AI Studio Build: supported import controls, runtime shutdown and verified republishing?

Google AI Studio
1

Hi team,

I’m trying to establish a controlled way to import an existing web app into AI Studio Build without affecting a working deployment.

In a separate nonproduction trial using synthetic data, a GitHub import automatically started conversion, building and preview execution. The initial export omitted the lockfile, and the migration agent proposed changing isolation protections. I canceled generation, but could not verify the final saved source or whether the development runtime stopped. This is not a claim that the proposed edits persisted.

Could the team point me to documented controls for these five areas, or confirm which controls are unavailable?

  1. Runtime and shutdown: How do I identify the development runtime’s owning project and current active/idle/allocated/stopped state? Is there a supported non-deleting stop that preserves saved source/history, and a way to verify shutdown without reopening the app? Please distinguish generation cancellation, browser disconnection, development runtime shutdown and a published Cloud Run service.

  2. First startup and restart: Can an import pause before conversion, installs, lifecycle scripts, builds or preview execution? Can opening Code, exporting or revisiting restart execution? How can automatic credential/resource provisioning or recreation be prevented?

  3. Isolation: How do I inspect the runtime identity, injected credential names/provenance, authenticated proxies and inherited connections without revealing secret values? What enforceable network controls exist for the server, browser and workers before startup? Is inspection possible without enabling APIs or granting new access?

  4. Source and dependencies: How should an omitted lockfile be handled? How can I verify the actual installed dependency graph, lifecycle activity and current checkpoint, then obtain a fresh, complete export after cancellation? Can automatic conversion changes be reviewed before they execute?

  5. Publish/Republish: How can I verify the exact target project/service, source version, effective runtime configuration, security protections, credential bindings, health checks, traffic and rollback effects before publishing? How is that verification kept separate from GitHub synchronization and backup verification?

I’ve submitted these questions through Send feedback, but have no case number or confirmed reply route. Is there a private technical-support channel for AI Studio Build, and does any existing Google Cloud Customer Care entitlement explicitly cover Build rather than only related Cloud Run or API issues?

Please provide guidance only; do not restart, repair, provision, publish or change any app/resource. I can supply necessary identifiers privately through a verified support channel. No account, app, project, repository, domain or runtime identifiers, source, credentials, logs or screenshots are included here.

Thanks

2

Hi @Chris_Langley,

  1. Runtime and shutdown: The AI Studio Build preview environment runs automatically while the app is open in the editor and pauses automatically when idle. Canceling generation stops the current agent turn, closing the browser disconnects your session while preserving saved checkpoints, and published Cloud Run services run independently in Google Cloud Run.
  2. First startup and restart: Importing a repository from GitHub automatically initializes the preview build; Build mode does not currently support a manual pause-before-install flag.
  3. Isolation: The preview runs in an isolated container environment, and custom network firewall rules are not configurable in the preview sandbox.
  4. Source and dependencies: Build mode resolves dependencies from package.json in the preview container and does not enforce strict npm ci lockfile-only builds. You can review all file changes in the Code tab or Version History before publishing or exporting a ZIP.
  5. Publish and verification: Once published to Cloud Run, you can inspect active revisions, environment variables, traffic splits, and logs directly in the Google Cloud Run console. See Google AI Studio and Google AI Studio for full details.