Importing and publishing

Agent Lab provides an experiment and revision workflow around app source. Import and publish are explicit snapshot operations, not continuous synchronization with a running session.

Import a saved app

Open Agent Lab, select a saved app, and choose Import app. Review its copied source and settings in the new experiment before running it.

Import uses the saved app. It does not capture unsaved editor changes, copy live positions, or preserve the current in-memory script state. Later edits to the app do not automatically update the experiment.

Select what to publish

Save pending experiment changes and select the intended revision in History. Check the revision shown in Publish revision against the one used by the reviewed run.

A selected run and a selected publication revision are separate choices. The experiment's AI chat and annotations provide research context; they do not prove which snapshot was published.

Create or update deliberately

Create app saves a separate app from the selected revision. Update imported app targets the app originally imported and checks whether it has changed since import.

If that check finds a conflict, import the current app into a fresh experiment and reconcile the changes, or create a separate candidate. After a successful update, a later update requires a fresh import; the old experiment is not a continuing synchronization link.

Open the saved result

Use Open published app and verify instrument, source, sizing, and intended execution environment. Publication does not launch a server, resume a historical test as live execution, or place orders.

Launch the saved result separately. Read Lab publishing for revision details and launching for the next step.