Tool Calling

Enable LLMs to call functions and interact with external systems.

When to Use This Skill

Invoke this skill when:

• Adding function calling to agents
• Designing tool schemas
• Integrating external APIs
• Implementing validation and error handling

Parameter Schema

Parameter	Type	Required	Description	Default
task	string	Yes	Tool calling goal	-
provider	enum	No	anthropic, openai, langchain	anthropic
strict_mode	bool	No	Enable strict validation	true

Quick Start

Claude Tool Use

from anthropic import Anthropic

tools = [{
    "name": "get_weather",
    "description": "Get weather for a location",
    "input_schema": {
        "type": "object",
        "properties": {
            "location": {"type": "string"}
        },
        "required": ["location"]
    }
}]

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    tools=tools,
    messages=[{"role": "user", "content": "Weather in Tokyo?"}]
)

LangChain Tools

from langchain_core.tools import tool

@tool
def search(query: str) -> str:
    """Search the web for information."""
    return web_search(query)

Schema Best Practices

# Good: verb_noun, clear description
{
    "name": "search_products",
    "description": """Search product database.
    USE WHEN: User asks about products.
    DO NOT USE: For order status (use get_order instead)."""
}

# Bad: vague
{"name": "helper", "description": "Helps with stuff"}

Troubleshooting

Issue	Solution
Tool not called	Improve description
Wrong tool selected	Add "DO NOT USE" conditions
Invalid arguments	Enable strict mode
Execution timeout	Add timeout, retry logic

Best Practices

• Use verb_noun naming convention
• Keep tool count under 20
• Include usage examples in descriptions
• Return errors as tool results

Related Skills

• ai-agent-basics - Agent loops
• llm-integration - API setup
• agent-safety - Input validation

References

• Anthropic Tool Use
• OpenAI Functions