Azure/mcp-kubernetes

A Model Context Protocol (MCP) server that enables AI assistants to interact with Kubernetes clusters. It serves as a bridge between AI tools (like Claude, Cursor, and GitHub Copilot) and Kubernetes

60 ★20 forksGoUpdated 24d ago

What you need to know

A Go MCP server that bridges AI assistants to Kubernetes clusters, translating natural language into kubectl operations with configurable access levels.

Install

Docker: ghcr.io/azure/mcp-kubernetes
Or download the mcp-kubernetes binary locally
Requires kubectl and optionally helm

Usage

•Mount your kubeconfig into the server
•Ask questions like cluster status, pod diagnostics, or scaling
•Control permissions via --access-level (readonly, readwrite, admin)

Key features

✓Unified call_kubectl tool or legacy specialized tools
✓Access-level filtering at registration time
✓Additional helm, cilium, and hubble tools
✓Namespace allowlisting
✓OpenTelemetry tracing via OTLP
✓Telemetry opt-out via env var

Best for

Kubernetes operators who want AI-assisted cluster management with strict permission controls.

Caveats

⚠Telemetry is on by default; opt out with KUBERNETES_MCP_COLLECT_TELEMETRY=false
⚠stdio transport only

Clients: Claude Desktop · Cursor · GitHub Copilot · Any MCP-compatible client

Reviewed 2026-08-11

Topics

kubernetesmcp-servermodel-context-protocol

Stars: 60★
Forks: 20
Language: Go
License: MIT
Created: 2025-04-25
Last push: 2026-08-11