Skip to main content

Tool

Taxonomy​

Tool execution:

  1. Built-in: atomic toolkit.
  2. Function calling.
  3. Bash: composable static scripts.
  4. Codegen: dynamic programs.
  5. Agent calling: agent tools, subagents, multi-agent systems.

Documentation​

Documentation is important:

  • Use clear name and description.
  • Simplify parameter lists, describe input and output parameters.
  • Add targeted examples.
  • Provide default values.
def get_product_information(product_id: str) -> dict:
    """
    Retrieves comprehensive information about a product based on the unique product ID.

    Args:
        product_id: The unique identifier for the product.

    Returns:
        A dictionary containing product details. Expected keys include:
            'product_name': The name of the product.
            'brand': The brand name of the product
            'description': A paragraph of text describing the product.
            'category': The category of the product.
            'status': The current status of the product (e.g., 'active', 'inactive', 'suspended').

    Example return value:
        {
            'product_name': 'Astro Zoom Kid's Trainers',
            'brand': 'Cymbal Athletic Shoes',
            'description': '...',
            'category': 'Children's Shoes',
            'status': 'active'
        }
    """

Agentic​

Describe actions, not implementations:

  • What over how.
  • Explain tool interactions over dictate (规定) workflows.
  • Don't duplicate instructions.

Granularity​

  • Define clear responsibilities.
  • Don't create multi-tools (❌多功能工具❌).

Output​

Concise output:

  • Don't return large responses.
  • Use external data storage for long-term memory.
  • Provide descriptive error messages.

References​