Conciseness

The body is lean and bullet-driven: it assumes competence (never explains what Shape Up, a breadboard, or a fit check is), keeps every section to decision-relevant guidance, and its only prose ("Spikes are not mini-projects. They exist to remove a tail of risk.") is load-bearing framing rather than padding. Every token earns its place, matching the top anchor; it is not a 4 because there is no over-explanation left to trim.

Actionability

For an instruction-only skill the guidance is highly concrete: a 7-step workflow, four named mitigation moves (patch/out-of-bounds/cut/spike), a five-field spike specification, three outcome classifications with follow-on decisions, and per-step pointers to specific templates. It falls short of fully executable (5) because step 4 "Execute the spike" and the wiring/integration/performance/state spike shapes remain necessarily abstract with no concrete worked steps or example embedded inline.

Workflow Clarity

Steps 1-7 are clearly numbered and sequenced, step 5 embeds an explicit decision branch (proceed/revise/drop) that acts as a checkpoint, and the Quality bar section is a checklist ("Timebox honoured", "Spike reduces risk *or* triggers a scope cut"). It sits below 5 because there is no explicit verify/re-validate loop — e.g. no step confirming the timebox or success criteria were actually met before classification — leaving minor validation gaps per the anchor.

Progressive Disclosure

SKILL.md is a clear overview-only workflow; all detail is delegated one level deep to real, existing files — templates/risk-register-template.md, spike-plan-template.md, spike-report-template.md (all verified present with substantive content), shapeup-risks-rabbit-holes-notes.md, and examples/sample-spike.md — each referenced inline at the step where it is needed and summarized in the Bundled resources section. Content is appropriately split with easy navigation, matching the top anchor exactly.

Specificity

The description lists several concrete actions — "running timeboxed spikes, documenting proof, and updating the shaped concept with patches, cuts, and out-of-bounds" — matching the anchor for several specific actions with minor gaps. It is not a 5 because phrases like "so the work stays thin-tailed within the appetite" lean abstract rather than concrete, and not a 3 because multiple distinct actions are named, not just 1-2.

Completeness

It explicitly answers both questions: "This skill should be used when shaping work and needing to de-risk rabbit holes (technical unknowns, design gaps, interdependencies)" is an explicit, concrete when-clause, and the what is stated as running timeboxed spikes, documenting proof, and updating the shaped concept with patches/cuts/out-of-bounds. This matches the top anchor with concrete trigger phrases; a 4 would require the when-clause to be less explicit or specific, which it is not.

Trigger Term Quality

Relevant keywords like "rabbit holes", "spikes", "de-risk", and "shaping" appear, but common variations a user might naturally say — "experiment", "prototype", "proof of concept", "validate technical assumptions", "technical risk" — are absent. This matches the anchor for some relevant keywords missing common variations or synonyms; it is above a 2 because the included terms are domain-natural, and below a 4 because coverage is not broad.

Distinctiveness Conflict Risk

The Shape Up framing (spikes, rabbit holes, appetite, shaped concept) carves out a distinct niche with domain-specific triggers, matching the anchor for mostly distinct with minor overlap risk. It is not a 5 because a user asking broadly to "investigate" or "research a technical question" could plausibly trigger it or a general research skill interchangeably.