codex - 💡(How to fix) Fix Codex Desktop automations should support missed-run catch-up after the app or computer was offline

What feature would you like to see?

Codex Desktop automations should have a clear, user-visible catch-up policy for scheduled runs that were missed because the app was closed, the computer was asleep, or the computer was powered off.

A useful default would be:

• when Codex starts or wakes, detect automations whose scheduled time passed while Codex was unavailable
• for each automation, either run one catch-up occurrence or explicitly skip it according to a configured policy
• show the user what happened, e.g. “missed at 09:30, ran on wake at 11:12” or “missed at 09:30, skipped because catch-up is disabled”

This could be exposed as a per-automation setting such as:

• catch_up = true|false
• misfire_policy = run_once|skip|run_all_due
• max_catch_up_age = 24h

The exact names do not matter much; the important part is that Codex does not silently wait until the next weekly/daily occurrence when the user reasonably expected a missed local automation to run after their machine comes back online.

Why this matters

I created a weekly Codex Desktop automation for a low-frequency monitoring task. I then realized that if my Mac is off at the scheduled time, the automation will not run, and it appears it will not automatically run when I start the computer later. For weekly checks, that means a single missed wake window can delay the task by a full week.

That behavior is surprising for user-facing automations, especially for tasks like:

• weekly repo/skills monitoring
• CI or PR follow-up
• inbox or calendar summaries
• lightweight reminders
• periodic local health checks

Most users will not think in terms of scheduler “misfires”; they will just expect Codex to notice that a scheduled task was missed and either run it soon after wake/startup or make the skip explicit.

Desired behavior / acceptance criteria

1. Codex stores enough state to know that an automation occurrence was missed while the app/device was unavailable.
2. On app startup or wake/resume, Codex applies a documented catch-up policy.
3. The automation run history distinguishes normal scheduled runs from catch-up runs.
4. The UI makes missed/skipped/caught-up runs visible enough that users can trust the scheduler.
5. There are guardrails to avoid runaway catch-up after long downtime, e.g. run only the latest missed occurrence by default.
6. The policy is documented for local Desktop automations so users know whether their laptop must be awake at the exact scheduled time.

Related

This is related to #8317, especially the --misfire=run-once|skip idea, but the request here is specifically for existing Codex Desktop automations and the common local-machine case where the app/computer was offline at the scheduled time. It is also adjacent to automation reliability bugs such as #17893, but this feature request is about the intended behavior after offline/sleep/power-off, not about automations failing while the scheduler is otherwise active.

Additional information

Environment where the expectation came up:

• Codex Desktop on macOS
• Local cron automation scheduled weekly
• User asked whether it would run if the computer was off, and whether it would catch up after the computer is turned on later