Conciseness

The body is lean and assumes domain competence — no explanation of what a clinical trial or ICH E6 is, and the anti-fabrication rule is stated once as a terse callout ('LOOK UP, DON'T GUESS'). Minor redundancy keeps it below 5: the drug/device tool names appear in both the Step 2 table and the Step 3 core-tools list, and the relationship to `tooluniverse-clinical-trial-design` is explained three times (intro paragraph, NOT-for section, Integration section). Not 3: there is no genuine padding or teaching of known concepts.

Actionability

Mostly executable: copy-paste-ready commands for all three calculator modes plus `--self-test`, and exact ToolUniverse tool names per section (e.g. `get_clinical_trial_eligibility_criteria`, `FDA_OrangeBook_get_exclusivity`). One real gap: Step 5 directs 'Fill `templates/protocol_template.md`', but that file does not exist in the bundle, so the final assembly step has no executable target. Not 3: the bulk of the guidance is concrete and runnable, not pseudocode or high-level hints.

Workflow Clarity

A clear 5-step sequence with explicit branch points (Step 1 if/else on feasibility inputs, Step 2 drug/device fork), explicit validation (`--self-test` verifies all three formulas), and a genuine feedback loop ('if you don't have a defensible effect-size assumption yet, go back to Step 3'). Not 5: the final step dead-ends at the missing template file and there is no completeness check on the assembled document. Not 3: checkpoints are explicit and recoverable, not implicit.

Progressive Disclosure

Structure and signaling are good — a Reference Files table with per-file descriptions, one-level-deep references, and section-by-section detail correctly delegated to `references/protocol_sections_guide.md`. But scored against the actual bundle, one of the four referenced files, `templates/protocol_template.md`, does not exist, and it is the file the entire Step 5 assembly depends on. A dangling reference that breaks navigation at the deliverable-assembly step is more than the 'minor organization gaps' of the 4 anchor.

Specificity

The description lists multiple concrete actions with comprehensive coverage: 'Generate a full clinical trial protocol document (ICH E6(R2)/FDA-style)... including background, objectives/endpoints, study design, population/eligibility, intervention details, assessment schedule, statistical analysis plan with sample-size justification, safety monitoring, and regulatory/ethical sections', and states the deliverable format ('stakeholder-ready Markdown deliverable'). It uses third person ('Generate', 'Produces') with no first/second person. Not 4: the action coverage is comprehensive across every protocol section rather than having minor gaps.

Completeness

Both questions are answered explicitly: 'what' is the enumerated protocol-section deliverable, and 'when' is the concrete 'Use when users say...' clause with six trigger phrases. It even adds an explicit non-goal ('not a feasibility score') clarifying the output type. This matches the 5 anchor exactly ('clearly and explicitly answers both what AND when with concrete trigger phrases').

Trigger Term Quality

Six natural user phrasings are quoted verbatim: 'generate a clinical trial protocol', 'draft a protocol for [drug/device]', 'write up the study design as a protocol document', 'create an FDA submission protocol', 'help me write section X of a trial protocol', 'calculate the sample size for this trial'. These cover full-document requests, partial-section requests, regulatory-submission framing, and the sample-size subtask with synonym variations. Not 4: no common natural phrasing is obviously missing.

Distinctiveness Conflict Risk

It carves a clear niche (full protocol document generation) and differentiates with 'not a feasibility score', but the 'calculate the sample size for this trial' trigger overlaps with the sibling feasibility skill's effect-size/sample-size territory, and the description itself does not name the sibling skills to disambiguate (the body does). 'Mostly distinct; minor overlap risk with closely related skills' fits better than the 5 anchor, whose triggers have essentially no cross-skill ambiguity.