A Model Context Protocol (MCP) server that provides RAG (Retrieval-Augmented Generation) capabilities for document querying
Two tools with significant definition gaps. rag_docs has a reasonable description (194 chars, within baseline) and proper schema with typed query parameter. configure_rag lacks actionable description (only 50 chars, below 20-char floor) and exposes api_token as a parameter, a critical security violation. Neither tool has output schemas documented. No error handling guidance. rag_docs returns inconsistent types (sometimes TextContent list, sometimes dict), violating response structure expectations. configure_rag modifies global state (AVAILABLE_TOOLS), creating composition and idempotency concerns.
Configure RAG tools with API token and base URL
It performs an intelligent query on the RAG document database. This tool searches company documents and provides answers based on actual content, including references to source documents. Ideal for finding specific information, details about people, processes, products or any documented content.
API token exposed as tool parameter (configure_rag). Credentials must use server-side secret injection via environment variables or vault. Agent traces log all parameters, secrets leak into logs and prompt history.
configure_rag description is 50 chars and non-actionable ('Configure RAG tools with API token and base URL'). Does not explain WHEN to call it, WHAT it modifies, or that it has side effects. Below 20-char floor triggers 0-20 description score.
Output schemas not documented. rag_docs returns either TextContent list or dict depending on success path, inconsistent and undocumented. LLMs cannot plan downstream calls without knowing response structure.
No error handling guidance. Timeout and API errors return raw messages without recovery hints. LLM cannot determine if error is retryable, user-fixable, or fatal.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | <=2025-11-25 | v2 |
configure_rag modifies global AVAILABLE_TOOLS state and must be called before rag_docs works. This creates hidden dependencies and non-idempotent behavior. Agents cannot safely retry or compose tools.