Scored across 19 tools
The six find_* tools are cleanly separated by dimension (stale, anomalies, deteriorating, recovering, unreliable, schema drift), and the verify_* triple (published artifacts vs. live transport vs. signature) is well delineated. However, a family of five per-dataset readers—get_dataset, get_data_passport, get_provenance, get_evidence, trust_verdict—all return published artifacts for one dataset, creating real selection risk that only the elaborate cross-referencing in descriptions mitigates.
Most tools follow a clean verb_noun or verb_prep_noun pattern (get_dataset, search_datasets, find_stale, verify_evidence, find_by_licence). The two exceptions—trust_verdict and usage_summary—drop the verb prefix, a minor deviation from an otherwise predictable convention.
19 tools is on the heavy side and sits at the borderline the rubric flags, but the domain (catalogue freshness, provenance, drift, reliability, reconciliation, attestation, analytics) is genuinely broad and each tool maps to a distinct capability. It is slightly over-provisioned rather than padded.
The surface covers the full lifecycle: discovery (search_datasets, find_by_licence), single-dataset detail (get_dataset, get_data_passport), evidence and citation (get_provenance, get_evidence), trust and verification (verify_dataset, verify_evidence, verify_attestation, trust_verdict), risk enumeration (all find_* tools), reconciliation, catalogue summary, and usage analytics. No obvious gaps or dead ends remain.