Nrel

Multiple tools have overlapping boundaries: ask_pipeworx and ask_pipeworx_beta are exact duplicates in practice, and ask_pipeworx_grounded, deep_research, entity_profile, validate_claim, and bet_research all serve query-driven research with fuzzily separated purposes. The polymarket_* cluster also spreads opportunity discovery, arbitrage, fill-risk, and edge decay across tools that an agent could easily misselect between. Detailed cross-referencing helps, but not enough to remove the clutter.

All names use snake_case and are readable, but the pattern is mixed: verb-first tools like ask_pipeworx and compare_entities coexist with noun-first tools like entity_profile and pipeworx_trending, plus bare-verb tools like forget and recall. The pipeworx_ and polymarket_ prefixes provide some grouping, but there is no uniform verb_noun convention across the set.

31 tools is heavy for the apparent purpose, especially because the set mixes unrelated utilities (generate_llms_txt, scan_dependency, memory trio) with a large prediction-market suite and multiple overlapping query entry points. Many of these could be consolidated into a handful of core data-lookup tools with parameters, and the beta duplicate inflates the count.

The research/data domain is quite thoroughly covered: universal routing, grounded hallucination-resistant answers, claim verification, entity profiles, comparisons, deep multi-source research, and prediction-market analytics are all present. Minor gaps remain—no dedicated tool for fetching a pipeworx:// citation directly, and no way to update/correct the data itself—but these work around via resources and MCP resource/read functionality.