crimes documentation

crimes scans TypeScript, JavaScript and Python repositories for change risk and agent risk. Use its findings to plan edits, identify related code and choose tests.

Pick a starting point:

• Command and detector reference: generated from the build.
• Pin migration: preserve reviewed decisions across identity changes.
• Agent setup and updates: install and refresh the bundled skills.
• JSON report types: generated from the public declarations.
• Triage: record and revisit risk decisions.
• Agent usage: how an AI coding agent should invoke crimes.
• Configuration: crimes.config.json reference.
• Scoring: what each per-finding score means and how agent_risk is computed.
• CI integration: --fail-on, baseline, and the diff workflow.

Browse detector families under Finding types in the sidebar.

The raw markdown sources live under docs/ so agents that read GitHub directly get the same content.