Contract
- Input: problem or task defined by the skill body.
- Output: Markdown artifact or structured result with completion criteria met.
- Side effects: none (design/review/documentation only unless explicitly stated).
- Dependencies: none (self-contained unless linked to other skills).
- Stop condition: all process steps completed; artifact saved; criteria checked.
- Risk: low.
- Boundary: produces reasoning or documentation artifacts; does not modify external systems unless explicitly instructed.
Skill Template Generator
Contract
- Input: skill name, domain, description, and optional capability list.
- Output: a non-overwriting sandbox Skill, completed frontmatter, and validation command.
- Boundary: generate only under
.skill-sandbox/; promotion remains a separate decision.
Use node .agents/skills/platform/skill-lab.mjs template <name> --domain <domain> to create a safe starting point in .skill-sandbox/. Ask for the problem, users, capabilities, outputs, dependencies, and risk before generating; never overwrite an existing template.
Validate the result with node .agents/skills/platform/skill-lab.mjs validate .skill-sandbox/<name> --json, then promote only after the sandbox checks pass.