Contributing
Praetor is a local-first AI company operating system. Useful contributions keep
the product understandable, inspectable, and safe for a solo founder running it
on their own machine.
Good contribution areas:
- role templates and company templates
- mission templates and demo workspaces
- executor adapters and dry-run/demo runtime improvements
- filesystem-first artifact handling
- approval, decision, and memory lifecycle fixes
- documentation, quickstarts, and troubleshooting
Development
pixi install
pixi run check
Run targeted smoke tests for the area you touch:
pixi run app-smoke
pixi run planner-smoke
pixi run organization-smoke
pixi run safety-policy-smoke
pixi run app-security-smoke
pixi run app-auth-smoke
pixi run app-api-smoke
pixi run app-fallback-smoke
pixi run stack-smoke
pixi run bridge-smoke
pixi run dry-run-provider-smoke
pixi run demo-workspace-smoke
Do not commit local runtime state, workspaces, backups, secrets, API keys, model logs, or generated caches.
Product Rules
- Keep workspace artifacts visible in user-readable files.
- Keep CEO-facing behavior framework-driven, not hard-coded template chatter.
- Keep approvals sparse, but mandatory for external actions, spending,
destructive changes, credentials, security/legal/privacy boundaries, and
governance changes.
- Keep raw agent discussion as evidence; promote durable facts through
decisions, memory updates, playbook entries, or reports.
Pull Requests
Pull requests should include:
- a concise summary of user-visible behavior
- verification commands run locally
- any security, privacy, or deployment risk
Changes touching executor behavior, file permissions, authentication, CSRF, deployment secrets, or model-provider data flow should be treated as security-sensitive.