SymPy MCP server has well-structured tool definitions with complete JSON schemas and clear descriptions. All 8 tools follow verb_noun naming (sympy_solve, sympy_simplify, etc.) and include parameter descriptions. However, descriptions are minimal (10-50 chars), lacking context on WHEN to use each tool or dependencies between them. Parameter descriptions are present but terse. No output schema documentation visible. Error handling is basic, no recovery guidance or actionable error messages. Tool #8 (sympy_list_operations) is a discovery tool but lacks explanation of when/why to call it. No tool annotations (readOnlyHint, destructiveHint) despite all tools being read-only.

• Expand tool descriptions to 100-200 chars. For each tool, state: (1) What it does, (2) When to use it vs similar tools, (3) What it returns. Example: 'Solves symbolic equations for a variable. Use when you need exact algebraic solutions (e.g., x**2 - 4 = 0). Returns a list of solution strings. For numerical evaluation, use sympy_evaluate instead.'
• Document output schemas for all tools. Specify: 'Returns: {type: array, items: {type: string}, description: "List of solution strings"}' or similar. This enables LLMs to plan chained calls.
• Add readOnlyHint: true to all tool definitions in the MCP schema. This signals to agents that these tools are safe to call without side effects and can be retried freely.
• Enhance error messages with recovery guidance. Instead of returning raw Pythonx errors, catch them and return: 'Failed to parse expression. Ensure variables are single letters (x, y, z) and operators are valid (**, +, -, *, /).' This guides LLM self-correction.
• Add parameter constraints to sympy_evaluate. Document that 'substitutions' is optional and should be a JSON object mapping variable names to numeric values (e.g., {"x": 2, "y": 3}).