Conciseness

The body is mostly efficient — dense tool parameter/return documentation Claude cannot know — but includes unnecessary or duplicated material: exclusivity codes (NCE/ODE/PED) are explained in both Phase 4 and the Reasoning Framework, the FAERS medicinalproduct param is noted twice, and the Reasoning Framework re-explains concepts Claude already knows (what an ANDA is, what Cmax/AUC/half-life are). This fits anchor 3 (mostly efficient, some unnecessary explanation or tightening possible) better than anchor 4.

Actionability

Highly actionable: every tool has exact parameter names, types, defaults, and return shapes, plus concrete calls and gotchas (UPPERCASE brand names, ttys="IN", overall_status as array). It falls short of anchor 5 because the Python snippets use an undefined 'tu' binding with no import/setup shown anywhere, so they are not fully copy-paste ready.

Workflow Clarity

The 8-phase workflow is clearly sequenced with an overview diagram, explicit cross-phase dependencies (search for UNII before get_substance, find class_id before get_class_members), and a Common Mistakes checklist. Verification steps exist but are implicit (e.g., 'check len(studies) > 0', read generics_available) rather than explicit validate-then-proceed checkpoints, matching anchor 4 rather than 5. The destructive/batch cap does not apply since all operations are read-only lookups.

Progressive Disclosure

No bundle files exist (no references/, scripts/, or assets/), and ~150 lines of per-tool API reference (params, return schemas, code aliases) are inlined in a single 280-line SKILL.md. Section headers and per-phase organization provide structure, but reference-grade material that belongs in a separate file is inline — matching anchor 3 rather than the well-split anchor 4.

Specificity

Lists several specific capabilities — 'FDA substance registry, ATC/EPC classification, EMA decisions, generic-drug status, FDA Orange Book exclusivity, NDA/BLA pathways' — but coverage has minor gaps: EMA decisions are claimed with no backing tooling in the body, and label parsing, clinical trials, and pharmacovigilance (three of eight body phases) are unmentioned. This matches anchor 4 (several specific actions, minor gaps) rather than anchor 5 (comprehensive coverage).

Completeness

Clearly answers both 'what' (substance registry, classification, EMA decisions, generic status, exclusivity, NDA/BLA pathways) and 'when' via the explicit clause 'Use for jurisdiction-aware approval status (FDA vs EMA), generic vs brand availability, exclusivity expiry tracking, and regulatory pathway selection.' This matches anchor 5: explicit what AND when with concrete trigger phrases.

Trigger Term Quality

Good natural-term coverage: 'FDA', 'generic', 'Orange Book', 'ATC', 'NDA', 'approval', 'exclusivity' are terms users would naturally say. A few common terms are missing from the description — notably 'patent' (e.g., 'when does the patent expire'), 'drug label', and 'adverse reactions' — so it falls short of the comprehensive synonym coverage of anchor 5.

Distinctiveness Conflict Risk

The regulatory/approval niche (Orange Book, exclusivity, NDA/BLA, UNII) is mostly distinct with minimal conflict risk, but terms like 'ATC code' and 'drug class' overlap with sibling drug skills (e.g., mechanism-research), fitting anchor 4 rather than the fully distinct anchor 5.