What problem does it solve?
Projects need a reliable, reproducible runtime and policy scaffold before AI agents can execute governed work. Manual bootstrapping of runtime state, health checks, and agent rules is error-prone and interrupts onboarding or first-run workflows. This Skill automates those first-time tasks so teams can start missions consistently.
Core Features & Use Cases
- Runtime bootstrap: Create the .geas/ directory structure including state, memory/agents, and recovery subfolders.
- Initial state & health: Generate .geas/state/run.json (versioned run state) and .geas/state/health-check.json with all health signals computed.
- Project rules and conventions: Detect project stack and commands, generate or update .geas/rules.md with a Code section, and optionally append .geas/ to .gitignore.
- Safe idempotence & validation: Commands are safe no-ops when artifacts already exist and the CLI enforces schema checks.
- Use case: Onboarding a new repository or preparing an existing repo for Geas-driven missions so automated agents can operate under a locked contract and evidence pipeline.
Quick Start
Run the setup skill to initialize the .geas runtime, generate rules.md from detected project conventions, and optionally add .geas/ to .gitignore.