Skip to main content
Glama
dkeeno

AWS EKS MCP Server

by dkeeno

AWS EKS MCP Server

Comprehensive MCP server for creating and managing AWS EKS (Elastic Kubernetes Service) clusters using Terraform.

Features

Terraform Generation & Execution

  • generate_eks_terraform - Generate complete Terraform configs (VPC, EKS, node groups, IRSA)

  • eks_terraform_init - Initialize Terraform

  • eks_terraform_plan - Preview infrastructure changes

  • eks_terraform_apply - Create/update EKS cluster

  • eks_terraform_destroy - Destroy cluster and all resources

Kubectl Configuration

  • configure_kubectl - Automatically configure kubectl to access the cluster

Add-on Management

  • install_addon - Install common EKS add-ons:

    • AWS Load Balancer Controller

    • Metrics Server

    • Cluster Autoscaler

    • External DNS

    • Cert Manager

Cluster Management

  • upgrade_eks_cluster - Upgrade Kubernetes version

  • list_eks_clusters - List all EKS clusters

  • describe_eks_cluster - Get detailed cluster info

  • get_eks_node_groups - List and describe node groups

  • scale_node_group - Scale node groups up/down

Related MCP server: Kubernetes MCP Server

What Gets Created

When you generate and apply an EKS Terraform configuration, you get:

Networking

  • VPC with public and private subnets across 3 AZs

  • NAT Gateways for private subnet internet access

  • Internet Gateway for public subnets

  • Proper subnet tagging for EKS/ALB

EKS Cluster

  • EKS control plane with specified Kubernetes version

  • IRSA (IAM Roles for Service Accounts) enabled

  • Managed node group with auto-scaling

  • EBS CSI driver (optional)

  • Security groups configured

IAM

  • EKS cluster IAM role

  • Node group IAM role

  • IRSA OIDC provider

  • EBS CSI driver IAM role (if enabled)

  • Cluster Autoscaler IAM role (if enabled)

Prerequisites

  • Python 3.10+

  • Terraform >= 1.0

  • AWS CLI configured

  • kubectl installed

  • Helm (for add-on installation)

  • boto3 Python library

Installation

cd ~/.mcp-servers/eks-mcp-server
pip install -r requirements.txt

Configuration

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "eks": {
      "command": "python3",
      "args": ["/Users/youruser/.mcp-servers/eks-mcp-server/server.py"],
      "env": {
        "EKS_WORKSPACE": "/Users/youruser/eks-clusters"
      }
    }
  }
}

Usage Examples

Create a New EKS Cluster

User: "Generate Terraform for an EKS cluster named production-cluster in us-west-2 with Kubernetes 1.29"

User: "Use t3.large instances with 3 nodes minimum and 10 maximum"

User: "Enable EBS CSI driver and cluster autoscaler"

Claude: [generates complete Terraform configuration]

User: "Initialize and plan the Terraform"

Claude: [runs terraform init and plan]

User: "Apply the configuration"

Claude: [creates the entire EKS cluster]

User: "Configure kubectl for this cluster"

Claude: [updates kubeconfig]

Manage Existing Cluster

User: "List all EKS clusters in us-east-1"

User: "Describe the production-cluster"

User: "Show me the node groups"

User: "Scale the node group to 5 desired nodes"

User: "Install the AWS Load Balancer Controller"

User: "Install metrics-server"

Upgrade Cluster

User: "Upgrade production-cluster to Kubernetes 1.30"

Claude: [initiates rolling upgrade]

Generated Terraform Structure

~/eks-clusters/
└── my-cluster/
    ├── main.tf       # VPC, EKS, node groups, IRSA
    ├── variables.tf  # Configurable parameters
    └── outputs.tf    # Cluster endpoints, IDs, kubectl command

Terraform Modules Used

  • terraform-aws-modules/vpc/aws - Production-ready VPC

  • terraform-aws-modules/eks/aws - EKS cluster and node groups

  • terraform-aws-modules/iam/aws - IRSA IAM roles

Default Configuration

  • VPC CIDR: 10.0.0.0/16

  • Subnets: 3 public + 3 private (one per AZ)

  • Node Instance Type: t3.medium

  • Node Count: 2 desired, 1-4 scaling range

  • Disk Size: 20 GB

  • Kubernetes Version: 1.29

  • IRSA: Enabled

  • EBS CSI Driver: Enabled

Add-on Installation

The server can install these add-ons via Helm or kubectl:

AWS Load Balancer Controller

Enables ALB and NLB ingress for your applications.

