CloudDrove Claude Skills

A curated collection of DevOps skills for Claude Code — the CLI tool by Anthropic. These skills extend Claude with deep domain knowledge for Kubernetes, Terraform, AWS, and other infrastructure tools.

What are Skills?

Skills are structured knowledge packs that Claude Code loads on demand. When you ask Claude a Kubernetes question, the k8s skill automatically activates — giving Claude access to troubleshooting workflows, manifest patterns, Helm guides, security references, and diagnostic scripts without you having to provide any context.

Available Plugins

Plugin	Description	Status
k8s-skills	Kubernetes operations, troubleshooting, and platform engineering	✅ Available
docker-skills	Docker — Dockerfile, Compose, registry, image optimization	✅ Available
terraform-skills	Terraform modules, state management, and IaC patterns	🚧 Planned
aws-skills	AWS services, architecture, and operations	🚧 Planned

Quick Start

Prerequisites

• Claude Code CLI installed and configured

Installation

# 1. Add the CloudDrove marketplace (one-time)
/plugin marketplace add clouddrove/claude-skills

# 2. Install the plugins you need
/plugin install k8s-skills@clouddrove-claude-skills

Usage

Skills trigger automatically based on context. Just ask naturally:

> My pod is stuck in CrashLoopBackOff, how do I debug it?
> Write a production-ready Deployment with HPA and PDB
> Which apps are failing in production?
> Compare resource usage between staging and production
> Is everything healthy in the cluster?

Or use slash commands for direct access:

/k8s-skills:k8s-health                          # Cluster health check
/k8s-skills:k8s-health production               # Namespace health check
/k8s-skills:k8s-debug my-pod production          # Debug a specific pod
/k8s-skills:k8s-deploy rollback myapp production  # Rollback a deployment

k8s-skills

The Kubernetes skill covers both day-to-day operations and platform engineering:

Area	What's Included
Troubleshooting	Decision trees for CrashLoopBackOff, OOMKilled, ImagePullBackOff, Pending, networking, storage
Manifests	Production-ready YAML for Deployments, Services, Ingress, HPA, PDB, NetworkPolicy, and more
Helm	Chart management, values handling, Helmfile, chart authoring, testing
Security	RBAC patterns, Pod Security Standards, network policies, secrets management
Monitoring	Prometheus alerting rules, ServiceMonitor patterns, Grafana dashboards, USE/RED methods
GitOps	ArgoCD and Flux — Application CRDs, sync policies, app-of-apps, image automation, secrets
Commands	/k8s-health, /k8s-debug, /k8s-deploy — direct access slash commands
Scripts	diagnose.sh, cluster-health.sh, rbac-audit.sh, namespace-setup.sh
Examples	Complete Helm chart, multi-environment Helmfile (dev/staging/prod)

Directory structure

plugins/k8s-skills/
├── commands/
│   ├── k8s-debug.md              # /k8s-debug — pod/deployment diagnosis
│   ├── k8s-deploy.md             # /k8s-deploy — deploy, rollback, restart
│   └── k8s-health.md             # /k8s-health — cluster/namespace health check
├── skills/k8s/
│   ├── SKILL.md                  # Core skill — command reference, troubleshooting tree, workflows
│   ├── references/
│   │   ├── troubleshooting.md    # Detailed debugging workflows for every error state
│   │   ├── manifests.md          # Production-ready YAML templates
│   │   ├── security.md           # RBAC, Pod Security Standards, network policies
│   │   ├── monitoring.md         # Prometheus, Grafana, alerting rules
│   │   ├── helm.md               # Helm operations, chart authoring, Helmfile
│   │   └── gitops.md             # ArgoCD and Flux — sync policies, app-of-apps, secrets
│   ├── scripts/
│   │   ├── diagnose.sh           # Pod diagnostic tool
│   │   ├── cluster-health.sh     # Cluster health overview
│   │   ├── rbac-audit.sh         # RBAC permissions audit
│   │   └── namespace-setup.sh    # Production namespace generator
│   └── examples/
│       ├── helm-chart/           # Complete production Helm chart
│       └── helmfile/             # Multi-environment Helmfile setup
└── README.md

Contributing

Contributions are welcome! Whether it's improving existing skills or adding new ones.

Adding a New Skill

1. Fork and clone this repository
2. Create your skill directory under plugins/<plugin-name>/skills/<skill-name>/
3. Write a SKILL.md with YAML frontmatter:

   ---
   name: my-skill
   description: "When to trigger this skill and what it does..."
   ---
   
   # Skill instructions here

4. Add reference docs in references/ and scripts in scripts/ as needed
5. Register the plugin in .claude-plugin/marketplace.json
6. Submit a pull request

Improving an Existing Skill

• Fix inaccuracies, add missing patterns, improve troubleshooting workflows
• Add new reference docs for uncovered topics
• Improve script diagnostics and output

Guidelines

• Keep SKILL.md lean (under 500 lines) — move detailed content to references/
• Write in imperative/third-person style for AI consumption
• Include working examples, not pseudo-code
• Scripts should support --help and work as black-box tools

How It Works

Skills use a three-level progressive loading system:

1. Metadata — Skill name and description are always loaded (~100 words). Claude uses this to decide when to activate.
2. SKILL.md — Core instructions loaded when the skill triggers. Contains quick references, decision trees, and pointers to deeper docs.
3. References — Detailed guides loaded on demand. Only pulled into context when needed for a specific task.

This keeps Claude's context efficient while making deep knowledge available when needed.

License

This project is licensed under the Apache License 2.0 — see the LICENSE file for details.

About CloudDrove

CloudDrove is a DevOps consultancy helping teams build and scale cloud infrastructure. We build open-source tools for the community.

• Website
• GitHub
• Terraform Modules