On July 21, 2026, between 07:41 UTC and 11:57 UTC, the SSH Authentication service was degraded and some SSH connections failed to authenticate. On average, 12.2% of SSH authentication requests failed, peaking at 15.7%. Both user RSA keys and deploy keys were impacted. This was due to a change in how our SSH service handled one public-key authentication method that caused the affected authentication attempts to be rejected as invalid. We mitigated the incident by reverting the change, after which SSH authentication returned to normal. We are working to expand our automated test coverage for our SSH public-key authentication flows to catch more edge cases and to improve observability and alerting on SSH authentication failures, to reduce our time to detection and mitigation of issues like this one in the future.
Between 06:39 and 18:11 UTC on July 20, 2026, the Copilot service experienced a degradation of the GPT 5.3 model due to an issue with our upstream provider. The upstream model provider returned intermittent errors for GPT 5.3 Codex requests, which caused some responses to fail. Auto mode requests that had selected GPT 5.3 Codex were also impacted. On average about 2% of GPT 5.3 Codex requests failed during this window. Copilot automatically routed eligible traffic away from the impacted provider to reduce customer impact. No other models were impacted.We worked with the upstream provider throughout the incident and confirmed sustained recovery before resolving.
Between July 19, 2026, at 23:05 UTC and July 20, 2026, at 03:55 UTC, Actions self-hosted and larger runners were unable to connect to GitHub. During this period, Actions jobs were delayed or failed when trying to acquire a runner. Jobs using standard and Mac hosted runners were not affected. Reconnection traffic from affected runners also increased load on GitHub APIs, resulting in 3-4 seconds of additional average request latency and elevated 5xx error rates.The incident was caused by a certificate lifecycle management failure in a subset of internal services, resulting in an SSL certificate expiration that disrupted runner connectivity. We restored service by rotating the affected certificate. Recovery began at 02:45 UTC. By 03:55 UTC, queued workflow backlog had been processed and workflow delay rates returned to normal.To prevent recurrence, we are strengthening certificate renewal automation, adding fallback expiry monitoring and alerting, and improving circuit-breaker protections during runner API disruptions to reduce the risk of cascading impact to other APIs.
Between July 19, 2026, at 23:05 UTC and July 20, 2026, at 03:55 UTC, Actions self-hosted and larger runners were unable to connect to GitHub. During this period, Actions jobs were delayed or failed when trying to acquire a runner. Jobs using standard and Mac hosted runners were not affected. Reconnection traffic from affected runners also increased load on GitHub APIs, resulting in 3-4 seconds of additional average request latency and elevated 5xx error rates. The incident was caused by a certificate lifecycle management failure in a subset of internal services, resulting in an SSL certificate expiration that disrupted runner connectivity. We restored service by rotating the affected certificate. Recovery began at 02:45 UTC. By 03:55 UTC, queued workflow backlog had been processed and workflow delay rates returned to normal.To prevent recurrence, we are strengthening certificate renewal automation, adding fallback expiry monitoring and alerting, and improving circuit-breaker protections during runner API disruptions to reduce the risk of cascading impact to other APIs.
From 22:21 UTC - 23:50 UTC on July 16, 2026, the REST API experienced significant degradation. During this period, about 39% of REST API requests failed with HTTP 500 level responses, with the errors peaking at 44.3%. We identified the issue as an infrastructure change that wrongly marked the majority of API backends in a single region as unhealthy. As a result, requests routed to those backends failed before reaching the application layer. To prevent this from happening again, we're improving our systems to catch this kind of invalid configuration before it reaches production. We'll also audit the related systems to make them more resilient to future changes, and we're increasing our monitoring sensitivity so we're alerted to problems like this sooner.
On July 16, 2026, GitHub Copilot users experienced elevated errors when using Claude Fable 5 from 17:33 UTC until mitigation at 22:04 UTC. The average error rate was 1.4%, with a maximum error rate of 30.85%. The issue was caused by degradation at an upstream model provider; other Copilot models were not significantly affected, and users could avoid the impact by selecting another model or Auto. Service recovered after the provider mitigated the degradation.
On July 16, 2026, between 08:50 UTC and 09:50 UTC, the GitHub MCP Server’s web_search tool experienced elevated failures. The average error rate was 42% and peaked at 82% of requests to the tool. Other GitHub MCP Server tools were unaffected. This was caused by degradation at a downstream web search provider.The incident was mitigated when the downstream provider recovered, after which we confirmed that the tool’s success rate had returned to normal.We are improving the tool’s resilience and failure handling to reduce the customer impact and duration of similar incidents.
On July 14, 2026, between 15:17 and 15:37 UTC, a rollout to GitHub's internal webhook delivery pipeline caused a subset of webhook delivery records to not be written to our webhook deliveries store after being processed and delivered successfully. Affected deliveries would be missing from the webhook delivery UI and API and won’t be available for redelivery. The root cause was an uncoordinated rollout: a change to how delivery records are handed off between pipeline components was deployed before the upstream components producing those records were updated to match. While the rollout was in progress, affected records were silently skipped rather than persisted, with no automatic retry. The impact ended as soon as the rollout was completed. About 2.4M delivery records were skipped (approximately 4% of the 20-minute impact window, 0.04% of a typical 24-hour period). Importantly, 95% of these deliveries reached customer endpoints successfully, only the record of the delivery is missing. Of the ~5% that failed to reach customer endpoints, only ~1.4% (5,463) map to webhooks that retried their deliveries in the past 28 days. To prevent recurrence, we are making delivery persistence safe-by-default, adding alert on drops in the persisted-delivery rate, and tightening rollout coordination for changes that span multiple components in the pipeline.
On July 14, 2026, the GitHub Codespaces service was degraded during two periods — between 06:00 UTC and 09:56 UTC, and again between 10:54 UTC and 12:53 UTC — and some users experienced intermittent failures or delays when creating new codespaces. Impact was concentrated in a subset of geographic regions. During the first period, the error rate averaged 0.5% and peaked at 4.6% of codespace creation requests. The second period was more pronounced, peaking at approximately 30% of codespace creation requests in the most-affected region before recovery. Both periods were caused by an unexpected surge in codespace creation from an abusive actor that drained the available compute capacity in the affected regions faster than it could be replenished. We mitigated the impact by identifying and stopping the sources of the excess creation volume, reducing the resources that could be consumed in the affected regions, and rebalancing traffic across regions to restore capacity. Codespace creation success rates returned to normal after each period. We are working to add automated, low-latency controls to throttle abnormal codespace creation and to strengthen our detection and safeguards, so we can reduce our time to detection and mitigation of issues like this in the future.
On July 13, 2026, between 13:11 and 13:53 UTC, some customers experienced failures starting and running GitHub Actions workflows, which also affected Copilot cloud agent sessions and GitHub Pages builds since they depend on Actions. During the peak of the incident, 30% of Actions jobs failed to start and 2% were delayed more than 5 minutes. The incident was triggered by a configuration change in an internal autoscaling component that contained outdated capacity threshold values. This caused a critical Actions service to scale below its required baseline, reducing capacity for workflow processing. We identified the regression, rolled back the change, and restored service capacity. New workflow executions recovered by 13:39 UTC. Full recovery was reached by 13:53 UTC after the queued backlog was drained. To prevent recurrence, we have added deployment guardrails to validate that autoscaling inputs are current and to detect drift between planned and live scaling state before autoscaling changes are applied.