Kubernetes MCP ServerUnclaimed

Kubernetes MCP Server is a Model Context Protocol (MCP) server that lets AI assistants work with Kubernetes and OpenShift clusters. It supports create, update, get, list and delete operations on any Kubernetes or OpenShift resource, along with Pod-specific actions such as viewing logs, running commands inside a pod and checking resource usage. It also covers namespaces, events, OpenShift projects, Helm releases, and Tekton pipelines and tasks.

The server is written in Go and talks directly to the Kubernetes API server rather than wrapping kubectl or helm. The page says no external tools need to be installed. It ships as a single native binary for Linux, macOS and Windows, and is also available as an npm package, a Python package and a container image. It can work with several clusters defined in a kubeconfig file, and it is configured with command-line arguments, TOML files or environment variables.

The project is aimed at developers and operators who want an AI assistant such as Claude Desktop, VS Code, Cursor or Goose to inspect and manage cluster resources. The README gives setup steps for each of these clients, using npx with a short JSON or YAML entry. Guardrails include a read-only setting and denied resources, which the page describes as best-effort, and it names Kubernetes RBAC as the access control to rely on. Optional OpenTelemetry tracing is also available.

The code is published under the Apache-2.0 license. The page does not list a hosted service or paid plans.