Skip to content
Conduit

skill

claude-skills-devops

AI-powered DevOps skills repository for Claude

anikievev/claude-skills-devopscommunitylistedunreviewed

Set up

Skills are installed by placing their folder where your agent loads skills.

The download comes from github.com/anikievev/claude-skills-devops itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 22, 2026

DevOps Claude Skills — AI-Powered Templates for Docker, Terraform, AWS CDK, Ansible, GitHub Actions & GitLab CI

Supercharge your DevOps workflow with Claude Code skills — slash commands that generate production-grade infrastructure code, hardened Dockerfiles, and battle-tested CI/CD pipelines in seconds.

License: MIT


What Are DevOps Claude Skills?

Claude Code supports skills — reusable prompt files you install once and invoke as slash commands (e.g. /docker, /terraform). Each skill in this repository teaches Claude exactly how to write expert-level DevOps code for a specific tool, with every best practice baked in from the start.

Instead of manually looking up whether your Dockerfile uses a non-root user, whether your Terraform variables are validated, or whether your GitHub Actions workflow pins SHA hashes — just invoke the skill and get production-ready output immediately.

These devops Claude skills cover:

  • Security hardening out of the box
  • Minimal image sizes and optimized build caching
  • Least-privilege IAM, encrypted storage, and secret management
  • Idempotent, lint-compliant, role-structured code
  • CI/CD pipelines with proper job separation, caching, and deployment gates

Available Skills

Slash CommandToolWhat Claude Writes
/dockerDockerMulti-stage Dockerfile + .dockerignore — minimal image, non-root user, pinned versions, healthcheck
/terraformTerraformModule structure (main, variables, outputs, versions, backend) — remote state, least-privilege IAM, validated variables
/aws-cdkAWS CDKConstruct-separated stacks — encryption at rest, grant* IAM, CDK Aspects, env-agnostic design
/ansibleAnsibleRole-based playbooks — idempotent tasks, handlers, Vault for secrets, block/rescue, ansible-lint compliant
/github-actionsGitHub ActionsSHA-pinned workflows — least-privilege permissions, concurrency groups, matrix builds, dependency caching
/gitlabGitLab CIStaged pipelines — YAML anchors, rules (not only/except), DAG via needs, artifact expiry, SAST included

Installation

Skills can be installed globally (available in every project) or per-project.

Global install (recommended)

# Clone this repository
git clone https://github.com/your-username/claude-skills-devops.git

# Install all skills globally
cp claude-skills-devops/skills/*.md ~/.claude/skills/

Per-project install

# Install a single skill into your project
mkdir -p .claude/skills
cp path/to/claude-skills-devops/skills/docker.md .claude/skills/

Verify installation

Open Claude Code in any project and run:

/help

You should see the installed skills listed under your available commands.


Usage

Navigate to any project and invoke the relevant skill. Claude will inspect your project context and generate code tailored to your stack.

Docker

/docker

Claude analyzes your project (language, entrypoint, ports) and generates:

  • A multi-stage Dockerfile with distroless or alpine runtime, non-root user, pinned base image, layer-cache-optimized dependency installation, and a HEALTHCHECK
  • A .dockerignore excluding dev dependencies, secrets, test files, and IDE config

Terraform

/terraform

Claude generates a complete module with main.tf, variables.tf, outputs.tf, versions.tf, and backend.tf — including remote state (S3 + DynamoDB), validated variables, sensitive output markers, and a mandatory tagging strategy.

AWS CDK

/aws-cdk

Claude writes environment-agnostic CDK stacks with construct separation, explicit RemovalPolicy, encryption at rest for all resources, grant*-based IAM, CDK Aspects for cross-cutting tagging, and CfnOutput for cross-stack references.

Ansible

/ansible

Claude scaffolds a full role structure with idempotent tasks, handlers for service restarts, Ansible Vault integration for secrets, block/rescue/always error handling, no_log on sensitive tasks, and ansible-lint–compliant FQCN module names.

GitHub Actions

/github-actions

Claude generates a CI/CD workflow with SHA-pinned action versions, permissions: {} at workflow level with per-job grants, concurrency groups to cancel stale runs, dependency caching keyed by lock file hash, matrix testing, and a protected deploy job gated on github.ref == 'refs/heads/main'.

GitLab CI

/gitlab

Claude writes a .gitlab-ci.yml with explicit stages, rules-based triggers (no deprecated only/except), YAML anchors for DRY job templates, needs-based DAG for parallel execution, lock-file–keyed caching, artifact expiry, environment-scoped deployment tracking, and included SAST/Secret Detection templates.


Why Use These Claude Skills for DevOps?

Without skills: You write a Dockerfile, realize it runs as root, add a user, then later notice the base image is :latest, then later find the build output isn't cached properly.

With /docker: Claude generates a hardened, optimized, production-ready Dockerfile in one shot — every best practice applied from the start.

These Claude Code skills for DevOps act as an expert reviewer, code generator, and best-practices enforcer rolled into a single slash command. Whether you're writing your first Terraform module or setting up a GitLab CI pipeline for a microservices platform, each skill ensures you don't miss the details that matter in production.


Contributing

Contributions are welcome. To add a new skill or improve an existing one:

  1. Fork the repository
  2. Create your skill file in skills/<tool-name>.md using the frontmatter format:
    ---
    name: skill-name
    description: One-line description shown in /help
    ---
    
    [Prompt content...]
    
  3. Focus the skill on writing code, not running commands
  4. Embed concrete best practices — be specific about what Claude should and should not do
  5. Open a pull request with a brief description of what the skill generates and why the best practices were chosen

License

MIT — see LICENSE.

↑↓ navigate openesc close