Replay & debug

How replay works

Play runs in this browser, against the live page. It is not a headless grid and not a cloud runner.

• Real input events via the Chrome DevTools Protocol
• Retries when selector resolution is flaky
• Waits for the page to settle before the next step
• Follows tabs the flow opens, plus iframes and open shadow DOM

A debug loop

1. 1Play the journey (or the selected folder).
2. 2If a step fails, open the failure screenshot and the step that broke.
3. 3Edit that step — wait, assert, offset, variable — instead of re-recording the whole flow.
4. 4Play again. Run history keeps the newest 50 runs; older runs are pruned.

What you will see

Each step gets a clear status as it runs. Failures capture a screenshot locally. Chrome will show a debugger banner while the debugger API is attached; that is expected. See permissions.

Before you hit Play

• Pick the environment if the flow uses variables.
• Fill variable values for anything that was masked at capture.
• For file-upload steps, set an absolute path after record — replay cannot guess the file you picked last time.

Honest limits

• Closed shadow DOM is session-scoped and can still fail on some hosts.
• Virtualized lists need a recorded scroll.
• Complex rich-text editors may need a script step.
• Debugger cannot attach to chrome:// or the Chrome Web Store.
• This is Chrome-only in v1. It is not Playwright, Selenium, or a CI worker — how they differ.