codex - 💡(How to fix) Fix Feature Request: Auto-resume CLI session when usage limit resets [1 participants]

When Codex hits the usage limit mid-task, the error already tells me exactly when the quota resets (e.g. "try again at 6:34 AM"). If that reset time falls while I'm asleep, the information is wasted — the task just dies and waits for me to wake up, find the session, and re-issue the prompt. The reset timestamp is already structured data in the protocol (codex-rs/protocol/src/error.rs → UsageLimitReachedError.resets_at) and the session already updates rate limits on this error (codex-rs/core/src/session/turn.rs ~line 1075). Today we just throw the timestamp away and crash the turn. The ask is: use it.

TL;DR

Implement an auto-resume feature that utilizes the existing resets_at timestamp in the UsageLimitReachedError to sleep until the quota resets and then re-run the same turn.

Guidance

• Modify the turn.rs file (~line 1075) to handle the UsageLimitReachedError by sleeping until the resets_at time instead of crashing the turn.
• Add an opt-in configuration option (--auto-resume-on-limit or auto_resume = true in ~/.codex/config.toml) to enable this feature without changing existing user behavior.
• Update the error handling logic to parse the resets_at timestamp and schedule a retry at that time.
• Consider implementing a queueing mechanism to handle multiple turns that are waiting for the quota to reset.

Example

// Pseudo-code example, actual implementation may vary
if let UsageLimitReachedError { resets_at } = error {
    let sleep_duration = resets_at - current_time;
    sleep(sleep_duration);
    retry_turn();
}

Notes

The implementation details may vary depending on the specific requirements and constraints of the Codex project. This guidance provides a general outline of the steps needed to implement the auto-resume feature.

Recommendation

Apply workaround: Implement the auto-resume feature as described, allowing users to opt-in to this behavior and providing a seamless experience when hitting usage limits.