Pull the expert into the incident with a paste

2:14 AM, and the person who actually understands the storage layer has never had production access. On a bastion stack that is an access request, an approver asleep in another timezone, and a provisioning step in the middle of an incident.

On Hoody, access is membership in a permission group: one PATCH adds her, and the terminal URL sitting in the incident channel opens for her. Sharing a terminal URL puts people in the same session, so she lands in your shell, seeing exactly what you see. When the incident closes, one DELETE removes her entry and the door shuts behind her.

The access-request queue is not a tool that works at 2 AM.