GSD Orchestrator exposes only 2 tools with minimal documentation. CallToolAsync has a generic description that does not explain WHEN to use it or what makes it different from direct MCP calls. ListToolsAsync lacks any description. Both tools have input schemas present but descriptions are far below the 50-200 char LLM-optimized baseline (10 and 0 chars respectively). No parameter descriptions beyond bare type info. No output schema documentation. No error handling guidance. No tool annotations. This server is a thin wrapper around MCP client operations, not a domain-specific tool suite.

• Add a 100-150 char description to ListToolsAsync explaining it discovers available MCP tools and when to call it (e.g., 'Call this first to see what tools are available before planning multi-step operations').
• Expand CallToolAsync description to 80-150 chars: explain that it wraps tool calls with Polly retry + circuit breaker, handles secondary rate limits with 65s backoff, and is the preferred way to invoke MCP tools from this orchestrator.
• Document the 'args' parameter: 'JSON object containing tool-specific arguments. Structure depends on the tool being called, use ListToolsAsync to discover available tools and their parameter schemas.'
• Add output schema documentation: CallToolAsync returns {text: string, isError: boolean}; ListToolsAsync returns {tools: [{name: string, description: string, inputSchema: object}]}.
• Add error handling guidance to CallToolAsync: 'On secondary rate limit (HTTP 429), automatically retries after 65s backoff. On circuit breaker open, returns non-transient error, check downstream service health. On other errors, returns isError=true with error text in the response.'

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

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