Getting Started with DevOps-OS MCP Server#

Welcome! DevOps-OS MCP Server is the Model Context Protocol implementation that connects AI assistants (Claude, ChatGPT, etc.) with DevOps-OS generators.

📌 About the Two Projects

• DevOps-OS MCP (this repo): MCP server for AI assistant integration
• DevOps-OS CLI (see cloudengine-labs/devops_os): Original command-line tool

Both share the same core scaffold modules. Choose based on your use case:

• Use MCP if you want to integrate with Claude, ChatGPT, Cursor, or other AI assistants
• Use CLI if you want to run generators in scripts, pipelines, or local automation

What is DevOps-OS MCP?#

DevOps-OS MCP is a toolkit that generates production-ready CI/CD pipelines, Kubernetes manifests, infrastructure hardening baselines, and SRE monitoring configs — through conversational AI using the MCP server.

Ask Claude or ChatGPT:

"Generate a complete GitHub Actions CI/CD workflow for a Python Flask API with Docker build, pytest tests, and deployment to Kubernetes."

You get production-ready YAML in seconds.

👋 New to DevOps-OS? Start Here#

Not familiar with DevOps terms? Read the Beginner’s Guide first — it explains concepts in plain English with real-world analogies.

Prerequisites#

Quick Start: 5 Minutes to First Generated Config#

Step 1: Clone and Install#

git clone https://github.com/chefgs/devops_os_mcp.git
cd devops_os_mcp

# Set up Python environment
python3 -m venv .venv
source .venv/bin/activate  # macOS / Linux
# .venv\Scripts\activate   # Windows (cmd)

# Install MCP server dependencies
pip install -r mcp_server/requirements.txt

Step 2: Configure Claude Desktop (or ChatGPT)#

For Claude Desktop:

1. Get your repo path: pwd (copy output)
2. Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)
3. Add this config:

{
  "mcpServers": {
    "devops-os": {
      "command": "python",
      "args": ["-m", "mcp_server.server"],
      "cwd": "/your/repo/path/here"
    }
  }
}

4. Restart Claude Desktop. You’ll see a wrench icon 🔧 at the bottom right.

For ChatGPT or other MCP clients:

See MCP Setup & Configuration for detailed instructions.

Step 3: Generate Your First Config#

Ask Claude:

Generate a GitHub Actions workflow for a Python Flask API with:
- Lint and test stages (using pytest)
- Docker build and push to Docker Hub
- Deployment to Kubernetes

Done! Claude will generate the complete workflow YAML.

🎯 Example Prompts to Try#

Once connected, try these prompts:

Learn More About DevOps-OS Philosophy#

Process-First SDLC Philosophy →

Understand why each generator exists and how it maps to software development best practices.

See the Process-First guide for the full reference.

3 — Generate your first CI/CD pipeline#

All scaffold generators are subcommands of the unified devopsos scaffold command. Use --help on any subcommand to see all options.

GitHub Actions#

python -m cli.devopsos scaffold gha --name my-app --languages python,javascript --type complete

Output: .github/workflows/my-app-complete.yml

GitLab CI#

python -m cli.devopsos scaffold gitlab --name my-app --languages python --type complete

Output: .gitlab-ci.yml

Jenkins#

python -m cli.devopsos scaffold jenkins --name my-app --languages java --type complete

Output: Jenkinsfile

4 — Generate Kubernetes / GitOps configs#

# ArgoCD Application CR + AppProject
python -m cli.devopsos scaffold argocd --name my-app \
       --repo https://github.com/myorg/my-app.git \
       --namespace production
# Output: argocd/application.yaml + argocd/appproject.yaml

# Flux CD configs
python -m cli.devopsos scaffold argocd --name my-app --method flux \
       --repo https://github.com/myorg/my-app.git
# Output: flux/ directory

5 — Generate infrastructure hardening baselines#

# Kubernetes policy baselines
python -m cli.devopsos scaffold hardening --standard cis-k8s --type kyverno --environment production

# Operating system compliance profiles
python -m cli.devopsos scaffold hardening --standard cis-rhel9 --type inspec

Output: hardening/ directory containing Kyverno policies, InSpec profiles, Checkov checks, and a compliance mapping file.

See Infrastructure Hardening for supported standards and validation examples.

6 — Generate SRE configs#

python -m cli.devopsos scaffold sre --name my-app --team platform

Output: sre/ directory containing:

• alert-rules.yaml — Prometheus PrometheusRule CR
• grafana-dashboard.json — Grafana importable dashboard
• slo.yaml — Sloth-compatible SLO manifest
• alertmanager-config.yaml — Alertmanager routing stub

7 — Generate unit test configs#

# Python — generates pytest.ini, conftest.py, and a sample test file
python -m cli.devopsos scaffold unittest --name my-app --languages python

# JavaScript with Jest
python -m cli.devopsos scaffold unittest --name my-app --languages javascript --framework jest

# TypeScript with Vitest
python -m cli.devopsos scaffold unittest --name my-app --languages typescript --framework vitest

# Multi-stack (Python + JavaScript + Go) in one command
python -m cli.devopsos scaffold unittest --name my-platform --languages python,javascript,go

See CLI Reference for all options and output file paths.

8 — Interactive wizard (all-in-one)#

python -m cli.devopsos init              # interactive project configurator
python -m cli.devopsos scaffold gha      # scaffold GitHub Actions
python -m cli.devopsos scaffold gitlab   # scaffold GitLab CI
python -m cli.devopsos scaffold jenkins  # scaffold Jenkins
python -m cli.devopsos scaffold argocd   # scaffold ArgoCD / Flux
python -m cli.devopsos scaffold hardening # scaffold hardening baselines
python -m cli.devopsos scaffold sre      # scaffold SRE configs
python -m cli.devopsos scaffold cicd     # scaffold GHA + Jenkins in one step
python -m cli.devopsos scaffold unittest # scaffold unit test configs

9 — Use with an AI assistant#

Via MCP (Recommended — Native Integration)#

Use DevOps-OS as an MCP server in Claude Desktop or ChatGPT for seamless integration.

For Claude Desktop (5-minute setup):

pip install -r mcp_server/requirements.txt

Then see MCP Quick Start for step-by-step instructions.

For ChatGPT, HTTP endpoints, and production deployment: See MCP Setup & Configuration for detailed setup guide.

Via API (Alternative — Direct API Calls)#

Load tools from skills/ directory and call Claude or OpenAI APIs directly:

10 — Next steps#