GitHub's Actions automation platform and Pages hosting service both suffered failures on Thursday, marking the latest in an increasingly long run of stability problems for the Microsoft-owned developer platform. The outage affected core CI/CD workflows and related tools including Copilot code review, Copilot coding agent, hosted runners and GitHub Enterprise Importer migrations, disrupting developers' ability to build, test and deploy code, and reinforcing growing concern that GitHub has become unreliable for serious production use.

GitHub first flagged degraded Actions performance at 15:22 UTC, with availability issues and outright workflow failures following within the hour, alongside errors on the Actions REST API and unexpected rate limiting; Pages was added to the list of affected systems soon after. By 17:40 UTC, engineers said mitigations had been applied and a further fix was rolling out, though webhook deliveries remained delayed. This is the sixth incident logged in August alone, following 26 in July, 23 in June, 23 in May and 26 in April; GitHub apologised for outages in April and pledged fixes in June, but the incident rate has not improved, with critics such as Ghostty developer Mitchell Hashimoto arguing the platform can no longer be trusted for serious work.