GitHub, the world’s largest developer platform owned by Microsoft, experienced a roughly nine-hour outage of its core automation service, GitHub Actions. The disruption triggered a cascading impact on its AI coding assistant Copilot and static web hosting service GitHub Pages.

According to GitHub Status, the company’s official service status page, GitHub first detected performance degradation in GitHub Actions around 12:22 a.m. KST on August 7. Some workflows failed to start entirely or failed mid-execution, while Actions API requests also returned errors. Users also reported encountering unexpected rate limits.

GitHub Actions is a CI/CD (continuous integration and continuous deployment) service that automatically runs testing, building, and deployment tasks when developers push code to a repository. During this outage, jobs piled up in queues or timed out and were forcibly terminated. Even self-hosted runners operated by enterprises experienced issues properly fetching jobs.

The impact quickly spread to other major GitHub services. Webhook throughput—which links code pushes and pull requests to automated tasks—plummeted to as low as 15% of normal levels at one point. Jobs for GitHub Enterprise Importer, a tool for migrating enterprise repositories, were completely halted. Additionally, Copilot code review, Copilot coding agent, and GitHub Pages availability all suffered severe performance degradation.

GitHub’s engineering team identified the root cause: jobs that were no longer valid were being continuously and incorrectly assigned to runners. After applying a fix, the job success rate for running workflows recovered to 99%, and the backlogged job queue gradually cleared. At 9:01 a.m. KST, GitHub officially announced that new jobs and webhook-based workflows had begun processing normally, and that GitHub Pages and Copilot-related features had also been restored.

However, migration jobs using Enterprise Importer remain suspended as a precautionary measure. GitHub has not yet provided specific details on the exact cause of the outage.

This incident once again highlights GitHub’s ongoing service instability over recent months. According to a tally by IT publication The Register, GitHub recorded 26 incidents in April, 23 in May, 23 in June, and 26 in July. A similar Actions outage occurred just days ago on July 29. In August, six incidents have already been logged in just six days, indicating worsening service instability.

Developer community frustration over these frequent outages is mounting. Mitchell Hashimoto, developer of the popular terminal emulator Ghostty, publicly criticized the platform, stating it has become too unstable to host “serious work” anymore. Reports indicate that a significant number of developers are considering migrating projects to other platforms.

GitHub issued a formal apology in April for the persistent outages, but the situation has reportedly not improved significantly since then. The company has previously cited system overload from surging AI usage as one cause of the disruptions. In June, Jakub Oleksy, GitHub’s senior vice president, promised structural improvements that would “permanently eliminate failure modes.” However, The Register noted that based on the subsequent incident history, those improvements either have not yet taken effect or were not properly implemented.

GitHub did not respond to The Register’s request for additional information regarding this latest outage.