You can interact with Amazon EKS clusters programmatically through the EKS MCP Server, a Model Context Protocol (MCP) server that exposes Kubernetes operations as callable tools. This lets you manage pods, nodes, and other cluster resources from MCP clients and AI assistants in a structured, repeatable way.

How to use

You connect to the EKS MCP Server from an MCP client by configuring a local or remote MCP endpoint that runs the server process. The server provides tools you can invoke to perform Kubernetes actions, such as listing pods or inspecting nodes. Start the server in your environment and reference it in your MCP client settings so the client can send tool requests and receive structured results.

How to install

Prerequisites and setup steps ensure you have the necessary software and access to your EKS environment.

# Prerequisites
# Node.js (v18.0.0 or higher)
# npm (comes with Node.js)
# kubectl
# AWS CLI

# Install Node.js from the official site or package manager
# Example for macOS using Homebrew
# brew install node

# Verify installations
node --version
npm --version
kubectl version --client
aws --version

Additional setup and configuration

Configure access to the EKS cluster by updating kubeconfig and ensuring your AWS credentials allow cluster interaction.

# Update kubeconfig for the EKS cluster
aws eks update-kubeconfig --region ap-northeast-2 --name Shared-cluster

# Verify access
kubectl config get-contexts
kubectl get nodes

Note: Replace the region and cluster name with your actual values.

## Operational notes

The server runs as a local process and listens for MCP protocol commands. You start the server, connect a client, and then issue tool requests. The server logs indicate it is listening on standard input/output when active.

Start the MCP server

node server.js

You should see a startup message indicating the server is listening on stdio

To stop, press Ctrl+C.

Troubleshooting

Common issues involve kubeconfig contexts, AWS permissions, or network reachability. Ensure the correct context is active and your AWS credentials permit cluster access.

# Check current context
kubectl config get-contexts

# If needed, switch contexts
kubectl config use-context eks-bastion

# Validate cluster access
kubectl get nodes

If access problems persist, re-run the kubeconfig update with the right cluster name and region.

## Demo scenarios

The following practical example demonstrates listing all pods across namespaces, using the available tool to gain visibility into the cluster state.

Example: list all pods across all namespaces

kubectl --context="eks-bastion" get pods --all-namespaces

For more detailed operations, extend your MCP client to invoke the list_pods tool and parse its JSON output.

Available tools

list_pods

Lists all pods across all namespaces in the cluster and returns a JSON array of pod identifiers.