Scoring was not performed

• Document output schema for each tool. For list_api_endpoints, specify that it returns {tools: Array<{name: string, description: string, resource: string, operation: string, tags: string[]}>}. For invoke_api_endpoint, specify what the endpoint result looks like (structured object, string, array, or varies).
• Expand tool descriptions to 50-150 chars explaining PURPOSE and WHEN TO USE. Example: 'list_api_endpoints: Search available API operations by name, resource, tag, or method. Call this first to discover endpoints before invoking.'; 'invoke_api_endpoint: Execute a discovered endpoint with arguments. Use get_api_endpoint_schema first to see required params.'
• Add explicit constraints to search_query parameter. Example: 'An optional search query (max 100 chars, case-insensitive) to filter endpoints by name, resource, operation, or tag. Partial matches included.'
• Restructure invoke_api_endpoint parameter description for args: 'An object containing arguments for the endpoint. Keys and values must match the schema returned by get_api_endpoint_schema tool. See that tool first to understand required and optional fields.'
• Add error handling that guides recovery. In the invoke_api_endpoint handler, catch errors and return messages like: 'Endpoint not found. Available endpoints: [list]. Try list_api_endpoints with a search query.' or 'Invalid arguments for endpoint. Expected schema: [show schema]. Check get_api_endpoint_schema for details.'
• Add parameter constraints (min/max string length, enum values for operation if applicable) using JSON Schema validation, not just descriptions.