Guide to get licit 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 0.3.0

2. Initialize

Navigate to your project directory and run:

cd my-project/
licit init

This:

• Automatically detects languages, frameworks, CI/CD, AI agents, and security tools.
• Generates .licit.yaml with the adapted configuration.
• Creates the .licit/ directory for internal data.

Initialized licit in my-project
  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

Displays 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

6. Track code provenance

licit trace                      # Analyze full 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.

7. 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.

What’s next?

Once future phases of licit are completed, you will be able to:

# Complete impact assessment (FRIA)
licit fria

# Generate Annex IV technical documentation
licit annex-iv --organization "My Company" --product "My App"

# View compliance report
licit report

# Identify gaps
licit gaps

# CI/CD gate (exit code 0 = pass)
licit verify

Generated structure

After licit init, your project will have:

my-project/
├── .licit.yaml          # Configuration (version control)
├── .licit/              # Internal data
│   ├── provenance.jsonl # Traceability (DO NOT version control)
│   ├── changelog.md     # Config changelog
│   ├── fria-data.json   # FRIA data (DO NOT version control)
│   ├── 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.