Civic

Many tools are clearly scoped (memory, subscriptions, CIViC, dependency checks), but there is substantial overlap among the Pipeworx meta-tools: ask_pipeworx, ask_pipeworx_beta, ask_pipeworx_grounded, deep_research, validate_claim, entity_profile, recent_changes, and compare_entities all answer data-lookup questions. The prediction-market suite also has heavily overlapping tools (polymarket_edges, polymarket_arbitrage, polymarket_fill_risk, polymarket_edge_tracker, bet_research). Descriptions do a lot to steer selection, so it is not chaotic, but boundaries remain fuzzy.

Names are almost entirely lower snake_case, with clear domain prefixes such as civic_, polymarket_, pipeworx_, and kalshi_. They are not uniformly verb_noun — several are noun phrases like entity_profile, company_facts, civic_assertions, recent_alerts — but the convention is still readable and consistent enough.

40 tools is very heavy for one MCP server and includes duplicate/experimental variants such as ask_pipeworx_beta plus many meta-tools. The set spans several unrelated domains (CIViC cancer variants, prediction markets, generic research, memory, subscriptions, AI visibility), so it feels like a platform gateway rather than a well-scoped server. This exceeds the typical 3–15 tool sweet spot by a wide margin.

The surface is broad and covers most lifecycle needs for its apparent domains: CIViC gene/variant/evidence/assertion lookup, company profiles and facts, prediction-market research and arbitrage, entity resolution, memory, subscriptions, and feedback. Minor gaps exist — e.g. no explicit CIViC disease/therapy enumeration tool and no write/update operations — but for a mostly read-only structured-data gateway, coverage is strong.