Agent is the core AI controller. It takes screenshots of the virtual desktop, interprets the screen using a vision-language model, and performs actions like clicking, typing, and navigating.
Constructor
| Parameter | Type | Default | Description |
|---|---|---|---|
model | str | None | None | Default VLM for all calls on this agent. When None, the SDK sends no model override and the platform picks server-side. See Models for the full list of accepted values. |