Skip to main content
Custom tools let you define a tool interface (name + JSON schema) and let the assistant choose when to call it. When it calls your tool, it returns a tool_use content block with structured input.

​
Prerequisites

​
Requirements

In the tools array, provide:
  • name: unique tool name
  • description: what the tool does
  • inputSchema: JSON Schema describing the tool input

​
Basic request and response

This example validates a request that defines a tiny calculator tool and forces its use. 
curl -X POST "https://{BASE_URL}/api/gpt/v1/messages/validate" \
  -H "Authorization: Bearer {API_TOKEN}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "default",
    "max_tokens": 64,
    "messages": [
      { "role": "user", "content": "Add 5 and 7." }
    ],
    "tools": [
      {
        "name": "simple_calculator",
        "description": "Add two numbers.",
        "inputSchema": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "a": { "type": "number" },
            "b": { "type": "number" }
          },
          "required": ["a", "b"]
        }
      }
    ],
    "tool_choice": { "type": "tool", "name": "simple_calculator" }
  }'
Example response: 
{ "valid": true, "errors": null }
To run the request, send the same body to POST /messages. When the assistant returns a tool_use block, execute your tool and include a tool_result block in the next message.