This is a fixture/demo server with minimal tool definitions. All 3 tools have basic descriptions (10-60 chars) and simple integer/string parameters with descriptions. However, the tools are intentionally trivial (add, echo, error-raiser) and lack the depth expected of production tools. No output schemas are documented. The 'boom' tool has an empty input schema, which violates schema requirements. Descriptions are present but sparse, below the 50-200 char LLM-optimized baseline. No error recovery guidance, no pagination, no composition patterns. This is appropriate for a test fixture but would not meet production standards.

• Document output schemas for all tools. E.g., 'Returns: {"result": integer}' for add, '{"text": string}' for slow_echo.
• Expand descriptions to 50-200 characters. Add context: 'Add two numbers and return the sum. Use for arithmetic operations.' Include what the tool returns.
• Rename 'boom' to 'trigger_error' or 'raise_error' to follow verb_noun convention and clarify intent.
• For 'boom', provide a structured error response with actionable guidance: 'This tool always fails intentionally. Use it to test error handling paths.'
• Add parameter constraints where applicable. E.g., for slow_echo, document that delay_ms must be >= 0 and suggest reasonable bounds (0-60000 ms).