Some Copilot Cloud Agent jobs not starting
This summary is created by Generative AI and may differ from the actual content.
Overview
GitHub Copilot cloud agent service experienced degradation on August 5, 2026, between 11:02 and 11:54 UTC. During this 52-minute period, 100% of newly submitted cloud agent jobs were delayed from starting. No jobs were lost, and the queued backlog was fully processed by 13:00 UTC. The incident was caused by an internal rate limit control that was enabled more broadly than intended, affecting unrelated coding agent traffic. The service recovered when the rate limit window expired, after which the control was tuned to prevent affecting unrelated traffic.
Impact
100% of newly submitted cloud agent jobs were delayed during the 52-minute incident window. No jobs were lost. Service degradation was limited to job start delays with backlog processing completed by 13:00 UTC.
Trigger
An internal rate limit control designed to protect service availability was enabled more broadly than intended, causing it to delay more traffic than expected.
Detection
Investigation began at 11:38 UTC following reports of impacted performance for GitHub services. Degradation was mitigated and monitoring confirmed stability by 12:01 UTC.
Resolution
The rate limit control was tuned to no longer affect unrelated coding agent traffic. The service recovered when the rate limit window expired. Ongoing work is planned to improve the control's scoping and monitoring and alerting to reduce time to detection and mitigation of similar issues.
Root Cause
An internal rate limit control intended to protect service availability was configured with overly broad scoping, causing it to delay unrelated coding agent traffic beyond its intended scope.