AI-Assisted Development

Use Studio’s integrated coding agent to build and revise a Python application while you inspect its live UI and viewport. The agent can edit your project files and help implement SDK calls, Python handlers, and frontend controls.

Assistant setup

Open the assistant panel in Studio. It uses Codex installed on your computer. Follow the setup instructions in the panel if Codex is unavailable, then sign in and select a model.

Studio uses your Codex account, configuration, and permissions. Studio’s own sign-in is separate.

Application requirements

Start from a sample and save it into your project folder. Ask for a concrete change, such as:

Add a button that creates a cube at the X position entered in a numeric field. Frame the scene after creating it.

The agent works on the saved project. Review its changes in the editor, run the app, and exercise the new control. Continue with specific feedback about the behavior you need.

Testing with the agent

Enable app QA for a request when you want the agent to run the project, exercise the relevant workflow, and inspect its result. Studio gives the agent access to the preview for that request. The panel explains the permissions and app information involved before you enable it.

You can also run the app yourself and give the agent the observed result. Use the console and engine logs to diagnose failures.

Development workflow

Keep the Python SDK calls and your UI code in ordinary project files. Add dependencies through project configuration. Use your preferred frontend framework for a custom UI, and follow Sharing apps when another developer needs to run the project.