Managing Kubernetes Clusters Using K8s MCP Server

Model Context Protocol (MCP) is a method for organizing context for large language models, allowing them to act intelligently and maintain continuity across tasks, tools, and sessions. This guide demonstrates how to set up a Kubernetes MCP server container to interact with a Kubernetes cluster using natural language and connect it to Claude Desktop for conversational cluster management.

To begin, ensure you have a Kubernetes cluster deployed and have installed kubectl, downloaded your cluster's kubeconfig file, downloaded Claude Desktop, and Docker Desktop. Start by displaying all available contexts defined in the kubeconfig file using `kubectl config get-contexts`. Next, pull the sample Kubernetes MCP server Docker container image with `docker pull ghcr.io/alexei-led/k8s-mcp-server:latest`.

Run the Docker container using the context value from the previous step. Verify the image is running in Docker Desktop. MCP Server tools include built-in support for multiple Kubernetes tools like kubectl, helm, istioctl, and argocd, all within a single containerized environment. It also supports cloud-native integration with major managed Kubernetes providers and offers secure by design features with non-root user privileges and strict command validation.

Command chaining is supported for complex workflows, and the server configuration is easily customizable through environment variables. To use Claude Desktop for interacting with the MCP server, start the application, locate the settings option, select Developer settings, click Edit Config, and add a configuration snippet to your Claude Desktop config file.

Replace `YOUR_USER_NAME` with your actual username and update the mount path to point to your real `.kube` directory. Restart Claude Desktop, confirm the server is running, and ensure the MCP server tools are available. With Claude Desktop, you can now manage your Kubernetes cluster using natural language. Supported operations include advanced tasks such as checking Istio service mesh status, creating ArgoCD applications, deploying Helm charts, setting up Ingress controllers, and troubleshooting issues.

Basic commands like listing all pods, services, and fetching pod logs are also supported. For further customization and extended functionality, consider extending the MCP server configuration to manage multiple cluster contexts and pairing it with ArgoCD and Istio tooling already included in the container.