MCP server for web search, URL fetching, and research with content summarization
Three tools with complete input schemas and descriptions, but descriptions are generic and lack LLM-optimized guidance. All parameters have types and descriptions, meeting baseline requirements. However, output schemas are undocumented (no structured response format visible), and error handling lacks recovery guidance. Tool names follow verb_noun convention (web_search, web_fetch, web_research) but descriptions don't explain when to use each vs. the others or what makes them distinct. No parameter constraints (enums) for provider/mode fields that accept specific values.
Fetch a URL, clean the page, and return a focused summary.
Search the web, fetch top pages, summarize them, and return a focused research answer with sources.
Search the web and return clean title, URL, and snippet results.
Output schemas undocumented. No structured response format defined for any tool. LLMs cannot plan downstream operations or extract specific fields from results.
Provider and mode parameters accept free-form strings instead of enums. 'provider: groq or copilot' and 'mode: summarize|lossless|chunks' should be declared as enum constraints to prevent hallucinated values.
Descriptions lack LLM-optimized guidance. No explanation of when to use web_research vs. web_fetch, or what makes web_search distinct. Descriptions are 50-80 chars; baseline for A+ is 50-200 chars with WHEN/WHY context.
No error handling guidance. If a URL fetch fails or a search returns no results, the tool returns raw errors with no recovery hints (e.g., 'Try a different query' or 'Check URL format').
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Parameter relationships undocumented. 'mode=chunks' requires 'top_k', but this dependency is not stated in descriptions. LLMs may pass mode=chunks without top_k, causing silent failures.