Overview
The Kubernetes MCP Server is a Go-based native MCP server that talks directly to the Kubernetes API server. It is not a wrapper around kubectl or helm; it requires no external dependencies. It supports both Kubernetes and OpenShift resources via MCP, exposing CRUD operations for any resource, plus specialized tooling for Pods, Namespaces, Events, OpenShift Projects, and Helm releases. It can automatically detect changes in kubeconfig and adapt, supports multi-cluster contexts, and runs as a single binary across Linux, macOS, and Windows. The server is available as a native binary and also as an npm package, a Python package, and container images; it can be deployed in various environments without Node/Python being required for the native binary. Configuration is via command-line options like port, log level, read-only, disable-destructive, toolsets, and multi-cluster toggles. Toolsets enable or disable groups of tools (config, core, helm, kiali) to tailor the feature surface and reduce context size for AI tools. Demos and documentation are provided, and development guidance includes running with mcp-inspector to debug and inspect the MCP server.
Features
Configuration
Automatically detect changes in the Kubernetes configuration and update the MCP server. View and manage the current Kubernetes kubeconfig or in-cluster configuration.
Generic Kubernetes Resources
Perform CRUD operations on any Kubernetes or OpenShift resource, including Create or Update, Get, List, and Delete.
Pods
Pod-specific operations: list pods (all namespaces or a given namespace), get a pod by name, delete, show logs, top, exec into a pod, and run a container image.
Namespaces
List Kubernetes Namespaces.
Events
View Kubernetes events in all namespaces or in a specific namespace.
Projects
List OpenShift Projects.
Helm
Install Helm charts, list releases, and uninstall releases in the current or a provided namespace.
