Skill effectiveness rubric

Structural validation (lib/validators/skills.mjs) is the floor. Effectiveness validation (lib/validators/skill-effectiveness.mjs) checks that skills are usable in production, not just well-formed.

Enforcement

Gate	Command
Scope + structure + effectiveness + composition	npm run lint:scopes
Full corpus	tests/skills-effectiveness.test.mjs
Composition graph	construct audit skills --composition
Inventory blocking	tests/certification/skills/inventory.json must have zero blockingFindings

Category expectations

Role overlays (skills/roles/**)

• ≥3 numbered failure modes with Symptom and Counter-move (or Counter)
• Base roles include Self-check, Methodology, or Ship Check
• scopes: frontmatter lists applicable work scopes
• applies_to matches ADR-0047 specialist bindings

Workflow skills (skills/docs/*-workflow.md)

• verificationBar ≥40 characters, testable against artifact release gate
• artifactType set (may differ from manifest type when orchestrating)
• ≥4 numbered steps; mentions construct artifact validate
• Specialist ids exist in registry; get_skill("…") paths resolve

Domain skills

• Body ≥400 characters after frontmatter
• Boundary in description or body (Use when, When not to use, etc.)

Quality gates and brand

• Executable checklists mapped to reviewers
• brand/output-vibe frontmatter name: brand-output-vibe; cited by customer-facing workflows

Composition

lib/skills/composition-graph.mjs audits routing.json targets, registry entitlements, and workflow get_skill chains. Unreachable non-role skills are blocking.

See also: rules/common/skill-composition.md.