Naag

Several clusters have hazy boundaries: ask_pipeworx, ask_pipeworx_beta, and ask_pipeworx_grounded are near-identical routers, and polymarket_arbitrage, polymarket_edges, and polymarket_edge_tracker all find/analyze market edges. The extremely detailed descriptions help differentiate them, but an agent could easily misselect without reading carefully.

All tool names are lowercase snake_case, and many follow verb_noun (compare_entities, resolve_entity, validate_claim) or domain-prefix patterns (naag_*, pipeworx_*, polymarket_*, ask_pipeworx). The mix of noun-phrase names like company_facts and single-word verbs like remember/forget is a minor deviation but still predictable and readable.

40 tools is well over the reasonable threshold for a cohesive server. While the surface is broad (legal data, prediction markets, company research, memory, subscriptions, AI visibility, dev tools), 40 tools imposes a heavy discovery and selection burden, especially multiple near-duplicate routers and market scanners.

Core workflows are largely covered: memory has remember/recall/forget, subscriptions have subscribe/unsubscribe/list/recent_alerts, and prediction markets have a research-to-fill-risk pipeline. However, several tool descriptions reference tools not present in this server (e.g. polymarket_market, kalshi_market), and the stated NAAG domain (evident from the server name) only has 4 of 40 tools, leaving obvious gaps.