Scrapingbee

Multiple tools have overlapping boundaries: ask_pipeworx, ask_pipeworx_beta, ask_pipeworx_grounded, and deep_research all answer natural-language questions through the same routing core, with the beta version currently identical to the stable one. The Polymarket family also has several overlapping 'find an edge/arb' tools, making misselection likely without very careful reading.

The set is uniformly lowercase snake_case and mostly readable, but it mixes verb-led names (compare_entities, discover_tools, validate_claim), noun phrases (company_facts, entity_profile, recent_changes), and brand-prefixed product names (polymarket_kalshi_spread, pipeworx_trending, scrapingbee_scrape). There is no consistent verb_noun pattern, though naming within each family is somewhat predictable.

39 tools is well beyond the 3-15 range of a focused server, and many are meta/infrastructure utilities such as pipeworx_feedback, pipeworx_trending, suggest_questions, and the remember/recall/forget memory trio. The surface feels like several product areas stapled together rather than one well-scoped toolkit.

The surface is deep in some pockets—Pipeworx data lookup and Polymarket/Kalshi analysis are heavily covered—but shallow elsewhere: a server called Scrapingbee offers only three scraping tools with no crawling, pagination, or scheduling coverage. The lack of a coherent domain makes coverage hard to assess, and obvious gaps on the scraping side would likely cause agent failures.