mcp-doctor exposes 2 well-named tools with clear descriptions and proper schema definitions. Both tools are read-only, properly annotated with tool safety hints, and have explicit input schemas with type information. However, parameter descriptions could be more detailed in some cases, and output schemas are documented in prose rather than formal JSON Schema. The tool set is narrowly scoped to evaluation/reporting, which is appropriate for its domain.

• Add formal JSON Schema definitions for output schemas of both tools. For check_server, define the structure of the returned report object with fields like 'evaluation', 'overall_grade', 'dimensions', 'ai_review'. For list_dimensions, define the 'dimensions' array schema.
• Enhance parameter descriptions for 'format', 'mode', and 'model' in check_server: add constraints like 'format must be one of: json, markdown' and document error scenarios, e.g., 'If mode=ai and OPENAI_API_KEY is not set, returns error with hint to set the environment variable.'
• Document the structure and fields of the error response object returned by check_server (e.g., 'error' and 'hint' fields) so agents know what to expect on failure.
• Add a brief note to list_dimensions explaining when/why an agent should call it (e.g., 'Call this before check_server to understand the 6 dimensions being evaluated').