Loading ToolBench
MCP Server for accessing TogetherCrew's data and agents.
Single tool 'query-community-resources' has critical definition gaps. Input schema uses Zod with proper types (string, string) but lacks parameter descriptions entirely, both 'communityId' and 'question' have null descriptions. Tool description is missing. Output is unstructured text with no schema documentation. Error handling returns generic 'Error fetching data...' with no recovery guidance. Naming is acceptable (verb_noun pattern) but insufficient to overcome missing descriptions and schemas.
Tool description is completely missing. LLMs cannot determine when or why to select this tool without a description explaining its purpose, use case, and what it returns.
Both input parameters lack descriptions (communityId and question have null descriptions). LLMs cannot infer what values to pass or how to obtain them without explicit parameter documentation.
Output schema is not documented. Tool returns unstructured text wrapped in a generic content array with no field documentation. LLMs cannot plan downstream operations or extract structured data.
Error handling is non-actionable. Generic 'Error fetching data...' message provides no recovery guidance, error classification, or details about what failed. LLM cannot self-correct or retry intelligently.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
No parameter constraints or format documentation. 'communityId' and 'question' lack guidance on expected format, length, or valid values. LLMs may pass invalid inputs without validation feedback.