mcp-doctor is a linter/analyzer tool, not a functional MCP server. The 6 tools listed are EXAMPLE tools from test fixtures (bad_server_ts, good_server_ts, bad_server) used to demonstrate linting rules. The actual mcp-doctor codebase does not implement these tools, it parses and audits them. This is a fundamental category error: we are evaluating a static analysis CLI, not an MCP server. The 'tools' are synthetic examples embedded in test cases. Scoring reflects the example tools' quality, which is intentionally mixed (bad_server examples fail checks; good_server examples pass). Per-tool scores below reflect this mixed-quality test corpus.

• Clarify that mcp-doctor is a static analysis tool for auditing MCP servers, not an MCP server itself. Separate the linter codebase from example fixtures.
• In bad_server examples, add comments explaining which checks they fail (missing descriptions, untyped params, generic names) to make the pedagogical intent explicit.
• In good_server examples, ensure all parameter types are consistent (integer vs number for days) and all descriptions meet the 10 - 1024 character guideline.
• Add a 'perfect_server' example that passes all 86 quality rules, demonstrating the target state for users.
• Document the 54 patterns and 86 rules that mcp-doctor checks, with links to arcade.dev/patterns for each.