Three tools with basic descriptions but significant schema and parameter documentation gaps. Tool names are action-oriented (discover, validate, check_command) but descriptions lack LLM-optimized structure. Input schemas visible but lack parameter-level type info and constraints. No output schemas documented. Error handling is minimal, no recovery guidance or actionable error messages. Naming is clear but descriptions are generic and under-optimized for agent selection.

• Document output schemas for all three tools. E.g., discover returns {manifest: McpManifest, source: string, errors: string[]}. validate returns {valid: boolean, errors: string[]}. check_command returns {available: boolean}.
• Add parameter-level type info and constraints. E.g., discover's 'input' param: type=string, description='Local file path, direct .json URL, or base URL. Must be valid path or URL format.' Add format validation (regex or enum) to reject invalid inputs early.
• Rewrite descriptions to be LLM-optimized (50-200 chars). E.g., discover: 'Discover and parse MCP manifest from file, URL, or well-known endpoint. Returns parsed manifest or list of errors. Call this first to load server config.' validate: 'Validate manifest structure and semantic correctness. Call after discover to check for config key mismatches and missing required fields.'
• Add error recovery guidance. E.g., discover error response: 'Discovery failed: {errors}. Try: 1) Verify URL is accessible, 2) Check file path exists, 3) Ensure manifest is valid JSON, 4) Call check_command to verify server availability.'
• Add tool composition hints. E.g., discover description: 'Returns manifest object suitable for validate(). If validation fails, use check_command to verify server is installed.' This guides multi-step agent planning.