All tests work for my automation but it fails every night with this message: “The run was rate-limited due to too many concurrent runs. Retry after a short delay or reduce the number of parallel runs.”
It is a simple automation, with no parallel agents or runs. At one point I even included specific instruction to never use a sub agents or multiple agents and still got this result. I use the API to kill any active cloud agents each day after testing and still this fails.
I’m at a loss as to what to do and the Cursor agent has not had any luck diagnosing why.
Has anyone seen something similar or have commands I can use to see all the agents?
I am seeing the exact same problem. I have nothing else running and this error occurs with a single automation. Running the same automations with “test” succeeds every time.
well I wish I had some good update info for you but I don’t. The last 2 days have worked but nothing was changed. Seems random or a Cursor backend issue
Thanks for raising this. The error message is a bit misleading. What you likely ran into were GitHub rate limits.
There’s nothing you can do in your configuration to fix this, and I can see that there have been elevated rates of rate limiting since this past weekend. I’ve flagged this for the team.
In the short term, shifting your automation cron off the top of the hour may reduce exposure to this.
Interesting. Any way to confirm this from the Github side? Seeing that the cloud agent doesn’t even spawn, (or is spawned, fails and self deletes) is this something that happens in a precheck of the Github PR tool ahead of the automation? Because that check usually happens in the agent chat, which leaves evidence of at least trying to load the tools. Good tip for moving off the hour I will try it. Failed again last night.
Good news: a fix for this went out on our side on Aug 19, and I can confirm your runs have launched and completed successfully every day since. @jonlervold this applies to your automation as well. Let us know if you still run into any issues.