MCP Kubernetes Server

Example Use Cases

• 1

  Query and fetch Kubernetes resources (pods, deployments, services, CRDs) as JSON or human-readable descriptions to analyze cluster state.

• 2

  Run kubectl or helm commands and perform read/write/delete operations (apply, create, scale, rollout, patch) against a cluster using a provided kubeconfig.

• 3

  Diagnose pod and node issues by retrieving logs, events, rollout history/status, and resource usage (top) and to manage node lifecycle actions like cordon/drain.

Description

This open-source server bridges AI assistants (that support MCP) and Kubernetes clusters by translating natural language requests into kubectl commands or Kubernetes API calls and returning structured results. It supports read, write, and delete operations, Helm commands, and diagnostic tooling such as logs, events, rollout status, and top metrics. The server can run via Docker or uvx, supports stdio/sse/streamable-http transports, and requires a kubeconfig plus kubectl/helm on the host. It is intended to enable agents to query, manage, and diagnose Kubernetes resources programmatically and conversationally.