Source: wshobson/agents · Maintainer: community: Seth Hobson · License: MIT · Checked: 24 October 2026
What is the Kubernetes manifest generator skill?
The k8s-manifest-generator skill guides a coding agent through writing production Kubernetes manifests: Deployments, Services, ConfigMaps, Secrets and PersistentVolumeClaims, with resource limits, health checks, security contexts and standard labels set from the start. It is for developers who ask Claude Code, Codex or Cursor to containerize a service for Kubernetes and want YAML that follows conventions rather than a minimal example.
It is part of the kubernetes-operations plugin in wshobson/agents, Seth Hobson's catalog of plugins, agents and skills for several coding tools. The skill's description reads: "Create production-ready Kubernetes manifests for Deployments, Services, ConfigMaps, and Secrets following best practices and security standards. Use when generating Kubernetes YAML manifests, creating K8s resources, or implementing production-grade Kubernetes configurations."
As of 24 October 2026, its skills.sh page listed 11.2K installs (snapshot, 24 October 2026).
What does it do?
The agent loads the skill when you ask for Kubernetes YAML or K8s resources. The SKILL.md is a short guide: when to use it, ten best practices, troubleshooting commands and next steps. The detail is in six supporting files.
| File | What it holds |
|---|---|
references/details.md |
A ten-step workflow from gathering requirements to validation, plus common patterns such as a stateless web app |
references/deployment-spec.md |
Deployment fields, rollout strategies, probe types, affinity and spread, sidecar and init container patterns, a production checklist |
references/service-spec.md |
ClusterIP, NodePort, LoadBalancer, ExternalName and headless Services, ports, session affinity, traffic policies, discovery |
assets/deployment-template.yaml |
A Deployment with standard labels, a zero-downtime rollout, startup, liveness and readiness probes, and a non-root security context |
assets/service-template.yaml |
Seven Service templates, from internal ClusterIP to multi-port with metrics |
assets/configmap-template.yaml |
Seven ConfigMap templates, including file, JSON, per-environment and Prometheus configs |
The workflow starts with questions, which is useful in an agent session: the app's name and purpose, image and tag, storage needs, exposed ports, secrets and config files, CPU and memory, and whether it is reachable from outside the cluster.
The ten best practices in the SKILL.md are: set resource requests and limits, add health checks, use specific image tags, apply security contexts, keep config in ConfigMaps and Secrets, label everything, follow naming conventions, validate before applying, keep manifests in Git, and document with annotations. The deployment template uses the app.kubernetes.io/* labels from the Kubernetes recommended labels and sets maxUnavailable: 0 for rollouts.
How do you install it?
The repo README documents a skills-only route and plugin routes. Pick one per tool so you don't load the skill twice.
Just this skill. The command from skills.sh, using the skills CLI:
npx skills add https://github.com/wshobson/agents --skill k8s-manifest-generator
Or GitHub CLI 2.90 or later, as listed in the README:
gh skill install wshobson/agents k8s-manifest-generator
Both copy only the skill folder, under its bare name: .claude/skills/k8s-manifest-generator/ in a project, or ~/.claude/skills/k8s-manifest-generator/ for all projects, per the Claude Code skills docs. Codex and Cursor read the same SKILL.md format; see Codex skills and Cursor skills.
The kubernetes-operations plugin in Claude Code. Inside a session:
/plugin marketplace add wshobson/agents
/plugin install kubernetes-operations@claude-code-workflows
The plugin includes four skills (gitops-workflow, helm-chart-scaffolding, k8s-manifest-generator, k8s-security-policies) and a kubernetes-architect agent. This skill then runs as /kubernetes-operations:k8s-manifest-generator.
Codex plugin.
codex plugin marketplace add wshobson/agents
codex plugin add kubernetes-operations@claude-code-workflows
claude.ai. Under the MIT License you can zip the folder and upload it in Customize > Skills, as described in Anthropic's skills help article.
How do you use it?
Give the agent the facts it would otherwise ask for: image, ports, resources, config and how the service is exposed.
Example prompts (suggestions, not quotes from the repo):
Write Kubernetes manifests for our API: image ghcr.io/acme/api:1.4.2, port 8080, 3 replicas, /healthz for probes, 250m CPU and 256Mi memory requests.Add a ConfigMap for config.yaml and mount it at /etc/app, and a Secret template for DATABASE_URL with no real value.Our Service has no endpoints. Check the selector against the Deployment labels and fix it.Convert this docker-compose.yml into a Deployment, Service and PersistentVolumeClaim per service, in one multi-document file.Review k8s/ against the skill's production checklist and list what's missing.
Requirements: kubectl if you want the agent to run the validation or troubleshooting commands, and access to a cluster for server-side dry runs. Writing the YAML needs nothing extra.
Tips and limits
- Secrets need more than a manifest. The Kubernetes Secrets docs say Secrets "are, by default, stored unencrypted in the API server's underlying data store (etcd)." The skill's reference suggests Sealed Secrets, External Secrets Operator or Vault. Keep real values out of the files the agent writes.
- One validator is unmaintained. The validation step lists
kubevalalongsidekube-scoreandkube-linter. The kubeval README says the project is no longer maintained and points to kubeconform as a replacement.kubectl apply --dry-run=serverworks without extra tools. - Plain YAML only. For templating, the skill points to Helm or Kustomize; the plugin's
helm-chart-scaffoldingskill covers Helm. - Commands touch your cluster. The troubleshooting section uses
kubectl describe,kubectl get eventsand a temporarybusyboxdebug pod. Check which context your kubeconfig points to before letting the agent run them. - Security audits. On 24 October 2026 the skills.sh page showed Pass from Gen Agent Trust Hub, Socket and Snyk.
- License. MIT, from the repo's LICENSE (copyright 2024 Seth Hobson).
Save it in your Kipt library
Kipt stores your prompts and skills as Markdown in a private GitHub repo you own. Open Skills and paste this skill's GitHub folder link into the import box; Kipt brings in the SKILL.md with its references and templates. The copy lives in your own repo, so you can change the template defaults there (registry, label values, resource sizes). Install your whole library into Claude Code or Codex with npx skills add https://github.com/<you>/<repo>.git --all -g. Kipt is not affiliated with Seth Hobson or wshobson/agents.
Similar skills
- Docker skill: multi-stage Dockerfiles for the images your Deployments run.
- GitHub Actions skill: workflow templates that build the image and apply these manifests.
- Terraform skill: HashiCorp's style guide skill for the cluster and cloud resources underneath.
- Security and hardening skill: general security rules for the code inside the containers.