your classified AI system. Requires running classify first.

eu-ai-act status
eu-ai-act status --json

Example output:

  📊 Compliance Status
  ────────────────────────────────────────

  System:     AI System (Provider)
  Tier:       GPAI
  Classified: 2026-03-30

  Compliance: ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0/9)

  Next deadline: High-risk requirements apply
  2026-08-02 (125 days remaining)

eu-ai-act generate [template]

Generate pre-filled compliance documentation from built-in templates.

# Single template
eu-ai-act generate conformity --system "MyAI" --provider "Acme" --purpose "Customer support"

# Save to file
eu-ai-act generate risk-assessment --system "Scorer" --provider "Corp" --purpose "Credit scoring" --output risk.md

# Generate ALL templates for a tier
eu-ai-act generate --tier high-risk --system "MyAI" --provider "Co" --purpose "Hiring" --output ./docs/

Available templates: conformity, risk-assessment, model-card, impact-assessment, transparency-notice

Required options: | Flag | Description | |------|-------------| | --system <name> | AI system name | | --provider <name> | Provider/organization name | | --purpose <text> | Intended purpose of the system | | --output <path> | Output file path (default: stdout) | | --tier <tier> | Generate ALL templates for a tier |

eu-ai-act report

Export a full compliance report combining classification, checklist progress, and deadlines.

eu-ai-act report                        # Print Markdown to stdout
eu-ai-act report --output report.md     # Save to file
eu-ai-act report --json                 # JSON output
eu-ai-act report --format json -o r.json

Options: | Flag | Description | |------|-------------| | --format <type> | Output format: md (default) or json | | --output <path> | Write to file instead of stdout | | --json | Shorthand for --format json |

Configuration

.eu-ai-act.json

The CLI stores classification state in a .eu-ai-act.json file in the current directory. This file is created automatically when you run eu-ai-act classify and is used by status, checklist, and report commands.

Structure:

{
  "version": "1.0.0",
  "system": {
    "name": "AI System",
    "provider": "Provider",
    "classifiedAt": "2026-03-30T17:00:00.000Z"
  },
  "classification": {
    "tier": "high-risk",
    "subTier": null,
    "conformityAssessment": "third-party"
  },
  "checklist": {}
}

The CLI walks up the directory tree to find this file, so you can run commands from subdirectories.

Environment Variables

| Variable | Description | |----------|-------------| | NO_COLOR | Set to any value to disable colored output (respects no-color.org) | | FORCE_COLOR | Set to 1 to force colored output even when not a TTY |

Global Options

| Flag | Description | |------|-------------| | -v, --version | Display the current version | | -h, --help | Display help for a command |

Error Handling

The CLI provides clear, actionable error messages:

  Error: Invalid tier "invalid-tier"

  Valid tiers: prohibited, high-risk, gpai, gpai-systemic, limited, minimal

  Run eu-ai-act classify to determine your tier.

  Error: No .eu-ai-act.json found in current directory tree.

  Run eu-ai-act classify first to classify your AI system.

JSON Output

Every command supports --json for machine-readable output, making it easy to integrate with CI/CD pipelines, dashboards, or other tools:

eu-ai-act classify --gpai --json | jq '.tier'
eu-ai-act status --json | jq '.progress.percent'
eu-ai-act checklist high-risk --json | jq '.items | length'

License

MIT