Fix problems
Grok Bot routines not running: checks in the right order
By Tal Slavin · Last checked · Next review 2026-11-06
Grok Bot routines not running: checks in the right order
Routines that never fire, run late or show Failed: the checklist, hourly limits, expired approvals, webhook routines and the October 2026 incidents.
◈Key Takeaways (5)
- •Check status first: routines failed on Cursor's side on October 8, 9, 13 and 19, and missed runs were not replayed.
- •Run history tells you whether it never fired or fired and failed.
- •Limits: 300 runs an hour across routines and 200 per routine; extra runs fail without retry.
- •Unattended approvals expire and the action then does not run.
- •Routines that use your own Mac or PC need the Grok Bot window active.
☰On this page (5 sections)
Rule out the platform first, because during an incident nothing you change helps. Then check the routine itself.
Checklist#
- Enabled, not Paused. A routine can pause itself after Grok Bot asked and got no reply.
- Schedule and time zone are what you meant (Settings, General, Bot, Timezone; Auto-detect follows the computer you are on).
- The owning Bot still exists.
- Plugins and connectors the routine uses are still signed in.
- Run history: no entry means it never fired; a Failed entry means it started and did not finish. "Automation run rate limit exceeded" means a cap was hit.
- Usage has not run out; see usage limits.
Common causes#
| Symptom | Cause | Fix |
|---|---|---|
| Many routines late or Failed at once | Platform incident | Check status; leave routines on |
| Run-rate-limit error | 300 an hour in total or 200 per routine | Narrow the trigger or widen the schedule |
| Action missing from a completed run | Approval expired | Notifications on; narrow rules, not broad Always allow |
| Routine on your Mac or PC failed | Local helper pauses when the screen is locked or the window is idle | Bring the Grok Bot window forward before the job |
| Fires hours off schedule | Time zone; new-Bot routines used UTC before 0.78.1 | Update and recheck |
| Webhook routine has no URL | Older bug (before 0.53.0) or a hidden Details field | Update the app |
Webhook and event-triggered routines also need a working endpoint, and short-interval routines suffer most when the scheduler is behind.
The October incidents#
- October 8 to 9: from about 10 AM ET, scheduled routines were held back, ran late or were marked Failed. Staff said pausing, editing or recreating routines did not help.
- October 13, morning US time: held back again for a few hours.
- October 13, 17:52 to 20:29 UTC: many webhook, Slack, git, email and Linear triggers did not run; events were not retried.
These were on Cursor's side, not a configuration error. After an incident, run anything important that was missed by hand. Use Test run only with a harmless input, because it performs real external actions.
Our full routines guide has the fix order for each case.
Frequently asked questions#
Why did my routines fail on October 8 or 13?#
Both were problems on Cursor's side. From about 10 AM ET on October 8, scheduled routines were held back, ran late or were marked Failed into October 9. On October 13 many webhook, Slack, git, email and Linear triggers did not run for a few hours, and the events were not retried.
What is the hourly limit?#
Since 0.75.5 your routines can run up to 300 times an hour in total and 200 per routine. Runs past either limit fail and are not retried.
Why did an action in my routine not happen?#
Approvals raised by unattended work expire, and the action then does not run. Turn on notifications for the Bot that owns the routine.