Critical gaps across all dimensions. Tool definitions exist only in test file (not in actual server code visible), making them inferred rather than explicitly registered. No input schemas visible for 6 of 8 tools. Descriptions are present but generic (avg ~120 chars, below 194-char baseline). No parameter descriptions for required inputs. No output schemas documented. No error handling guidance. No tool annotations. This server fails basic definition quality standards.

• Move tool registration from test file into server.py. Use mcp.server.Server.tool() decorator or register_tool() with explicit Tool objects containing name, description, and inputSchema.
• Add JSON Schema inputSchema to all tools. For analyze_codebase and project_statistics, document any optional parameters (e.g., language_filter, include_metrics). For find_* tools, add 'symbol' and 'function' parameter descriptions: type, length constraints, and examples of valid formats.
• Expand tool descriptions to 100-200 chars. Include WHAT the tool does, WHEN to use it, and WHAT it returns. Example: 'Find all references to a symbol in the codebase. Returns file paths, line numbers, and context. Use after find_definition to understand usage patterns.'
• Document output schemas in descriptions or as explicit outputSchema fields. Specify returned fields (e.g., 'Returns array of {file: string, line: number, column: number, context: string}').
• Add error handling guidance. For find_* tools: 'If symbol not found, returns empty array. Try a partial name or check spelling.' For analysis tools: 'If project has no code, returns empty metrics. Ensure project_root points to a valid codebase.'

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

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