artic

The descriptions work hard to distinguish tools, but there is real overlap: ask_pipeworx, ask_pipeworx_beta, ask_pipeworx_grounded, deep_research, validate_claim, entity_profile, compare_entities and recent_changes all answer natural-language questions over largely the same underlying data. ask_pipeworx and ask_pipeworx_beta are explicitly stated to currently be identical, which is a direct misselection hazard. The Polymarket cluster (edges, arbitrage, edge_tracker, fill_risk, kalshi_spread, bet_research) is better delineated but still requires careful reading.

Everything is snake_case, which is good, but the verb/noun ordering and prefixing are mixed: some are verb-first (get_artwork, list_subscriptions, resolve_entity, validate_claim, scan_dependency) and others are noun-first (entity_profile, ai_visibility_check, bet_research, recent_alerts, recent_changes). Only some namespaces get a prefix (pipeworx_, polymarket_), so no single predictable pattern spans the set. Still readable, but inconsistent.

35 tools is heavy for any server, and several are near-duplicates (ask_pipeworx vs ask_pipeworx_beta, polymarket_arbitrage vs polymarket_edges) rather than distinct capabilities. The breadth spans unrelated domains (SEC data, art museum, AI marketing, memory, subscriptions), so no coherent scoping justifies the count. Many tools could be consolidated or folded into the router tools.

Lifecycle coverage is strong in the key areas: memory has remember/recall/forget, subscriptions have subscribe/list/unsubscribe/recent_alerts, Polymarket has research/scan/telemetry/fill-risk, and entity resolution is thorough. Minor gaps exist (no general update/edit tools, no direct raw-source query tools outside the router), but the major workflows are covered and agents rarely hit dead ends.