Kubernetes MCP Server

Kubernetes MCP Server

v0.0.65
Apache-2.0
kubernetesopenshiftdevopscloudhelmcontainers

Summary

Native Go MCP server for Kubernetes and OpenShift — talks to the API server directly, no kubectl required.

Features

  • Direct Kubernetes API access — no kubectl dependency
  • Multi-cluster support driven by your kubeconfig
  • Pod operations: list, get, logs, exec, run, delete
  • Generic CRUD for any Kubernetes resource type
  • Helm chart management
  • Optional Tekton, KubeVirt, NetObserv, and Istio/Kiali integrations
  • Read-only and stateless modes for safer production access

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Kubernetes MCP Server" tools will be available.
{
  "mcpServers": {
    "kubernetes-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "kubernetes-mcp-server@latest"
      ]
    }
  }
}

Description

kubernetes-mcp-server, maintained under the containers GitHub organization, lets an agent operate Kubernetes and OpenShift clusters by talking to the API server directly. It is a native Go implementation rather than a wrapper that shells out to kubectl, so it does not depend on external binaries being installed or on parsing CLI output.

It covers generic resource CRUD for any resource type, pod operations (list, get, logs, exec, run, delete), Helm chart management, and multi-cluster access driven by your existing kubeconfig. Optional integrations extend it to Tekton pipelines, KubeVirt virtual machines, network observability through NetObserv, and service-mesh insight via Istio and Kiali.

Two operational modes make it safer to point at real infrastructure: read-only mode, which removes every mutating tool, and stateless mode for HTTP deployments. It ships as a single cross-platform binary and is also distributed via npm, pip, and Docker, under Apache-2.0.

Related MCP Servers

Browse all MCP servers →