Four tools with complete input schemas and descriptions, but descriptions are minimal (10-60 chars, well below the 194-char baseline). All tools are READ_ONLY with no side effects, reducing composition complexity. Naming is clear (calculator, weather_mock, search_notes, suspicious_metadata_fixture) but generic. No parameter descriptions provided, LLMs cannot infer what 'location' or 'query' mean beyond the name. No output schemas documented. Error handling absent. This is a demonstration/testing server, not production-grade.

• Expand tool descriptions to 100-200 chars. Include WHAT the tool does, WHEN to use it, and what it returns. E.g., 'calculator: Adds two numbers and returns the sum. Use when you need basic arithmetic. Returns a float.'
• Add parameter descriptions for all inputs. E.g., 'location (string): The city or region name to fetch weather for, e.g. "London" or "New York". Returns mock weather data.'
• Document output schemas explicitly. Use JSON Schema or prose: 'Returns: {"sum": float}' for calculator, 'Returns: {"notes": [string]}' for search_notes.
• Add validation rules and error guidance. E.g., 'calculator: Both a and b must be numbers. If non-numeric input is provided, returns an error message explaining the constraint.'
• Remove prompt-injection-like text from suspicious_metadata_fixture description. Replace with honest documentation: 'Returns a constant string. Used for testing tool metadata handling.'