End-to-end SRE incident flow on Kuberly clusters: drive the Platform MCP incident_diagnosis workflow to a cited, read-only root cause by tying together aws, cluster, grafana, and vpc_flow_logs signals, then synthesize a postmortem. Use when an incident needs structured diagnosis across multiple signal sources, when you must separate confirmed evidence from hypotheses, when mutating diagnostic steps must be reported not performed, or when a resolved incident needs a postmortem. Keywords: incident, outage, root cause, RCA, triage, postmortem, sev1/sev2, blast radius.
