Seamless resume mechanism after quota interruption

What variant of Codex are you using?

CLI

What feature would you like to see?

The 5-hour quota currently allows for roughly a ~30-minute Astra Max session, which makes it very difficult to use effectively for longer tasks.

What is especially frustrating is that there is no seamless continuation after the quota is exhausted. When the quota becomes available again, I tell the agent to continue because the previous session was interrupted, but it often resumes from an older state instead of exactly where it stopped.

This wastes a significant amount of quota. In some cases, the agent uses as much as a quarter of the entire new 5-hour quota just to reconstruct the previous state and get back to the point where the session was interrupted, even if the terminal was not closed.

I would strongly suggest one of two solutions:

1. Remove the 5-hour limit and keep only the weekly quota, as was previously the case; or
   2. Implement a seamless resume mechanism so that the agent does not even perceive the quota interruption as a new session.

For example, there could be a command such as /resume_after_quota_loss that requires no additional prompt and simply makes the agent continue from the exact state where it stopped, with the full working context preserved - option available at least if the terminal was not closed.

This would make quota usage much more efficient and significantly improve the experience of using Codex for longer-running tasks.

Additional information

No response

The issue concerns the Codex CLI losing session continuity after quota exhaustion and proposes a /resume_after_quota_loss command; no files, tests, or implementation entry points are named. Start by locating the CLI's session and quota handling, then define how working context survives a quota reset. Done means an open terminal can continue from the exact interruption point without reconstructing prior state.