Kubernetes

🚀 MCP Kubernetes Server

This is an MCP (Model Context Protocol) server that controls Kubernetes clusters through interactions with Large Language Models (LLMs).

📚 Overview

This client allows you to perform common Kubernetes operations through MCP tools. It encapsulates kubectl commands and provides a simple interface to manage Kubernetes resources. MCP enables seamless interaction between language models and Kubernetes operations.

❓ What is MCP?

MCP is a framework that enables language models to interact with external tools and services in a structured manner. It provides:

• A standardized method to expose functionality to language models
• Operation context management
• Tool discovery and documentation
• Type-safe interaction between language models and tools

💻 Usage Examples

• Create a deployment named nginx-app with the nginx:latest image, and configure 3 replicas in the production namespace.
• Update the deployment nginx-app to version 1.19 in the production namespace.
• Scale the deployment nginx-app to 5 replicas in the production namespace.
• Get the pods in the production namespace.
• Get all namespaces in the cluster.
• Get all nodes in the cluster.
• Get all services in the cluster.
• Get all deployments in the cluster.
• Get all jobs in the cluster.
• Get all cronjobs in the cluster.
• Get all statefulsets in the cluster.
• Get the current context.
• List all contexts.
• Switch to the context .
• Get the logs of the pod in the production namespace.
• Get the events in the production namespace.
• Add an annotation key1=value1 to the pod in the production namespace.
• Delete the annotation key1 from the pod in the production namespace.
• Add the label key1=value1 to the pod in the production namespace.
• Delete the label key1 from the pod in the production namespace.
• Expose the nginx-app deployment on port 80 in the production namespace.
• Forward the local port 8080 of the pod, deployment, and service named to the production namespace.
• Delete the pod, deployment, service, job, cronjob, statefulset, and daemonset named in the production namespace.

📋 Planned Features

• Create a cluster role.
• Delete a cluster role.
• Create a cluster role binding.
• Delete a cluster role binding.
• Create a namespace.
• Delete a namespace.
• Create a service account.
• Delete a service account.
• Create a role.
• Delete a role.
• Create a role binding.
• Delete a role binding.

🔌 LLM Integration

This MCP client is designed to integrate seamlessly with Large Language Models (LLMs). Functions are decorated with @mcp.tool(), making them accessible to LLMs through the MCP framework.

📝 Example LLM Prompts

LLMs can interact with your Kubernetes cluster through natural language. Here are some example prompts:

• "Create a new deployment named nginx with 3 replicas in the production namespace."
• "Scale the nginx-app deployment to 5 replicas."
• "Update the nginx-app image to version 1.19."

The LLM will interpret these natural language requests and call the appropriate MCP functions with the correct parameters.

🌟 Advantages of LLM Integration

1. Natural Language Interface: Manage Kubernetes resources using conversational language.
2. Reduced Command Complexity: No need to remember precise kubectl syntax.
3. Error Prevention: LLMs can validate input and provide useful error messages.
4. Context Awareness: LLMs can maintain context between operations.
5. Structured Interaction: MCP ensures type-safe and documented interaction between language models and tools.

📋 Prerequisites

• A configured Kubernetes cluster
• kubectl installed and configured
• Python and its package management tool (e.g., pip) installed

📦 Installation Instructions

Install via Smithery

To automatically install the Kubernetes server for Claude Desktop via Smithery:

npx -y @smithery/cli install @abhijeetka/mcp-k8s-server