This proxy server exposes 11 tools that forward requests to a remote MCP server. Tool naming follows verb_noun convention (list_*, get_*, read_*, set_*, subscribe_*, call_*, complete). However, descriptions are minimal (10-40 chars), parameter schemas are sparse or missing entirely, and output schemas are undocumented. Most tools lack parameter descriptions. The 'complete' tool has vague parameter names ('ref', 'argument') with no type information. No error handling guidance, no recovery hints, and no validation constraints visible. Tool definitions appear to be dynamically registered based on remote capabilities, but the actual schema definitions in the source are incomplete.

• Expand tool descriptions to 80-150 characters. Include WHEN to use the tool and what it returns. E.g., 'List all available prompts from the remote MCP server. Call this first to discover what prompts are available before requesting a specific one.'
• Document output schemas for all list_* tools. Specify that they return an array of objects with fields like 'name', 'description', 'arguments'. Include example structure.
• Add parameter descriptions to 'complete' tool. Clarify what 'ref' and 'argument' represent, their expected types, and how they relate to completion requests.
• Add error handling guidance to tool descriptions. E.g., 'call_tool: If the tool does not exist on the remote server, an error will be returned. Verify the tool name using list_tools() first.'
• Add pagination support to list_* tools with optional 'limit' and 'offset' parameters. Document the default limit (e.g., 20) and maximum allowed (e.g., 100).

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

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry