Are Xcode Cloud Workflow Start Conditions broken this week?

Additional data from a controlled repro (all times UTC, 2026-07-23):

The failure is intermittent, starting roughly Jul 20–21: automatic PR triggers have failed on several days this week (requiring manual starts, e.g. runs 604 on Jul 21 and 609 on Jul 22), yet also succeeded in between — PULL_REQUEST_OPEN/UPDATE runs fired normally at Jul 22 23:51 (build 611), Jul 23 02:24 (612), and Jul 23 05:36–06:31 (614, 616, 617). The most recent successful automatic trigger was today 06:31:56 (build 617).

Today's first failure, reproduced in a controlled way: a new PR created at 19:46 (clean head commit, no skip markers, same workflow, base main) produced no build run and no check on the PR after 20+ minutes. Closing and reopening the PR to generate a fresh pull_request event also produced nothing.

Meanwhile, everything on the API side works:

• PR ingestion: GET /v1/scmRepositories/{repoId}/pullRequests returns the new PR with correct metadata (SCM PR id

098509a9-5243-4741-abca-877aaa5381c4).

• Manual triggering of the same workflow against that same PR works immediately: POST /v1/ciBuildRuns (workflow 2ab3fb84-1c94-4ed1-a84c-b2e6ab84d054,

that pull-request id) created run 888e622d-f3af-4c58-895a-d2f601ab6f3c (startReason MANUAL) at 20:12:17, which SUCCEEDED at 20:20:19.

So Apple's outbound path (ASC → GitHub with the App installation token) is fully healthy — repo access, PR metadata, build execution all work. What's failing, intermittently, is only the inbound event path: GitHub's pull_request events reaching Xcode Cloud's start-condition evaluation. These are different mechanisms, so "the integration has the permissions it needs" doesn't clear the webhook path — and since Xcode Cloud integrates as a GitHub App, its event webhook is registered on Apple's side, not in my repo settings, so I have no delivery log to inspect from my end. Repo id 2fa9eb0d-6b3b-4755-8534-22a039e3ccf0, app com.ArkonLabs.Protego.

I'm having the same issue with Code Push event on a self-hosted Gitlab. I have it set so that any code push to staging causes a new build. The triggers stopped working with week. I can go ahead and manually generate a build from the Xcode Cloud tab on ASC, which, as you said, confirms that ASC -> repo connection valid, but the auto generation via the web hook does not work

Feedback Assistant:

FB23948570

Update, six hours on: the degradation has moved a level deeper. This morning the PR was ingested and a manual API start worked around the dead trigger; tonight PR ingestion itself has stopped, so there is no manual path either.

Controlled data (all times UTC, 2026-07-24; same repo 2fa9eb0d-6b3b-4755-8534-22a039e3ccf0, same PR workflow 2ab3fb84-1c94-4ed1-a84c-b2e6ab84d054):

• 01:43:10 — new PR opened (clean head commit, no skip markers, base main). As before: no build run, no check on the PR.
• 02:11, 28 minutes later — GET /v1/scmRepositories/{repoId}/pullRequests (paginated) still does not return this PR. The newest PR it returns remains this morning's (098509a9-5243-4741-abca-877aaa5381c4). This is the step that still worked this morning.
• Ref ingestion is fine and now diverging from PR ingestion: GET .../gitReferences picked up the PR's branch promptly (0b5618c4-a9d4-45f1-af7f-2bfb38d7efe8).
• With no scmPullRequest object, the manual workaround is gone: POST /v1/ciBuildRuns with the workflow plus the branch's git reference is rejected ("The manual build cannot be created because the branch ... is not associated with the workflow" — correct behavior for a PR workflow, but it means a PR-scoped workflow currently has no start path at all), and starting the build from the App Store Connect UI doesn't work either.

So the inbound pipeline has gone from "pull_request events not evaluated against start conditions" to "pull_request objects not ingested at all", while everything outbound (repo access, ref ingestion, build execution) stays healthy.

Will attach this to FB23948570.

The Auto-Start for "Any Branch" is also a dead trigger specifing a exact branch name is a unhelpful work around. Can also be added to your Feedback.

Seeing similar issues. Recent branches aren't being seen by Xcode Cloud at all. Any new branch starting with release is supposed to trigger a build. After 2 hours nothing. I tried starting manually but the branch wasn't in the list. Verified multiple times that it was pushed and available.

It finally showed up long enough to get a build created only to disappear on me an hour or so later along with any history of the build. It's been 20+ hours and it's still missing.