Scored across 19 tools
The find_* family is clearly distinct, but there is meaningful overlap among get_dataset, get_data_passport, get_provenance, get_evidence, verify_dataset, trust_verdict, and verify_attestation, all of which return related trust/evidence information. The descriptions use explicit 'use X not Y' guidance, yet an agent could still misselect, especially since get_dataset itself mentions provenance/citation metadata.
Names are uniformly lowercase snake_case and mostly follow a verb_noun pattern like search_datasets, find_stale, and verify_evidence. A few names such as usage_summary and trust_verdict are noun-phrase style, but the overall convention is predictable and readable.
With 19 tools, the server sits in the 16-25 range that feels heavy. Many tools are narrowly scoped variants of finding risk signals or verifying evidence, so the surface could plausibly be consolidated without losing much functionality.
For a read-only data-trust server, the surface is remarkably complete: discovery, metadata, freshness summaries, stale/anomaly/trend/reliability/drift detection, reconciliation, provenance, evidence receipts, live verification, attestation, trust verdicts, and licence filtering are all present. No obvious dead ends or missing lifecycle operations for the stated domain.