k8s-incident-triage

To diagnose Kubernetes pods in CrashLoopBackOff or Pending states, this Skill performs read-only triage using kubectl to inspect cluster state, analyze events, and review logs. It pinpoints workload issues and forms an initial hypothesis about the root cause.

Yes, you can automate Kubernetes incident response because this Skill accepts structured JSON input alongside natural language queries. This multi-modal input enables flexible integration into automated workflows to rapidly gather diagnostic insights during downtime.

You can check node pressure and workload health by invoking this Skill to perform comprehensive kubectl diagnostics across namespaces. It safely inspects cluster state, covering pod states, events, logs, and node conditions to identify pressure issues.

Kubernetes triage with this Skill does not require write permissions and never modifies cluster resources. It operates in a strictly read-only manner, safely inspecting workloads, pods, and namespaces using kubectl commands without applying any changes.

Before diagnosing Kubernetes incidents, you need the kubectl binary installed and optional environment variables configured for cluster access. This setup allows the Skill to connect to your cluster and execute read-only triage commands effectively.

Your Kubernetes pods may be failing due to node pressure or configuration errors, and this Skill provides actionable next steps by analyzing recent events and relevant logs. It delivers rapid insights into root causes to guide your incident response.