MCP Kubernetes Server

TypeScript MCP stdio server for Kubernetes operational tasks.

Prerequisites

Required

• Node.js 18+ and npm
• kubectl installed and configured
• SSH access to Kubernetes master nodes

For Nested SSH (Master → Worker Nodes)

• sshpass installed on master nodes (not on your local machine)
• Only needed if you use check_node_logs or nested SSH commands

Installation

npm install
npm run build

Run

npm start

MCP Configuration

Option 1: Using Config File (Recommended)

Create a config file anywhere on your local machine (e.g., C:\Users\your-username\.ssh\mcp-config.json or D:\configs\mcp-config.json):

{
  "HOST_ALIASES": "pod1-master=kubemaster01.pod1.com,pod2-master=kubemaster01.pod2.com",
  "USERNAME": "your-username",
  "PASSWORD": "your-password"
}

Note: This same file can be used for both Kubernetes MCP and Vault MCP servers!

Then reference it in mcp_config.json using an absolute path:

{
  "mcpServers": {
    "kubernetes-ts": {
      "command": "npx",
      "args": ["-y", "mcp-kubernetes-server"],
      "env": {
        "MCP_CONFIG_FILE": "C:\\Users\\your-username\\.ssh\\mcp-config.json"
      }
    }
  }
}

The config file can be located anywhere:

• ✅ C:\Users\your-username\.ssh\mcp-config.json
• ✅ D:\my-configs\mcp-config.json
• ✅ C:\configs\mcp-config.json
• ✅ Any absolute path you prefer!

Option 2: Direct Environment Variables

{
  "mcpServers": {
    "kubernetes-ts": {
      "command": "npx",
      "args": ["-y", "mcp-kubernetes-server"],
      "env": {
        "SSH_HOST_ALIASES": "pod1-master=kubemaster01.pod1.com,pod2-master=kubemaster01.pod2.com",
        "SSH_USERNAME": "your-username",
        "SSH_PASSWORD": "your-password"
      }
    }
  }
}

Tools

Kubernetes Resources

• get_resources - Get any K8s resource (ingress, services, configmaps, etc.) in various formats
• get_secrets - Get secrets with automatic base64 decoding
• get_nodes - Get cluster nodes with optional metrics

Pod Management

• get_pod_status - Get pod status in tabular/json/yaml format
• describe_pod - Describe pod details
• get_pod_logs - Get current/previous pod logs
• diagnose_pod_failure - Comprehensive pod failure diagnosis

Cluster Operations

• list_contexts - List kubeconfig contexts
• switch_context - Switch active context
• get_service_nodes - Get nodes hosting service pods

SSH & Logs

• list_ssh_aliases - List SSH host aliases
• execute_ssh_command - Execute SSH commands on remote hosts
• check_node_logs - Check node logs (supports serviceName and custom logPath)