Metrics Server

Required for Horizontal Pod Autoscaler (HPA) and kubectl top.

Cluster Autoscaler

Automatically scales node groups based on pod resource requests.

External DNS

Automatically creates Route53 DNS records for Ingresses.

Cert Manager

Automates TLS certificate management with Let's Encrypt.

Cluster Upgrade Process

When you upgrade a cluster:

  1. Control plane is upgraded first (rolling, zero-downtime)

  2. Add-ons may need updating to match new K8s version

  3. Node groups should be upgraded separately (after control plane)

  4. Update Terraform config and run eks_terraform_apply to upgrade nodes

Cost Considerations

Typical EKS cluster costs:

  • EKS Control Plane: $0.10/hour (~$73/month)

  • EC2 Nodes: Varies by instance type

    • t3.medium: ~$0.0416/hour (~$30/month per node)

    • t3.large: ~$0.0832/hour (~$60/month per node)

  • NAT Gateways: $0.045/hour per AZ (~$32/month each)

  • Data Transfer: Varies

Cost Optimization:

  • Use single NAT Gateway for dev environments

  • Enable cluster autoscaler to scale down during off-hours

  • Use Spot instances for fault-tolerant workloads

  • Right-size instance types based on actual usage

Security Best Practices

  • Enable cluster logging (CloudWatch)

  • Use private API endpoint for production

  • Implement Pod Security Standards

  • Enable secrets encryption using KMS

  • Use IRSA instead of instance profiles

  • Regular security updates (upgrade K8s version)

  • Network policies to restrict pod communication

  • Use AWS WAF with ALB

Troubleshooting

Terraform fails with "Error creating EKS Cluster"

  • Check AWS service quotas

  • Verify IAM permissions

  • Ensure region supports EKS

Node groups stuck in "CREATE_FAILED"

  • Check EC2 instance limits

  • Verify subnet has available IP addresses

  • Review node IAM role permissions

kubectl can't connect

  • Run configure_kubectl tool again

  • Verify AWS credentials are current

  • Check cluster endpoint is accessible

Add-on installation fails

  • Ensure Helm is installed

  • Verify kubectl is configured

  • Check cluster has internet access (for pulling images)

Environment Variables

  • EKS_WORKSPACE: Directory for Terraform projects (default: ~/eks-clusters)

Advanced Usage

Custom Node Group Configuration

User: "Generate EKS cluster with mixed instance types: t3.large and t3.xlarge, 5-20 nodes"

User: "Set disk size to 50GB"

Multi-Region

User: "Create EKS cluster in ap-southeast-1"

User: "List clusters in eu-west-1"

Tags

User: "Generate EKS cluster with tags: Environment=Production, Team=Platform, CostCenter=Engineering"

Terraform State Management

⚠️ Important: For production, configure remote state:

terraform {
  backend "s3" {
    bucket         = "my-terraform-state"
    key            = "eks/production/terraform.tfstate"
    region         = "us-east-1"
    encrypt        = true
    dynamodb_table = "terraform-locks"
  }
}

Next Steps After Cluster Creation

  1. Install core add-ons (metrics-server, AWS LB controller)

  2. Set up monitoring (Prometheus, Grafana)

  3. Configure logging (Fluent Bit → CloudWatch)

  4. Deploy sample application

  5. Set up CI/CD pipelines

  6. Implement GitOps (ArgoCD, Flux)

Resources

  • EKS Best Practices: https://aws.github.io/aws-eks-best-practices/

  • Terraform AWS EKS Module: https://registry.terraform.io/modules/terraform-aws-modules/eks/aws

  • EKS Documentation: https://docs.aws.amazon.com/eks/


Production-Ready EKS Clusters | Terraform + VPC + IRSA + Add-ons

A
license - permissive license
-
quality - not tested
B
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/dkeeno/eks-mcp-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server