17 tools with basic schemas and descriptions, but significant gaps in parameter documentation and error handling. Most tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization). Parameter descriptions are sparse or missing entirely. No output schemas documented. Error handling is absent. Tool naming is clear and verb-based, but composition issues exist (ask_question conflates RAG and agent modes). This is a typical community server with functional but incomplete definitions.

• Expand all tool descriptions to 50-200 characters. Include WHAT the tool does, WHEN to use it, and any prerequisites. E.g., 'Create a new conversation within a collection to start a multi-turn chat session. Requires a valid collection_id from list_collections().'
• Document output schemas for every tool. Specify field names, types, and whether they are required. E.g., create_conversation should return {conversation_id: string, title: string, created_at: ISO8601}.
• Add range constraints and format hints to numeric and string parameters. E.g., 'top_k (integer, 1-100): Number of top results to retrieve. Default: 10.'
• Split ask_question into two tools: ask_question_rag (RAG-only, no tool calling) and ask_question_with_agent (uses MCP tools if configured). This clarifies intent and prevents mode confusion.
• Add error handling guidance to destructive tools. E.g., delete_collection: 'This operation is irreversible. Returns 400 if collection not found, 403 if user lacks permission. Consider archiving instead of deleting.'