Composer, models, and execution settings
The Composer is the request and follow-up area in Studio. It is the safest place to describe a change because the request is attached to the current project and its history.
Write and send a message
Enter a new outcome, correction, or question in the text area. Use the available attachment or voice controls only for material that belongs in the project. Review the selected route and model before sending. During an active run, the Composer can be disabled or provide a stop/continue action according to the current Agent state.
Model selector
The selector can separate platform-default models from custom connections. A model row may show a provider icon, display name, description, and configured credit conversion multiplier; custom choices may show only the available identity. A missing description or multiplier is not a promise that usage is free. The selected model must support the input and task you intend to send. Your organization can restrict the list.
Execution mode and reasoning depth
Execution controls can offer modes such as planning, discussion, execution, or a general interaction mode, plus a reasoning-depth setting. Use planning to request a proposed scope before implementation, discussion to resolve questions, and execution to ask the Agent to make and verify changes. Exact labels depend on the project and platform version. Higher reasoning settings may take longer or consume more of the applicable usage allowance.
Continue after a run limit
If the Agent reaches a continuation limit, read the message and the saved progress. Use the explicit continue action only when you want the same task to proceed. Each continuation is a user-directed step; review the resulting status before requesting another one. Do not repeatedly submit a second copy of the task while the first run is active.