The body is lean and operational — concrete discovery paths, a frontmatter template, decision rules, and a checklist — with no padding explaining concepts Claude already knows, fitting the 'lean and efficient; every token earns its place' anchor rather than the score-2 'mostly efficient but includes some unnecessary explanation'.

Provides copy-paste-ready concrete guidance — a frontmatter template, a numbered creation workflow, an explicit skill-vs-MCP/hooks/tools/plugin decision tree, and specific validation checks — so for an instruction-only skill the guidance is fully actionable, matching the top anchor rather than the score-2 'incomplete; missing key details'.

The Creation Workflow is a clearly sequenced 1–7 process ending in an explicit validation step (step 7: 'Validate by loading the skill through /skills...') plus a dedicated Validation Checklist, satisfying the 'clear sequence with explicit validation steps; checklists' anchor; the destructive/batch cap does not apply to this non-destructive context.

No bundle files exist, so scoring is on structure alone; the ~100-line body is cleanly sectioned (What A Skill Is, Minimum Shape, Writing Rules, Creation Workflow, Updating, Validation Checklist) and genuinely self-contained with no need for external references, matching the well-organized-sections rationale for a self-contained skill rather than the score-2 'could be better organized'.

Lists multiple concrete actions — 'Create or improve codewhale skills', plus the create/update/meta-guidance distinction ('wants a new skill, wants to update an existing skill, or needs guidance on when a skill should be a skill versus MCP, hooks, tools, or a plugin scaffold'), matching the 'lists multiple specific concrete actions' anchor rather than the score-2 'names domain and some actions'.

Explicitly answers both what ('Create or improve codewhale skills') and when ('Use when the user wants a new skill, wants to update an existing skill, or needs guidance...'), and the present 'Use when...' clause lifts it above the score-2 cap where the trigger is only implied.

Covers natural terms the target user would actually say — 'new skill', 'update an existing skill', 'MCP, hooks, tools, or a plugin scaffold' — which are the audience's native vocabulary, fitting the 'good coverage of natural terms users would say' anchor rather than the score-2 'some relevant keywords but missing common variations'.

Occupies a clear niche — skill authoring and the skill-vs-MCP/hooks/tools/plugin decision — with distinct triggers unlikely to fire for unrelated skills, matching the 'clear niche with distinct triggers; unlikely to conflict' anchor rather than the score-2 'could still overlap with similar skills'.