Schemas & Tools
ANTS Platform stores structured output schemas and tool definitions per project, so you can reuse them across Playground windows when you test structured output and tool calling.
Schemas and tools are created, edited and attached from the Structured Output and Tools sections of a Playground window. They are saved to the project and listed there for reuse. They are not used by evaluators or experiments.
Structured output schemas
A schema is a named JSON Schema that the model's response must follow, for example an object with required fields and enums.
- Give it a name and description, and write the JSON Schema.
- Attach it in a window's Structured Output section. The model is asked to return JSON that matches.
- Use it when your application parses the response and needs a fixed shape.
Tools
A tool is a function definition the model can choose to call.
| Field | Meaning |
|---|---|
| Name | The function name, for example get_weather. |
| Description | What the tool does and when to use it. The model reads this. |
| Parameters (JSON Schema) | The arguments the tool accepts. |
Attach tools in a window's Tools section. When the model decides to call a tool, the Playground shows the tool call and its arguments. It does not run your function. To continue the conversation, add the tool result as a message yourself.
Use tools in the Playground to check whether a model calls the right tool with the right arguments before you wire up real execution in your agent.