Open the bug at the line where it failed

Sentry fires. The alert email has a link: not to a stack trace web page, but to a VS Code URL. You click. The editor opens, the file is already open, the cursor is on line 247: exactly where the exception was raised. The container has a snapshot of production state from the moment the error fired.

You don't git pull. You don't reproduce locally. You don't ask "what version was deployed?" The snapshot is the answer to all of those questions.

You add a print, you add a breakpoint, you re-run the failing request against the snapshotted state. The bug is right there. Five minutes from page-fired to fix-pushed.

The branch lives just long enough to write the fix. Then the snapshot is destroyed, the container goes idle, and the production state is back to its own canonical line.