troubleshooting-kubernetes

Kubernetes pod crashes including CrashLoopBackOff occur when containers repeatedly fail to start. This Skill collects pod logs and events, identifies the root cause (memory limits, missing configs, image issues), and presents numbered remediation options you approve before applying—ensuring safe, confirmed fixes without auto-changes.

Service unavailability with empty endpoints typically stems from failed pod deployments or selector mismatches. This Skill gathers service, pod, and deployment state, diagnoses the underlying issue, and walks you through fix options—scaling resources, updating image pulls, or adjusting configs—before you confirm changes.

Stuck deployments (ImagePullBackOff, Pending status) require diagnosis before action to avoid cascading failures. This Skill interactively identifies the blocker, proposes 2–4 remediation paths with clear explanations, waits for your selection, applies your chosen fix, and verifies resolution.

Yes. This Skill handles multi-layer Kubernetes issues by gathering state from pods, services, deployments, and events simultaneously, matching symptoms to common failure modes (OOMKilled, ImagePullBackOff, endpoint gaps), and presenting unified remediation options across your cluster.

Yes. This Skill relies on kubectl to query cluster state (pods, services, deployments, events) and apply fixes. You must have kubectl configured and cluster permissions to read diagnostics and execute remediation steps you approve.

Interactive approval prevents unintended outages by letting you review root causes and fix options before changes deploy. This Skill diagnoses first, shows trade-offs, waits for your choice, then applies and verifies—keeping you in control of production changes.