Azure Functions-based MCP server for orchestrating LLM conversations with tool integration, memory management, and web search capabilities
Two tools defined with basic schemas and descriptions. search_web has a complete input schema with 5 parameters (query required, focus_mode enum, optional context/question/user_language), but output schema is not documented in visible code. hello_mcp lacks any visible schema or parameter documentation. Both descriptions are present but minimal (search_web: 95 chars, hello_mcp: 54 chars). No error handling guidance, no output structure documentation, no tool annotations. Naming follows verb_noun pattern (search_web is clear), but hello_mcp is generic and non-descriptive. Parameter descriptions in search_web are adequate but terse. No evidence of idempotency hints, permission gates, or recovery guidance.
Default MCP tool for testing tool integration
Perform a web search via Azure Function (SearXNG) with optional focus modes.
hello_mcp tool has no visible input schema, no parameter documentation, and a generic non-descriptive name.
Output schemas not documented for either tool. LLMs cannot plan downstream calls or extract required fields without knowing response structure. search_web response format is completely undocumented.
No error handling guidance. Tools do not indicate what errors are retryable, user-fixable, or fatal. No recovery suggestions provided to LLM on failure.
Tool descriptions are minimal (54-95 chars). search_web lacks context on when to use it vs other search tools, what focus_mode does, and what results look like. Descriptions should be 50-200 chars with WHAT/WHEN/WHAT-RETURNS guidance.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | <=2025-11-25 | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). search_web is marked READ_ONLY in metadata but this is not exposed in the tool definition where LLMs can see it.