All 5 tools lack input parameter schemas entirely, no JSON Schema definitions visible in source. Descriptions are minimal (8-27 chars), well below the 10-1024 char baseline and insufficient for LLM tool selection. No parameter descriptions, no output schema documentation, no error handling guidance. Tools are registered but definitions are skeletal. This is a wrapper library, not a production MCP server, and the examples demonstrate only the CLG integration pattern, not tool quality.

• Add JSON Schema input definitions to all tools. calculator should define two numeric parameters (operand_a, operand_b) with type, min/max, and description. create-invoice should define invoice fields (customer_id, items, amount, etc.).
• Expand tool descriptions to 50 - 200 characters. E.g., 'calculator: Add two numbers and return the sum. Accepts two numeric operands (integers or floats). Returns the result as a number.' This enables LLM selection and parameter inference.
• Document output schemas. E.g., calculator returns {type: 'object', properties: {result: {type: 'number'}}}. create-invoice returns {type: 'object', properties: {invoice_id: {type: 'string'}, status: {type: 'string'}}}.
• Add parameter descriptions and constraints. E.g., operand_a: 'First number to add (integer or float, range -1e9 to 1e9)'. customer_id: 'Unique customer identifier (string, required)'.
• Implement basic error handling. Validate inputs (e.g., calculator rejects non-numeric operands with 'operand_a must be a number'). Return structured error responses with actionable guidance for LLM retry.