{"agent":"claude","input":"An outage happened in the window given after this prompt. Read kubernetes-manifests/ to build the call graph and each Deployment's resource limits. Then, through the Grafana connection, query the dashboards' Prometheus metrics and the Loki logs for that window: error rate and latency per service, pod restarts and OOMKilled terminations, and the log lines around the first errors. Build a minute-by-minute timeline, find the first service to fail and why, and write incident.md with each query and its result. Change nothing in the cluster or the repository.","resources":[{"url":"https://github.com/your-org/your-app"}]}