Loading ToolBench
Single tool with minimal definition quality. Tool name 'web_search' is appropriately verb-prefixed and clear. Description is present but extremely brief (44 chars), below the 10-1024 char guideline and well below the 194-char baseline for A+ tools. Input schema exists with proper Zod typing but parameter description is minimal (20 chars, at the floor). No output schema documented, LLMs cannot predict response structure. No error handling guidance; errors return raw API messages without recovery hints. No pagination support despite search results potentially being large. Tool is READ_ONLY (good), but lacks idempotent/readonly annotations in the MCP protocol.
Search the web using Eureka Search API
Tool description is 44 characters, far below the 194-char baseline for A+ tools and insufficient for LLM selection reasoning. Current text 'Search the web using Eureka Search API' lacks context on when to use this vs other search tools, what data is returned, or prerequisites.
No output schema documented. LLMs cannot predict the structure of response.data from the Eureka API. This forces LLMs to guess at field names and types, increasing hallucination risk and wasting tokens on exploratory calls.
Parameter 'query' description is only 20 characters ('The search query'). No guidance on format, length limits, special characters, or examples. LLMs cannot infer constraints and may pass invalid queries.
Error handling returns raw API error messages without recovery guidance. E.g., 'Search API error: 401 Unauthorized' tells the LLM nothing, should be 'API key invalid or expired. Verify EUREKA_API_KEY environment variable is set correctly.'
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | 2026-07-28+ | v2 |
No pagination support. Search results could be large; tool accepts no limit or offset parameters. Returning unbounded results risks context window exhaustion and token waste.