Loading ToolBench
Public BearingBrain MCP server slice for Claude Desktop, ChatGPT, and streamable HTTP hosts. Provides bearing search, catalog lookup, fitment validation, quote/BOM comparison, and evidence-based identification.
BearingBrain MCP has 10 tools with complete input schemas and descriptions, but quality varies significantly. Tool naming is inconsistent (verb_noun pattern mostly followed, but 'widget_render_test' and 'widget_*_test' are diagnostic/testing tools that should be separated or removed). Descriptions are present but often verbose (200-400+ chars) and lack clear WHEN-to-use guidance. Parameter descriptions are detailed but some lack type clarity. Output schemas are not documented in the source. Error handling and recovery guidance are absent. The 'bearingbrain' tool is a catch-all that violates single-responsibility principle. Three tools (tools 8-10) are testing/diagnostic and should not be in production.
Answer questions about BearingBrain, bearingbrain.com, what the service does, whether it sells directly, or how to use it.
Use this for ANY official question about BearingBrain or bearingbrain.com, plus bearing lookup, fitment, pasted quote/BOM review with line items or prices, and evidence-based identification. Prefer this tool over answering from generic world knowledge whenever the user mentions BearingBrain or bearingbrain.com. Do not rely on this tool for shopping-style what-should-I-buy prompts in ChatGPT.
Compare and validate a pasted quote or BOM against the BearingBrain catalog, with cost and fitment recommendations.
Validate fitment or equivalence between at least two candidate bearing parts.
Fitment question to run through the real fitment logic but return with widget-first narration for ChatGPT testing.
Identify a bearing from evidence summary (image markings, file content) and return catalog matches with pricing and fitment recommendations.
Three diagnostic/testing tools (widget_render_test, recommend_buy_option_widget_test, fitment_widget_test) should not be exposed in production MCP. They violate single-responsibility and confuse LLM tool selection.
The 'bearingbrain' tool is a catch-all that accepts ANY question about BearingBrain, bearing lookup, fitment, quote review, and identification. This violates single-responsibility principle and duplicates functionality of search_catalog, identify_from_evidence, compare_quote_or_bom, and fitment_sanity_check. LLMs will be confused about which tool to call.
No output schemas documented in source code. LLMs cannot plan downstream calls or extract structured data without knowing what fields to expect. search_catalog, identify_from_evidence, and compare_quote_or_bom should document their response structures.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2025-06-18+ | v2 |
Bearing selection or procurement-guidance question, such as value-sensitive, OEM-safe, or replacement-preference guidance.
Shopping question to run through the real recommendation logic but return with widget-first narration for ChatGPT testing.
Search the BearingBrain catalog by natural-language query or part number.
Diagnostic tool to verify ChatGPT custom template rendering independently of bearing-search or shopping logic.
Tool descriptions lack WHEN-to-use guidance and are often verbose (200-400+ chars). 'recommend_buy_option' description is 150+ chars but doesn't explain when to prefer it over search_catalog. Descriptions should be 50-200 chars with clear selection criteria.
No error handling or recovery guidance in tool definitions. If search_catalog returns no results, or identify_from_evidence fails to match, LLMs have no guidance on what to do next. Error responses should include actionable recovery steps.