Conciseness

The body is efficient and assumes Claude's competence — no basic concept explanations (it says 'TAM/SAM/SOM' and 'burn multiple' without defining them), and the data-gathering and perspective-selection tables are dense and purposeful. It falls at anchor 4 ('efficient; minor instances... that could be trimmed') rather than 5 because of small duplications: the verdict scales are defined in Step 3 and then fully restated in Step 5, and the three reference files are listed in Step 3 and again in the closing Reference Files section. It is above anchor 3 because these are genuinely minor — no padded sections or unnecessary explanations.

Actionability

For an instruction-only skill the guidance is fully executable: a concrete table of exactly what data to gather, a table mapping the user's situation to which perspective to emphasize, per-lens core-area checklists, defined verdict scales (e.g., 'Strong Pass / Lean Pass / Lean Invest / Strong Invest'), and a precise report template ('titled "[Company Name] — Startup Analysis"' with named sections). Per the scoring note, absence of code is not penalized since the guidance is specific and actionable; anchor 5's 'specific examples cover the common cases' is met, including edge-case handling for insufficient public data.

Workflow Clarity

A clearly sequenced five-step workflow (Gather Information → Determine Perspectives → Analyze Each → Synthesize → Present) with explicit checkpoints: the Step 1 branch 'If web search does not return enough information to produce a meaningful analysis... ask the user to provide the company's website URL before proceeding' is a genuine feedback loop for the primary failure mode, and Step 5's 'Don't fabricate metrics; make informed inferences and state your confidence level' plus 'Scale each section to the evidence' act as quality gates. This matches anchor 5's 'clear sequence with explicit validation steps; feedback loops'; no destructive/batch-operation cap applies.

Progressive Disclosure

Scored against the actual bundle: all three referenced paths (`references/vc-framework.md`, `references/job-applicant-framework.md`, `references/ceo-framework.md`) exist as real files, each is one level deep, clearly signaled in its Step 3 subsection ('Read `references/vc-framework.md` for the full evaluation framework'), and re-indexed with one-line descriptions in a closing Reference Files section. SKILL.md stays a concise overview while the detailed criteria/benchmarks live in the references — anchor 5's 'clear overview with well-signaled one-level-deep references; content appropriately split'.

Specificity

The description lists concrete deliverables with comprehensive coverage: 'Analyze a startup from three perspectives — VC investor, job applicant, and CEO/founder — covering market, traction, unit economics, team, defensibility, runway, equity value, culture, and competitive position, with a verdict for each.' It enumerates nine specific evaluation areas plus three named perspectives and a defined output (a verdict for each). Anchor 5 ('Lists multiple specific concrete actions; comprehensive coverage') fits best; it exceeds anchor 4 because there are no minor gaps in coverage, and nothing is vague as anchor 3 would require.

Completeness

It explicitly answers both questions: what ('Analyze a startup from three perspectives... with a verdict for each') and when ('Use this skill whenever the user wants to evaluate a startup or private tech company: due diligence... or open-ended asks such as "what do you think of [company]"'). This mirrors anchor 5's exemplar structure of concrete capabilities followed by an explicit trigger clause with concrete phrases; anchor 4 would require the 'when' to be less explicit, which it is not.

Trigger Term Quality

Natural phrasings a user would actually say are comprehensively covered: 'due diligence', 'whether to invest in it or join it', 'weighing a startup job offer', 'company health', 'outlook', 'competitive position', plus open-ended variants '"what do you think of [company]" or "research [company] for me"'. This matches anchor 5's 'comprehensive coverage of natural terms including synonyms' — it includes both formal terms and casual user phrasings; anchor 4 would require noticeable missing natural terms, which is not the case.

Distinctiveness Conflict Risk

A clear niche (startup / private tech company evaluation) with distinct triggers anchored to that niche — 'evaluate a startup or private tech company', 'weighing a startup job offer', 'due diligence'. Even the broadest trigger phrases are explicitly scoped by the 'startup or private tech company' qualifier, matching anchor 5's 'clear niche with distinct triggers; minimal conflict risk'. It is above anchor 4 because overlap with generic research or document skills is minimal given the domain scoping.