The Kubernetes MCP Server, developed by Containers, bridges AI assistants with Kubernetes and OpenShift environments. This free MCP server, built with Go, provides a native binary for Linux, macOS, and Windows, and can also run as a container. It extends AI capabilities to manage cluster resources, offering a powerful way for agents to interact with and observe cloud-native deployments. The server is available as an mcpb extension for Claude Desktop, enhancing AI-driven operations.

What does Kubernetes MCP Server connect to?

The Kubernetes MCP Server acts as a Model Context Protocol (MCP) server, linking AI assistants to Kubernetes and OpenShift clusters. It functions as an intermediary, allowing AI agents to interact with these container orchestration platforms. This connection enables AI clients to perform operations and gather information from cluster environments. The server is designed to work with various AI clients, making it a central point for AI-driven cluster interactions.

Core features and tools

The server exposes callable tools for managing Kubernetes resources, including CRUD operations on any resource. Specific pod operations include list pods, get a pod by name, delete a pod by name, show logs for a pod by name, top gets resource usage metrics, exec into a pod, and run a container image. It also supports viewing Kubernetes events, executing "kubectl" commands, and managing clusters through natural language. Tools like list_resources, describe_resource, get_pod_logs, and rollout_restart are available. Specific tool groups can be enabled or disabled.

Who is it for?

The Kubernetes MCP Server is designed for AI assistants/agents, platform engineers, DevOps engineers, and SRE teams. It helps these groups manage and observe Kubernetes resources using natural language and AI-driven workflows. The server supports interaction with multiple clusters, making it suitable for environments with varied deployment needs. Compatible AI clients include Claude, Cursor, GitHub Copilot, and others like LangChain and OpenAI. Skip this if your environment does not use Kubernetes or OpenShift clusters.

How to get started

Access is granted through a standard OAuth authorization flow, or by using the default Kubernetes configuration from the local "~/.kube/config" file or an in-cluster service account. OAuth 2.1 with PKCE enforcement is supported for HTTP transports.

Use cases

• AI-assisted troubleshooting and diagnosis of deployments by querying pod logs and events via natural language.
• Natural language-driven Kubernetes resource management, enabling users to perform CRUD operations, list resources, and scale deployments using conversational prompts.
• Multi-cluster management and observability, allowing AI agents to interact with and monitor Kubernetes resources across multiple clusters configured in a kubeconfig file.

The Kubernetes MCP Server by Containers provides a valuable bridge for AI agents to interact with Kubernetes and OpenShift. It offers extensive resource management capabilities through callable tools and supports various authentication methods for secure operations. The server enhances observability and management across single or multiple clusters, making complex tasks more accessible via natural language.