I set up a Cursor automation to automatically fix failing CI on an open pull request. I’ve used the trigger “CI completed” on my repository. Our repo is a monorepo that intentionally skips some CI jobs based on which files are edited. When I test this automation on a PR, the Cursor UI returns an error message saying that the CI hasn’t yet completed. I strongly suspect this is because of the skipped tests.
Steps to Reproduce
Create an animation using the CI complexity trigger on a repo that has skipped CI jobs.
Expected Behavior
I would expect the automation to trigger when all jobs have either completed or were skipped.
Hey, thanks for the report. A couple of questions to narrow this down:
Are you seeing this when you click the “Test” button in the automation UI, or is the automation also failing when it’s triggered normally via webhook, like after CI actually finishes on a PR?
On the PR you’re testing, are there any other checks besides the skipped ones that might still be running or pending at the time? For example, third-party bots, coverage tools, and so on.
Can you share a link to a GitHub PR that shows this behavior? That’d help us confirm exactly how GitHub is reporting the check run statuses.
The “Test” button does a one-time check of CI status and doesn’t poll, so if any checks are still in progress at that exact moment, it’ll return the “CI not yet completed” error, even if the skipped jobs are fine. The webhook flow includes polling, so the behavior might be different there.
Even after all CI jobs are either done or skipped, we dont’t see the automation trigger.
the repo is plumefinderTeam/aiplumefinder but it’s private. if useful, I can work on setting-up an open repro. this is a bit lower prio for us, so I can’t commit on any ETA for this.