claws-skill-qualitylisted

# Claws Skill Quality — authoring and review standard ## What this is Workflow wrapper around the SKILL.md quality regime. The canonical schema, rubric axes, and scoring weights live at `docs/skill-construction-standard.md`. This skill owns the workflow: when to run the linter, how to interpret scores, and what failure modes to watch for. ## When to invoke - Authoring a new `.claude/skills/*/SKILL.md` file. - Reviewing a PR that adds or modifies a SKILL.md. - Asked "lint my skill", "is this skill compliant", or "score this SKILL.md". - Pre-dispatch check before wiring a skill into a titan blueprint. ## When NOT to invoke - To run a skill at runtime — use the Skill tool directly. - To navigate the skill adjacency graph — use [[claws-skill-graph]]. - For chain-inference rules — use [[claws-skill-chaining]]. - For MCP tool selection — use [[claws-mcp-cognition-engine]]. ## Steps 1. **Load schema.** Read `docs/skill-construction-standard.md` for the 6 rubric axes and scoring weights. 2. **Run linter.** `node scripts/claws-skill-lint.js --skill <slug> --strict`. Exit 0 = clean, 1 = violations, 2 = fatal. 3. **Map to axes.** Match each finding to its rubric axis; see repair guidance in the schema doc. 4. **Repair.** Edit the SKILL.md. Prefer dense prose. For genuine leaf skills, add `leaf: true` to frontmatter rather than inventing backlinks. 5. **Re-run until clean.** Repeat steps 2–4 until exit 0. 6. **Wire into graph.** New skill? Add wikilinks from ≥2 related skills' `#