Kubernetes MCP Server

What it does

Kubernetes MCP bridges the gap between AI assistants and Kubernetes clusters. It solves common issues like context window overflow by using yq expressions to filter resource data before it reaches the LLM. It provides a secure way to manage clusters through natural language, supporting multi-cluster configurations and complex RBAC policies.

Tools

• get_resource, list_resources, describe_resource: Retrieve and inspect K8s resources using GVR.
• apply_manifest, patch_resource, delete_resource: Modify cluster state with safety rails.
• scale_resource, get_rollout_status, restart_rollout, undo_rollout: Manage workload scaling and deployments.
• get_logs, exec_command, list_events: Debugging tools with output truncation for context efficiency.
• get_cluster_info, list_api_resources, list_namespaces: Cluster-wide discovery and metadata.
• get_current_context, list_contexts, switch_context: Seamlessly navigate multi-cluster environments.
• check_permission, get_pod_metrics, get_node_metrics: Security audits and performance monitoring.
• diff_manifest: Compare local manifests against cluster state.

Installation

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "kubernetes": {
      "command": "/path/to/kubernetes-mcp",
      "args": ["--config", "/path/to/config.yaml"]
    }
  }
}

Supported hosts

• Claude Desktop