Two tools with minimal descriptions and basic schemas. 'add' has a 20-char description (borderline); 'calculate' is 44 chars. Both lack context on WHEN to use them, prerequisites, or return value documentation. No error handling guidance (e.g., divide-by-zero is handled but not documented in the tool description). Parameters have types and descriptions, but lack constraints (e.g., no enum for operation in 'add', no range limits). Output schema is not documented, LLMs cannot plan downstream calls. No tool annotations (readOnlyHint, idempotentHint). Naming is acceptable (verb_noun pattern), but descriptions fall short of the 50-200 char LLM-optimized baseline.

• Expand tool descriptions to 50-200 chars. Example for 'calculate': 'Performs arithmetic operations (add, subtract, multiply, divide) on two numbers. Returns the numeric result. Use this for multi-operation workflows; use add() for simple addition only.'
• Document output schema in tool metadata or description. Specify: 'Returns a JSON object with field result (number) containing the computed value.'
• Add tool annotations: set readOnlyHint=true and idempotentHint=true for both tools to signal they are safe to retry.
• Document error conditions in 'calculate' description: 'Divide by zero returns an error message. Other operations always succeed.'
• Remove the 'add' tool and rely solely on 'calculate' with operation='add'. This eliminates redundancy and reduces LLM confusion.