Conciseness

The body is dense with non-obvious operational data (rate limits like "NCBI APIs at 3 req/sec without key, 10 with key", per-database pagination parameter names, credential tables), but contains minor trims: "Many APIs return paginated results — if you only read the first page, you may miss data" restates what Claude already knows, and several identifier-table rows (e.g., "ENSG###########", "rs########") restate common knowledge. It is not 5 because of these small padded spots, and not 3 because the padding is minor against overwhelmingly task-specific content.

Actionability

Guidance is fully executable: a copy-paste curl example with --fail-with-body flags, concrete endpoint patterns ("PubChem /compound/name/{name}/cids/JSON", "Ensembl /xrefs/symbol/homo_sapiens/{symbol}"), exact thresholds ("10,000 records, 100 API calls", "at most 5 independent API requests in flight"), and a per-database error-recovery ladder. Specific examples cover the common cases (identifier resolution, rate limits, pagination), matching the top anchor.

Workflow Clarity

The 7-step Core Workflow is clearly sequenced with explicit validation checkpoints: "count first when the API supports it", "reconcile counts", and "fail visibly if the final dataset is incomplete", plus confirmation gates before large retrievals. A dedicated Error recovery section provides the feedback loop (check identifier format → try alternatives → try different database → report), and the Completeness and Reproducibility section is an explicit 6-step checklist — matching the anchor with validation steps, feedback loops, and checklists.

Progressive Disclosure

The hub-and-spoke structure is exemplary: per-database detail lives in one-level-deep reference files (all 82 referenced paths verified to exist, e.g., references/alphafold.md, references/database_selection_guide.md, references/retrieval-contract.md), navigation is well-signaled via the Database Selection Guide and the Available Databases tables mapping each database to its file, and only cross-cutting quick-reference material (identifier formats, credentials) is kept inline. This matches the 'clear overview with well-signaled one-level-deep references' anchor.

Specificity

"Queries documented public database APIs with explicit endpoints, filters, pagination, and provenance" names multiple concrete capabilities (endpoint querying, filtering, pagination, provenance) with no vague filler. This matches the anchor for comprehensive concrete-action coverage rather than the 'several specific actions, minor gaps' anchor below it, since every capability named is concrete and none is generic.

Completeness

The 'what' is explicit ("Queries documented public database APIs with explicit endpoints, filters, pagination, and provenance") and the 'when' is an explicit "Used when a scientific, regulatory, financial, or other database-backed fact must be retrieved reproducibly from a named source rather than inferred from general knowledge" clause with concrete trigger conditions. Both halves are clearly and explicitly answered, matching the top anchor.

Trigger Term Quality

Natural domain keywords are present ("scientific, regulatory, financial", "database", "API", "fact", "named source"), giving good trigger coverage. It is not 5 because common natural phrasings users would say (e.g., "look up", "data source", specific database names like FRED or UniProt) are absent; it is not 3 because the keywords present are relevant and go beyond a single generic term.

Distinctiveness Conflict Risk

"documented public database APIs" and "retrieved reproducibly from a named source rather than inferred from general knowledge" carve a clear niche that is easily distinguished from generic search or general-knowledge skills. Conflict risk is minimal, matching the 'clear niche with distinct triggers' anchor rather than the 'minor overlap risk' anchor below.