Kubernetes MCP Server

A powerful MCP (Model Context Protocol) server for Kubernetes operations that allows you to manage your Kubernetes clusters using natural language through AI assistants.

🌟 Features

• � Resource Listing

  • List pods, services, deployments, and namespaces
  • Get detailed resource information

• 🔍 Monitoring & Debugging

  • View pod logs with customizable line counts
  • Describe pod details for troubleshooting
  • Get cluster events for debugging

• ⚡ Operations

  • Port-forward services to local ports
  • Scale deployments up or down
  • Execute kubectl commands through natural language

• 🎯 Developer Focused

  • Designed for common Kubernetes developer workflows
  • Easy to extend with additional tools
  • Comprehensive error handling

� Available Tools

Tool	Description	Parameters
list-pods	List pods in a namespace	namespace (optional)
list-services	List services in a namespace	namespace (optional)
list-deployments	List deployments in a namespace	namespace (optional)
list-namespaces	List all namespaces	None
describe-pod	Get detailed pod information	pod (required), namespace (optional)
get-logs	Get pod logs	pod (required), namespace (optional), lines (optional), follow (optional)
port-forward	Port forward a service	service, localPort, targetPort (required), namespace (optional)
scale-deployment	Scale a deployment	deployment, replicas (required), namespace (optional)
get-events	Get cluster events	namespace (optional)

📋 Requirements

• Node.js v18+
• kubectl installed and configured to access your cluster
• @modelcontextprotocol/cli (optional, for testing)

# Verify kubectl works
kubectl get pods

# Install MCP CLI (optional)
npm install -g @modelcontextprotocol/cli

🔧 Installation

# Clone or download this repository
cd kube-mcp

# Install dependencies
npm install

# Run tests to verify setup
node test.js

🏃 Running the Server

# Start the MCP server
npm start

# Or run directly
node server.js

The server runs on stdio transport and will output: Kubernetes MCP server running on stdio

🔌 Connecting to MCP Clients

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "kubernetes": {
      "command": "node",
      "args": ["/absolute/path/to/kube-mcp/server.js"]
    }
  }
}

Other MCP Clients

For clients that support MCP, configure them to connect to this server using:

• Command: node
• Args: ["/path/to/server.js"]
• Transport: stdio

💡 Example Usage

Once connected to an MCP-compatible client, you can use natural language like:

• "List all pods in the default namespace"
• "Show me the logs for the nginx pod"
• "Port forward the web service on port 8080 to my local port 3000"
• "Scale the api deployment to 3 replicas"
• "What events happened recently in the kube-system namespace?"

🧪 Testing

Run the test script to verify everything is working:

node test.js

This will check:

• ✅ kubectl availability
• ✅ Server syntax validation
• ✅ Server startup capability
• ✅ Cluster connectivity

🛠️ Development

To add new tools:

1. Add the tool definition to the tools array in server.js
2. Add a corresponding case in the CallToolRequestSchema handler
3. Implement the kubectl command and response formatting
4. Test with node test.js

⚠️ Security Notes

• This server executes kubectl commands on your system
• Ensure your kubectl context is set to the correct cluster
• Be cautious when scaling deployments or port-forwarding in production
• Review all operations before execution

📜 License

MIT License - feel free to modify and distribute

🤝 Contributing

Contributions are welcome! Please feel free to submit issues and enhancement requests.

Restart your MCP-enabled client (Cursor AI, Claude Desktop, VS Code MCP extension, etc.)

Now you can issue natural commands like:

• “Show me all pods in namespace payments”
• “Port forward postgres service to local 5432”
• “Get logs from api-pod”