Manifest Maker
Describe your app, get production-ready K8s YAML
Tell it what you're deploying and it generates complete, production-ready Kubernetes manifests — Deployment, Service, and, when appropriate, Ingress, HPA, PDB, and NetworkPolicy — with sane resource requests, health checks, security contexts, and labels. No more copy-pasting from old deployments and forgetting the readiness probe.
INGREDIENTS
PROMPT
Create a skill called "Manifest Maker". When I describe an application (what it does, language/runtime, port, expected traffic, dependencies), generate production-ready Kubernetes manifests. Always include: - Deployment with sensible resource requests/limits - Liveness/readiness/startup probes where applicable - Security context (non-root, minimal privileges, read-only fs where practical) - Service with correct port and protocol - Standard labels (`app.kubernetes.io/*`) Include only when appropriate: - Ingress - HorizontalPodAutoscaler - PodDisruptionBudget - NetworkPolicy - ConfigMap and Secret references Validate the output with `kubectl apply --dry-run=client` if kubectl is available. Output as a single multi-document YAML file.
How It Works
Describe your application in plain English — what it does, what port it
listens on, how much traffic it gets — and your Claw generates complete
K8s manifests following production best practices.
What You Get
- Core manifests: Deployment and Service
- Optional manifests when they fit: Ingress, HPA, PDB, NetworkPolicy, ConfigMap, Secret references
- Proper resource requests and limits based on your description
- Health check probes (liveness, readiness, startup) configured correctly
- Security contexts (non-root, read-only filesystem where possible, dropped capabilities)
- Labels and annotations following K8s conventions
Setup Steps
- Describe your application to your Claw (language, port, expected traffic, dependencies)
- Review the generated manifests
- Run `kubectl apply --dry-run=client -f` to validate
- Adjust and iterate as needed
Tips
- Be specific about your app's needs: port, protocol, expected memory/CPU, dependencies
- Ask for Kustomize or Helm output if you need environment-specific variations
- Request NetworkPolicy if you want pod-to-pod traffic restrictions
- HPA, PDB, and Ingress should be included only when they make sense for the workload
- Can also convert existing manifests to Helm charts or Kustomize overlays