GitHub outage hits Actions, API, PRs and SAML auth worldwide
A component failure has degraded CI, git ops, logins and Copilot since 13:40 UTC; GitHub found the cause but residual errors are still hitting builds.
Copy markdown
What's degraded: nearly the whole platform
API requests, Actions, Pull Requests, Issues, webhooks and git operations are all impaired. Web and API traffic hit roughly 20% error rates, while archive and raw-content downloads peaked near 50%.
Four-plus hours in and still not clear
GitHub began investigating at 13:40 UTC, said it identified the faulty component and applied fixes by 16:36, and moved to monitoring at 16:59 — but was still flagging residual degraded performance past 17:36 UTC.
SSO logins are failing, not just slow
SAML, OIDC, SCIM and Team Sync are all impacted, so orgs behind single sign-on can be locked out rather than merely slowed. If your CI or deploy authenticates through GitHub, expect intermittent 401s and jobs that never start.
Copilot went flaky too
Copilot availability degraded from 14:31 UTC alongside Actions and the API, since it rides GitHub's auth and backend. If completions stall in your editor, it's the outage, not your config.
Do this now: don't fight the red X
Residual errors mean retries often succeed, so re-run failed Actions jobs rather than rewriting them, hold non-urgent deploys, and resist panic force-pushes. Packages and Codespaces stayed largely up if you need a path around it.