skillforge Skill
When To Use
Use this skill when an agent needs to create, lint, test, render, package, or review a portable coding-agent skill bundle. It is especially useful for release gates around skill.yaml, SKILL.md, activation fixtures, host rendering, and safety notes.
Inputs
- A local skill directory with
skill.yamlandSKILL.md. - Optional activation fixture file.
- Optional host target such as
openclaworclaude-plugin. - Optional output directory or package archive path.
Side-Effect Boundaries
The lint, test, matrix, and report commands read local files only. Render and package commands write only to the requested output path. The CLI does not publish packages, install skills into an agent host, call network APIs, or approve skill proposals.
Workflow
- Run
skillforge lint <skill-dir> --format jsonfor a small machine-readable quality gate. - Run
skillforge test <skill-dir> --fixtures <skill-dir>/fixtures/activation.jsonto check activation examples and anti-examples. - Run
skillforge matrix <skill-dir> --format markdownto inspect host compatibility. - Run
skillforge report <skill-dir> --format jsonbefore release-candidate review. - Render or package only after diagnostics are clean or intentionally accepted.
Approval Requirements
Ask before writing rendered host layouts into a user-controlled skill directory. Ask before packaging a bundle intended for external sharing. Do not publish, install, or approve skills unless the user explicitly asks for that separate action.
Examples
skillforge lint examples/tdd-sentinel --format json
skillforge test examples/tdd-sentinel --fixtures examples/tdd-sentinel/fixtures/activation.json
skillforge report examples/tdd-sentinel --format markdown
skillforge render examples/tdd-sentinel --target openclaw --out /tmp/tdd-sentinel-openclaw
Verification
Run:
npm run check
npm test
npm run smoke