December’s Incident Summary
GitHub recorded five incidents in December that resulted in degraded performance across its services. The issues spanned API access, Copilot features, and AI governance tooling, with no single root cause recurring across events.
Enterprise AI Controls Data Outage
A configuration change deployed on November 25 caused a multi-day disruption for enterprise administrators using the AI Controls page. From November 26 until December 8, users could not list agent session activity in the AI Controls view. The failure did not affect agent activity in audit logs, direct navigation to individual session logs, or AI agent management.
The misconfiguration stopped data from publishing to an internal Kafka topic that fed agent session activity to the AI Controls interface. GitHub corrected the configuration on December 8 and is improving monitoring of data pipeline dependencies while strengthening pre-deployment validation to prevent similar configuration issues in production.
Copilot Code Review Failures
A service degradation on December 15 caused 46.97% of Copilot Code Review requests to fail between 15:15 UTC and 18:22 UTC. Users saw the error: “Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.” The remaining requests completed successfully.
The problem stemmed from elevated response times in an internal model-backed dependency. Timeouts and backpressure in the review pipeline caused sustained queue growth and incomplete reviews. Mitigation involved temporarily bypassing fix suggestions to reduce latency, adding worker capacity to clear the backlog, and deploying a model configuration that lowered end-to-end latency. Success rates recovered and stayed stable through peak traffic.
Post-incident work includes raising baseline worker capacity, adding instrumentation for queue health and worker utilization, and improving automatic load-shedding and fallback behavior to shorten time to detection and resolution.
Actions Runner Timeouts in West US
Intermittent timeouts affected some GitHub Actions runners on December 18, causing failures during runner setup and workflow execution. Network packet loss between runners in the West US region and a GitHub edge site was the cause. Roughly 1.5% of jobs on larger and standard hosted runners in that region (0.28% of all Actions jobs) were impacted between 08:15 UTC and 17:11 UTC.
Mitigation was achieved by routing all traffic away from the affected edge site by 17:11 UTC. GitHub is working on earlier detection of cross-cloud connectivity problems and faster paths to mitigation.
Copilot Policy Updates Blocked
Users, organizations, and enterprises could not update Copilot policies on December 18 between 16:25 UTC and 19:09 UTC. The service backing Copilot policies was degraded due to a database migration that caused schema drift. No other GitHub services, including other Copilot components, were affected.
The incident was resolved by synchronizing the schema. GitHub has hardened the service to prevent schema drift from causing future issues and is exploring deployment pipeline improvements to speed up mitigation.
Unauthenticated Traffic Spike
Unauthenticated requests to GitHub.com were degraded on December 22, between 22:01 UTC and 22:32 UTC, with slow or timed out page loads and API requests. Unauthenticated requests from Actions jobs, such as release downloads, were also affected. Authenticated traffic was unaffected. A severe spike in traffic, primarily targeting search endpoints, caused the degradation.
GitHub’s immediate response focused on pinpointing the source of the traffic increase; automated traffic management and targeted mitigation restored service. The company improved limiters for high-load endpoints and is working to better identify large traffic volume changes, strengthen resilience in critical request flows, and reduce time to mitigation.
