GitHub degraded: ~20% error rates across web, API, and Actions

GitHub opened an incident at 13:40 UTC on August 17, 2026, saying it was investigating degraded performance across some services. Over the following twenty minutes the status page added API Requests, Actions, Webhooks, Issues, and Pull Requests to the affected list. By 14:04 UTC the company put numbers on it: roughly 20% error rates for web experiences and API traffic, and roughly 50% for archive downloads and raw repository content downloads.

The higher failure rate on archive and raw content downloads is the part most likely to bite automated systems. Anything that fetches a tarball or a raw file during a build — dependency installs pointing at GitHub, container images pulling source, scripts curling a raw URL — sees one in two attempts fail rather than one in five. Combined with degraded Actions and Webhooks, that means CI pipelines can fail in ways that look like broken code rather than an upstream outage, and webhook-triggered deploys may simply not fire.

At the time of the last update in the report, the root cause was still unknown. GitHub said investigations were ongoing and that it would continue posting updates. The report does not state a resolution time, a trigger, or whether any data or queued events were lost, and it gives no indication of which underlying systems were implicated.

The practical read for anyone shipping today: treat retries on GitHub fetches as expected, and check the status page before debugging a red build.