kubectl

kubectl

Use this skill to keep Kubernetes resource management deterministic and safe across clusters and namespaces.

Intent Router

Request	Reference	Load When
Install, kubeconfig setup, contexts, env vars	resources/install-and-setup.md	User needs to install kubectl or configure KUBECONFIG/contexts
Core commands: get/describe/apply/delete/logs/exec	resources/command-cookbook.md	User needs daily kubectl operations
Resources, labels, namespaces, explain	resources/resource-management.md	User asks about resource types, labels, selectors, or declarative workflows
Pod crashes, ImagePull errors, events, debug	resources/troubleshooting.md	User encounters CrashLoopBackOff, OOMKilled, or needs cluster diagnostics

Quick Start

# Verify connectivity
kubectl cluster-info
kubectl get nodes

Confirm active context and namespace

kubectl config current-context kubectl config view --minify

List pods in all namespaces

kubectl get pods -A

Apply a manifest

kubectl apply -f deployment.yaml --dry-run=client kubectl apply -f deployment.yaml

## Core Command Tracks

- **Read state:** `kubectl get/describe <resource>` — inspect current cluster state
- **Apply changes:** `kubectl apply -f` — declarative reconciliation
- **Logs:** `kubectl logs <pod> [-c <container>] [-f] [--previous]`
- **Shell:** `kubectl exec -it <pod> -- bash`
- **Port forward:** `kubectl port-forward <pod> 8080:80`
- **Rollout:** `kubectl rollout status/history/undo deployment/<name>`
- **Scale:** `kubectl scale deployment/<name> --replicas=3`
- **Output formats:** `-o wide`, `-o json`, `-o yaml`, `-o jsonpath`

## Safety Guardrails

- Always confirm active context (`kubectl config current-context`) before applying changes to prevent applying to the wrong cluster.
- Always specify namespace explicitly with `-n <namespace>` for production operations; never rely on default namespace alone.
- Use `--dry-run=client -o yaml` to preview resources before applying.
- Use `kubectl diff -f <file>` to see what will change before applying.
- Confirm before running `kubectl delete` — especially with label selectors (`-l`) which can match many resources.
- Never run `kubectl delete namespace <name>` without explicit user confirmation; it deletes all resources in the namespace.
- Prefer `kubectl rollout undo` over direct manifest rewrites for reverting deployments.

```bash
# Troubleshoot a crashing pod: check recent events then inspect previous container logs
kubectl get events --sort-by=.lastTimestamp -n my-namespace
kubectl logs my-pod --previous -n my-namespace

Workflow

1. Confirm context: kubectl config current-context and kubectl config get-contexts.
2. Inspect current state with kubectl get and kubectl describe before making changes.
3. Preview changes with kubectl diff -f <file> or --dry-run=client.
4. Apply changes with kubectl apply -f <file>.
5. Monitor rollout with kubectl rollout status deployment/<name>.
6. On failure, check events with kubectl get events --sort-by=.lastTimestamp.

Related Skills

• helm — package manager for Kubernetes applications
• docker — container image building for Kubernetes workloads

References

• resources/install-and-setup.md
• resources/command-cookbook.md
• resources/resource-management.md
• resources/troubleshooting.md
• Official docs: https://kubernetes.io/docs/reference/kubectl/
• Command reference: https://kubernetes.io/docs/reference/generated/kubectl/kubectl-commands