Documentation / Browser lifecycle
CLOUDBROWSER DOCUMENTATION

Browser lifecycle

Start, pause, resume and branch your work with predictable behavior.

CREATE→RUNNING⇄PAUSED→RESUME

Asynchronous operations

Create, pause, resume, fork and delete return 202 with an operation ID. Poll the operation until it succeeds or fails. --wait provides this behavior in the Go CLI; timeout does not imply the operation failed.

ActionCLIBehavior
createcloudbrowser create --name work --waitReserve a slot, create Chrome, wait for readiness.
pausecloudbrowser pause br_123 --waitStop compute, revoke runtime-bound connections, preserve profile.
resumecloudbrowser resume br_123 --waitCheck allowance and concurrency, start Chrome; get a fresh CDP grant or reopen an existing share.
forkcloudbrowser fork br_123 --name copy --waitPause the source first; fork into a new paused browser by default.
deletecloudbrowser delete br_123 --yesRevoke access, delete browser; terminal operation.

Safe parallel work

Starting and resuming browsers reserve capacity atomically. A paused fork does not consume running capacity. An idempotency key prevents duplicate create/fork operations on retries.

Fork consistency

The local implementation requires the source to be paused. Explicit pause closes Chrome gracefully before snapshotting its filesystem, allowing buffered profile writes to finish. Paused fork has been verified with synthetic persistent Cookies and localStorage. Fork does not promise a copy of live JavaScript memory. Never attach one profile directory to two concurrent Chrome processes.

Browser automation activity is independent from connection lifetime. Closing a viewer or detaching a tool does not pause the browser.
Private preview · REST API · CLI designExplore console