Verified

Kubernetes MCP Server

Kubernetes cluster management and inspection for AI-powered development workflows.

Data last scanned today · Reviewed today

28 5 dangerous tools23 tools★ 1,576TypeScriptMIT

Overview

The Kubernetes MCP Server provides a direct interface for interacting with and managing Kubernetes clusters from within AI environments. It leverages existing local kubectl configurations to facilitate cluster visibility and operational control, supporting standard kubeconfig locations by default alongside support for Helm v3 for deployment tasks. Designed for integration via npx, the server is configured to pull credentials from your environment or home directory. It offers developers a way to query cluster resources and manage deployments through natural language commands, bridging the gap between local development tools and remote cluster orchestration.

Our verdict

This server carries a high-risk score of 53/100, primarily because the authentication methods and read-only mode capabilities have not yet been editorially reviewed. Given that it requires direct access to your local kubeconfig and interacts with sensitive cluster infrastructure, it is currently recommended only for use in isolated, non-production environments where security policies are strictly managed by the user.

Active development with commits within the last week
Supported by a community of 44 contributors
Small supply-chain surface with only 12 direct dependencies
Uses the permissive MIT license for flexible integration

Authentication methods have not undergone security review
Read-only mode functionality is not yet verified
Lacks official vendor maintenance and support

Setup

Claude Desktop

{
  "mcpServers": {
    "mcp-server-kubernetes": {
      "args": [
        "-y",
        "mcp-server-kubernetes"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
kubectl_get	Get or list Kubernetes resources.	low
kubectl_describe	Describe Kubernetes resources.	low
kubectl_create	Create Kubernetes resources.	medium
kubectl_apply	Apply YAML manifests to Kubernetes resources.	medium
kubectl_delete	Delete Kubernetes resources.	high
kubectl_logs	Get logs from Kubernetes resources.	low
kubectl_context	Manage kubectl contexts.	medium
explain_resource	Explain Kubernetes resources.	low
list_api_resources	List Kubernetes API resources.	low
kubectl_scale	Scale Kubernetes resources.	medium

Compatibility

Client	Local	Docker	Remote	Read-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›What software must be installed on my machine to use this server?

You must have kubectl installed and in your PATH. Helm v3 is also required if you intend to use Helm-related features.

›How does the server authenticate with my Kubernetes cluster?

The server uses your existing kubeconfig file. By default, it looks for the configuration at ~/.kube/config, though additional authentication methods are supported.

›Do I need special cluster access to run this?

You need access to a Kubernetes cluster that is already configured for kubectl, such as minikube, Rancher Desktop, or GKE. You can verify your access by confirming that the kubectl get pods command works in your terminal.

›Can I use this server in Cursor or VS Code?

Yes, both environments support the server by adding the provided JSON configuration block to your respective MCP settings. You can also use the provided one-click installation badge for VS Code extensions.