Two tools with minimal descriptions and basic schemas. 'add' (18 chars) and 'calculate' (40 chars) descriptions lack context on WHEN to use each tool, WHY they differ, or what the LLM should expect. No parameter descriptions visible in schema, only type and enum constraints. No output schema documented. No error handling guidance (e.g., divide-by-zero returns plain text, not structured error). Tool naming is acceptable but lacks verb clarity ('add' vs 'calculate' distinction is unclear to an LLM). Schemas are present but sparse. No tool annotations (readOnlyHint, idempotentHint). Composition is weak: 'add' and 'calculate' overlap (both do addition), should consolidate or clarify when to use each.

• Expand tool descriptions to 100-200 chars. For 'add': 'Performs simple addition of two numbers. Use this for basic arithmetic when you only need to add two values. Returns the numeric sum.' For 'calculate': 'Performs arithmetic operations (add, subtract, multiply, divide) on two numbers. Use this when you need multiple operation types in one call. Returns the numeric result or an error if division by zero.'
• Add descriptions to all parameters in the schema. E.g., 'a': 'First operand (numeric value)', 'b': 'Second operand (numeric value)', 'operation': 'Arithmetic operation to perform: add, subtract, multiply, or divide.'
• Document the output schema explicitly. E.g., 'Returns: { content: [{ type: "text", text: "<numeric result as string>" }] }' or return structured JSON with a 'result' field and optional 'error' field.
• Consolidate 'add' and 'calculate' into a single 'calculate' tool that accepts an operation enum (defaulting to 'add'). This eliminates confusion and reduces the LLM's decision space.
• Improve error handling for divide-by-zero and other edge cases. Return structured errors: '{ error: "DivisionByZeroError", message: "Cannot divide by zero. Ensure b is non-zero.", code: "INVALID_OPERAND" }' instead of plain text.