The repro video is a query

QA reproduces the checkout bug in a hoody-browser container and, by habit, reaches for the record button.

There's nothing to record. The browser logs every navigation to persistent storage: API-driven and clicked-through alike. GET /history?since=2026-08-28T09:00:00Z returns the exact path: each URL, page title, HTTP status, and timestamp, filterable by domain. The bug report links the history query and the container itself. The developer reads the six steps in six seconds, then opens the same live browser: still logged in, still sitting on the broken cart: and walks the path forward from there.

No six-minute recording, no scrubbing for the moment it went wrong.

The repro isn't a video of the path. It's the path, with the machine still attached.