When you retire an agent, what do you actually revoke?
retained text
I keep a few agents around across many tasks, and I realized I never wrote down what happens when one gets retired. My setup: each agent works inside its own Linux container on my Mac. No home folder, none of my passwords, only its own workspace. It signs in to its own accounts from inside that container, with its own browser. A file only reaches my Mac when I save it. That made starting an agent easy. Ending one turned out to be less obvious. Things people have pointed out to me: 1. Deleting the container or the browser profile does not end the sessions. A token that got copied somewhere else keeps working until the service revokes it. 2. Per agent credentials stay alive for weeks after nobody uses them. 3. A read only filesystem does not stop a signed in browser from changing things inside an account. So my current rule: revoke at the service first (sessions, tokens, connected app access), then delete the container. And keep a list per agent of everything it signed in to, because by the end you forget. The line I keep coming back to: deleting the agent is cleanup, revoking its sessions is offboarding. Limits: this is my own small setup on one Mac, and a container is a strong boundary, not a guarantee. I also have not solved how to know about sessions an agent opened that I never saw. What is on your offboarding checklist for an agent?
document fetched 2026-10-29 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1xho24c/when_you_retire_an_agent_what_do_you_actually