skill
Generate a repository-local skill from live repo discovery and user intent. Use when: "create QA skill", "generate repo skill", "make a local skill", "persona acceptance skill", "value proposition QA", "bespoke repo QA", "scaffold local skill". Creates concrete `.agents/skills/<name>/` guidance with harness-specific bridges when useful. Trigger: /create-repo-skill, /repo-skill, /create-qa-skill.
/create-repo-skill
Generate a repository-local skill from live repo discovery and user intent. Use when: "create QA skill", "generate repo skill", "make a local skill", "persona acceptance skill", "value proposition QA", "bespoke repo QA", "scaffold local skill". Creates concrete `.agents/skills/<name>/` guidance with harness-specific bridges when useful. Trigger: /create-repo-skill, /repo-skill, /create-qa-skill.
Support primitive in the Harness Kit operating loop.
This generated excerpt gives readers the beginning of the live primitive contract before they jump to GitHub.
Generate a local skill for this repo, not Harness Kit globally. This is a thin
skill authoring lane, not a persona runtime or scheduling engine.
## Route
| Need | Load |
|---|---|
| repo-specific QA skill | `references/qa.md` |
| persona/value-proposition acceptance skill | `references/persona-acceptance.md` |
| skill-writing rules | `references/authoring.md` |
| readiness profile present | Load `.harness-kit/agent-readiness.yaml` for gate commands, mock policy, stack feedback strength, and waivers |
## Contract
- Discover the live repo before drafting.
- Ask the user for missing product truth: value proposition, target users,
critical workflows, production/local target, and allowed side effects.
- Dispatch independent lanes: repo mapper, product/persona mapper, critic.
- Write only repo-local skill artifacts in the target repo.
- Prefer `.agents/skills/<name>/` as shared root; bridge `.claude/skills/`,
`.codex/skills/`, and `.pi/skills/` when those dirs exist.
- Do not shadow a first-party global skill unless the local repo explicitly
needs the same command name, such as `.agents/skills/qa/`.
- Include at least one eval seed or smoke oracle in the generated skill.
- Include an acceptance block with live repo evidence, exact command/path,
repo-fit check, and residual risk.
- Name the repo's observable surfaces when the generated skill verifies
behavior that should be watched after ship: healthchecks, logs, analytics
coverage, receipts, evidence directories, benchmarks, or release smoke.
- Keep generated `SKILL.md` under 300 lines unless the repo proves otherwise.
## Delegation Floor
Delegation floor applies: probe the roster first; dispatch two or more
providers for substantive work; direct solo only for mechanical, emergency,
user-forbidden, or fewer-than-two-providers cases. See
...