chief-of-agent

Decision instruments for turning submitted material into criteria a decider can check.

Quickstart

npx skills@latest add LilMGenius/chief-of-agent --global --agent '*'

Every skill here carries disable-model-invocation: true, so an agent does not pick one up on its own. Name the skill when you want it. A skill directory is self-contained, so copying one into the folder your agent reads works without its siblings.

Skill index

| Skill | Purpose | | --- | --- | | skills/decide/SKILL.md | Extracts checkable decision criteria from submitted material. | | skills/incorp/SKILL.md | Lands handed-over material at the tier an existing knowledge base already defines. |

Gates

Run from a checkout; the gates are repo tooling, not part of what an installed skill needs.

bash scripts/validate-skills.sh
bash scripts/check-links.sh

The first validates the skill catalog against the manifests; the second resolves every relative link in every tracked markdown file. Both name the file for each problem and exit non-zero when any problem was reported.

Authoring rules, the full list of what each gate enforces, and the commit convention live in AGENTS.md.