Quick Start

Guide to get licit up and running in your project in 5 minutes.

1. Install

pip install licit-ai-cli

Requires Python 3.12+. If you have multiple versions: python3.12 -m pip install licit-ai-cli

Verify:

licit --version
# licit, version 0.5.0

2. Initialize

Navigate to your project directory and run:

cd mi-proyecto/
licit init

This will:

• Automatically detect languages, frameworks, CI/CD, AI agents, and security tools.
• Generate .licit.yaml with a tailored configuration.
• Create the .licit/ directory for internal data.

Initialized licit in mi-proyecto
  Languages: python
  Frameworks: fastapi
  Agent configs: CLAUDE.md
  CI/CD: github-actions
  Config saved to .licit.yaml

If you only need a specific regulatory framework:

licit init --framework eu-ai-act     # EU AI Act only
licit init --framework owasp         # OWASP Agentic Top 10 only

3. View Status

licit status

Shows a summary of:

• Detected project (name, languages, frameworks, git)
• Loaded configuration
• Enabled frameworks
• Available data sources
• Active connectors
• AI agent configurations found

4. Connect Data Sources (optional)

If you use Architect or Vigil:

licit connect architect    # Read reports and audit logs from Architect
licit connect vigil        # Read SARIF findings from Vigil

To disconnect:

licit connect architect --disable

5. Version the Configuration

git add .licit.yaml
git commit -m "chore: initialize licit compliance tracking"

Add sensitive data to .gitignore:

.licit/provenance.jsonl
.licit/fria-data.json
.licit/.signing-key

5. Track Code Provenance

licit trace                      # Analyze entire git history
licit trace --since 2026-01-01   # From a specific date
licit trace --stats              # Show statistics
licit trace --report             # Generate Markdown report

Example output:

  Analyzing git history...
  Records: 45 files analyzed
  AI-generated: 18 (40.0%)
  Human-written: 22 (48.9%)
  Mixed: 5 (11.1%)

  Stored in .licit/provenance.jsonl

The trace command analyzes each commit with 6 heuristics (author, message, volume, co-authors, file patterns, time) and classifies each file as ai, human, or mixed.

6. Generate Agent Config Changelog

licit changelog                        # Markdown by default
licit changelog --format json          # JSON output
licit changelog --since 2026-01-01     # From a specific date

Example output:

# Agent Config Changelog

> 2 change(s) detected across 1 file(s): **1** major, **1** minor

## CLAUDE.md

- **[MAJOR]** Changed: model from gpt-4 to gpt-5 (`abc1234`) — 2026-03-10
- **[MINOR]** Changed: section:Rules from 5 lines to 8 lines (+3/-0) (`def5678`) — 2026-03-09

  Changelog saved to .licit/changelog.md

The command analyzes the git history of agent configuration files, detects semantic changes (not just line diffs), and classifies them by severity.

7. Complete the Impact Assessment (FRIA)

licit fria                # Interactive 5-step questionnaire
licit fria --update       # Update existing FRIA

licit auto-detects answers where possible (models used, guardrails, testing, etc.) and asks for confirmation. It generates .licit/fria-data.json and .licit/fria-report.md.

8. Generate Annex IV Technical Documentation

licit annex-iv --organization "Mi Empresa" --product "Mi App"

Auto-generates technical documentation with 6 sections from project metadata. Includes recommendations for sections with missing evidence. Generates .licit/annex-iv.md.

9. Verify Compliance

licit verify --framework eu-ai-act    # Evaluate EU AI Act
licit verify                          # Evaluate all enabled frameworks

Exit code 0 = compliant, 1 = non-compliant, 2 = partial. Ideal for CI/CD gates.

10. Generate Compliance Reports

licit report                              # Markdown report (default)
licit report --format json -o report.json # JSON
licit report --format html -o report.html # Self-contained HTML

The report includes an overall summary, a per-framework table, and per-requirement detail with evidence and recommendations.

11. Identify Gaps

licit gaps                           # All gaps
licit gaps --framework eu-ai-act     # EU AI Act only

Each gap includes a description, actionable recommendation, suggested tools, and effort level.

Generated Structure

After licit init, your project will have:

mi-proyecto/
├── .licit.yaml          # Configuration (version this)
├── .licit/              # Internal data
│   ├── provenance.jsonl # Traceability (DO NOT version)
│   ├── changelog.md     # Config changelog
│   ├── fria-data.json   # FRIA data (DO NOT version)
│   ├── fria-report.md   # FRIA report
│   ├── annex-iv.md      # Annex IV documentation
│   └── reports/         # Generated reports
└── ... your code ...

Quick Command Reference

Global options: --version, --config PATH, --verbose, --help

For more details, see the full documentation.