Errors with GitHub APIs delaying workflows

We have identified an issue with GitHub API requests that may cause some customers to see workflows stuck in a running state or workflows fail to start.

GitHub has reported incidents on the APIs

https://www.githubstatus.com/incidents/ph5nns5y4gxj

https://www.githubstatus.com/incidents/8vfyvq16hzh9

We are monitoring the stability of GitHub APIs and will update this page as more information is available.

We are also seeing errors processing push event webhooks due to the upstream API outage.

Users may need to retry pipelines for these push events.

We are seeing signs of recovery from the GitHub API.

We will stay in a monitoring state to verify that the API error rate has recovered.

GitHub API error rates have recovered to normal levels.

During the incident, some customers may have seen pipelines fail to start or not receive status updates. Some pipelines may be stuck in a running state.

• You should re-push commits or use the UI to manually trigger pipelines that didn't start.
• Pipelines in a stuck running state should be canceled and rerun.

If you encounter any issues please contact CircleCI support.

## Summary

On July 20, 2026 from 00:22 to 01:45 UTC, CircleCI customers using our GitHub integrations experienced failures running pipelines and experienced workflows getting stuck. During this incident, some pipelines triggered by users using GitHub failed to start or ran as error pipelines. Customers whose pipelines failed or ran as error pipelines during this window should re-run them.

This was caused by an upstream API degradation at GitHub, which impacted many GitHub APIs that CircleCI uses to trigger and run pipelines.

By 01:45 UTC, GitHub APIs recovered, and customer pipelines ran normally.

The original CircleCI status page can be found here.

## What Happened

(all times UTC)

Beginning 00:22 on July 20, 2026, GitHub began returning elevated errors for the following API requests:

• `applications/*/token`
• `repos/*/commits`
• `repos/*/*/contents/*`
• `repos/*/*/hooks`
• `repos/*/*/hooks/*`
• `repos/*/*/keys`
• `repos/*/*/pulls`
• `repos/*/*/statuses/*`

CircleCI relies on these requests to correctly trigger and run pipelines.

At 00:22, our internal monitoring alerted us to the problem. Our team began investigating, and found that a small number of pipelines belonging to GitHub projects either failed to start or ran as error pipelines. This included scheduled pipelines and workflows. In addition, a small number of customer workflows experienced stuck jobs and required a re-run of the workflow to fix.

At 01:45, GitHub recovered and pipeline processing returned to normal operating levels. Only pipelines triggered during the incident window were affected, and customers should re-run those pipelines.

## Future Prevention and Process Improvement

We are actively working to improve the resilience of pipeline processing during GitHub service disruptions to reduce the customer impact of similar incidents in the future. We are scoping features that will help customers recover with reduced manual steps after incidents resolve.

Customer experience is our top priority, and we commit to continually improving the reliability of our systems to match the trust that our customers place in us. Please reach out to our support team with any questions or concerns.