What does this skill do?

The Kubernetes Production Patterns Skill provides production-grade YAML templates and troubleshooting commands for deploying, managing, and scaling workloads on Kubernetes. It includes optimized configurations that avoid common anti-patterns—such as using the :latest tag or running as root—and ensures high availability through probes, resource limits, and PodDisruptionBudgets.

Production Deployments
Create deployment manifests with rolling updates, security context, and resource limits.
Least Privilege RBAC
Configure ServiceAccounts, Roles, and RoleBindings by applying the principle of least privilege.
Probes and Health
Define `startupProbe`, `livenessProbe`, and `readinessProbe` with separate endpoints to accommodate slow startup times.
Troubleshooting with kubectl
Diagnose pods experiencing CrashLoopBackOff, OOMKilled, or image pull errors using the exact commands.

Usage examples

🚀 Deployment Base
Generate a production deployment for a web API with 3 replicas, probes, and resource limits.
🔒 Secure RBAC
Create a ServiceAccount and a least-privilege Role that only allows reading ConfigMaps in my-namespace.
🩺 Pod debugging
My pod is in a CrashLoopBackOff state. Please help me diagnose the issue using kubectl and check the container logs.
📊 Scaling with HPA
Set up a Horizontal Pod Autoscaler for my Deployment that scales between 2 and 10 replicas based on CPU usage.

Features

Robust Deployment Template It includes rolling update, securityContext, runAsNonRoot, and readOnlyRootFilesystem by default.
Separate probes Configure `startupProbe` for slow startups, `livenessProbe`, and `readinessProbe` with different endpoints.
Resource Management Define CPU and memory requests and limits tailored to the type of workload (web API, worker, JVM).
Least-Privilege RBAC Disable token mounting if the app does not use the K8s API, and apply specific roles if it does.
kubectl Debug Sheet Exact commands for inspecting logs and events and diagnosing CrashLoopBackOff, OOMKilled, and pull errors.

Frequently asked questions

No. The skill generates YAML manifests and kubectl commands, but it does not connect to any cluster. You just need kubectl installed to run the generated commands.
Yes. The templates avoid using the :latest tag, running as root, skipping probes, or failing to define resource limits, in accordance with production best practices.
Of course. The templates are a starting point. Claude tailors responses, images, resources, and prompts based on the type of prompt you provide.
Yes. The skill handles scaling using Horizontal Pod Autoscaler and PodDisruptionBudgets to ensure availability during maintenance.
Kubernetes Production Patterns — Robust Deployments with Claude AI

¿Prefieres escuchar el contenido? Genera la narración de audio con un clic.