Tool Schema Definition
What you'll learn: How to explicitly structure tool descriptions so an LLM agent knows exactly what parameters a tool accepts, what types they are, and how to use them correctly.
The Challenge of Tool Invocation
When your AI agent decides to use a tool (like a web search function or database query), it needs precise instructions. The LLM doesn't inherently "know" what parameters search_products(query, max_results, category) expects. Without a clear schema, the agent might pass a string where an integer is required, omit mandatory fields, or misunderstand what each parameter does.
What is a Tool Schema?
A tool schema is a structured specification that describes:
- Tool name: The function identifier
- Description: What the tool does (guides action selection)
- Parameters: Each argument's name, type, description, and constraints
- Required vs. optional fields: Which parameters are mandatory
- Return value description: What the tool produces
Think of it like an API documentation page, but formatted specifically for LLM consumption—typically as JSON or structured text.
Example Structure
{
"name": "search_products",
"description": "Search the product catalog by keyword",
"parameters": {
"query": {
"type": "string",
"description": "Search keywords",
"required": true
},
"max_results": {
"type": "integer",
"description": "Maximum number of results",
"default": 10,
"minimum": 1,
"maximum": 50
},
"category": {
"type": "string",
"enum": ["electronics", "clothing", "books"],
"required": false
}
}
}
The schema guides the LLM to generate valid tool calls with correct types and constraints, reducing errors in the perception-action loop.
Key Takeaway: Tool schemas are structured specifications that teach LLMs exactly how to invoke tools correctly by defining parameter names, types, descriptions, and validation constraints—essential for reliable agent behavior.