Scored across 11 tools
Every tool has a distinct resource+action target, and descriptions include explicit 'Use when' / 'Do not use when' clauses that name the sibling tool to redirect to (e.g., screen_ofac_entity vs. get_sovereign_policy_brief vs. synthesize_regulatory_docket). The one apparent overlap — issue_safe_harbor_certificate bundling OFAC screening, collateral valuation, and HUD-1 fees — is explicitly framed as a composite deliverable distinguished from the point-solution tools. No realistic misselection path remains.
All 11 names are snake_case imperative verb_noun constructions (get_cusip_valuation, screen_ofac_entity, run_portfolio_stress_test, provision_agent_key). Verb choice is varied but semantically precise, and the pattern never breaks into camelCase or noun-only forms.
11 tools sits squarely in the well-scoped 3-15 band, and each maps to a discrete deliverable: four analytics oracles, one compliance screen, one composite certificate, two policy/document generators, one billing pair, and one forensic model. No filler tools and no obvious missing surface within the stated scope.
The read-only analytics, screening, document synthesis, and payment/API-key lifecycle are covered end to end, and the 'does not transmit comments' caveat is honestly scoped rather than a dead end. Minor gaps: no equivalent of redeem/provision for querying usage or revoking a key, no way to list supported dockets or available CUSIP reference data, and no batch/portfolio-construction input beyond the stress-test array.
