Conciseness

Mostly efficient (commands, flags, config paths, canon citations all earn their place), but the "estimate + named human owner" caveat is repeated in the intro ("Every output is an estimate with stated assumptions routed to a named human owner"), the anti-patterns ("Presenting a power estimate as fact. Every output is an estimate with a named owner who must sign."), and forcing-question 5; the surrogate-endpoint warning appears in four separate sections ("Penalizes unvalidated surrogate endpoints", "unvalidated serum-cytokine surrogate (cannot be primary)", "An unvalidated surrogate cannot anchor a PRIMARY endpoint", "Anchoring a primary on an unvalidated surrogate"); and the "Distinct from" table substantially duplicates "Do NOT use this skill to". This fits the anchor 'mostly efficient but includes some unnecessary explanation or could be tightened' — more than the 'minor instances' of the 4 anchor.

Actionability

Copy-paste-ready commands with real flag values throughout: "python3 scripts/sample_size_estimator.py --design proportions --p1 0.30 --p2 0.45 --dropout 0.15", "phase_gate_scorer.py --input study.json --profile <same> --phase {1|2|3|4}", onboarding ("python3 scripts/onboard.py --show"), and a full "/ar:setup --domain custom --name trial-feasibility ..." invocation. The "--sample" quick examples cover the common cases, matching the anchor 'fully executable; copy-paste ready code or commands; specific examples cover the common cases'.

Workflow Clarity

A clear 5-step numbered workflow (draft synopsis → select endpoint → estimate sample size → score feasibility → route for sign-off), each step with its exact command, plus conditional handling ("If >1 primary, plan multiplicity control") and a human sign-off checkpoint in step 5. Not 5 because there is no explicit error-recovery loop (e.g., what to revise and re-score when the verdict is REDESIGN/NO-GO is left implicit); not 3 because sequence and most checkpoints are present. These are analysis tools, not destructive/batch operations, so the workflow cap does not apply.

Progressive Disclosure

Good structure: a References section clearly signals three one-level-deep files with content summaries ("references/study_design_canon.md — ICH E8(R1)... ICH E9 + E9(R1) estimand addendum; CONSORT 2010..."), and a Scripts table maps each tool to its purpose. However, no bundle directories (references/, scripts/, assets/) exist alongside SKILL.md in this review, so the referenced paths (e.g., "assets/protocol_synopsis_template.md") cannot be verified as real, and the ~25-line forcing-question library is inlined detail that could live in a reference file. Fits the anchor 'good structure; most content is appropriately placed; references mostly clear; minor organization gaps' rather than the fully verified, cleanly split structure of the 5 anchor.

Specificity

Three concrete, parameterized actions are listed: "selecting and classifying endpoints (primary / key-secondary / exploratory, with surrogate-endpoint flagging)", "estimating sample size and power for two-arm designs (means / proportions / survival)", and "scoring a study plan for feasibility and a GO / GO-WITH-CONDITIONS / REDESIGN / NO-GO phase-gate decision". This matches the anchor 'lists multiple specific concrete actions; comprehensive coverage' rather than the 4 anchor, since coverage spans all three tools with concrete classification schemes and output types, leaving only minor gaps.

Completeness

It explicitly answers both: what (endpoint selection/classification, sample-size and power estimation for means/proportions/survival, feasibility scoring with named verdicts) and when ("Use when designing a prospective clinical study before submission — ... or scoring a study plan for feasibility..."). Matches the anchor 'clearly and explicitly answers both what AND when with concrete trigger phrases'; the when-clause is explicit and enumerated, so it exceeds the 4 anchor ('when' could be more explicit).

Trigger Term Quality

Natural phrases a clinical-research user would say are present: "designing a prospective clinical study", "sample size and power", "primary", "feasibility", "phase-gate decision", "surrogate-endpoint". Not 5 because a few common variations are missing ("clinical trial", "protocol", "power analysis"); not 3 because the included keywords are natural domain phrasing, not jargon-only.

Distinctiveness Conflict Risk

The niche (prospective clinical study design pre-submission) is clear and explicitly disambiguated from the nearest neighbor: "Distinct from ra-qm-team, which handles the regulatory/QM submission (ISO 13485, EU MDR, FDA 510(k)/PMA/QSR), not the study design". Matches the anchor 'clear niche with distinct triggers; minimal conflict risk' — the 4 anchor's 'minor overlap risk with closely related skills' is not triggered.