Server has 5 tools with complete input schemas and descriptions present. However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool or what to do with results. Parameter descriptions are minimal (e.g., 'Collection name' vs. 'The MongoDB collection to query; use list_collections to discover available collections'). No output schemas documented, LLMs cannot predict response structure. Error handling is present but not recovery-focused. Tool names follow verb_noun convention (good), but lack clarity on relationships (query_documents vs aggregate both search; no distinction in names). Missing: pagination guidance, field-level descriptions for complex params (filter, pipeline, projection), and idempotency notes for write operations.

• Document output schemas for all tools. Example: list_collections returns {collections: [{name: string, type: string}]}. Include field descriptions so LLMs know what to extract.
• Expand parameter descriptions to 50-150 chars with actionable guidance. E.g., 'Collection name (string). Use list_collections first to discover available collections. Returns error if collection does not exist.'
• Add state-modification warnings to insert_document: 'Inserts a new document (WRITE operation). Retrying with identical document will create duplicates. Consider using upsert pattern if idempotency is required.'
• Clarify query_documents vs aggregate in descriptions. E.g., query_documents: 'Simple MongoDB queries with filter, projection, limit. For complex multi-stage transformations, use aggregate instead.'
• Add pagination guidance to query_documents and aggregate. Document the limit cap (100), explain skip/limit pattern, and note when results are truncated.

Inferred effective spec: <=2025-11-25.

• Relies on Stateful initialize / Mcp-Session-Id (removed; protocol is stateless) - make each request self-contained