Multi-Cluster MCP server

The Multi-Cluster MCP Server provides a robust gateway for Generative AI (GenAI) systems to interact with multiple Kubernetes clusters through the Model Context Protocol (MCP). It facilitates comprehensive operations on Kubernetes resources, streamlined multi-cluster management, and delivered interactive cluster observability.

🚀 Features

🛠️ MCP Tools - Kubernetes Cluster Awareness

✅ Retrieve resources from the hub cluster (current context)
✅ Retrieve resources from the managed clusters
✅ Connect to a managed cluster using a specified ClusterRole
✅ Access resources across multiple Kubernetes clusters(via Open Cluster Management)
❌ Retrieve and analyze metrics, logs, and alerts from integrated clusters
❌ Interact with multi-cluster APIs, including Managed Clusters, Policies, Add-ons, and more

📦 Prompt Templates for Open Cluster Management (Planning)

Provide reusable prompt templates tailored for OCM tasks, streamlining agent interaction and automation

📚 MCP Resources for Open Cluster Management (Planning)

Reference official OCM documentation and related resources to support development and integration

📌 How to Use

Use with MCP Inspector

mcp dev ./src/multicluster_mcp_server/__main__.py

Configure the server using the following snippet:

{ "mcpServers": { "multicluster-mcp-server": { "command": "uvx", "args": [ "multicluster-mcp-server@latest" ] } } }

Note: Ensure kubectl is installed. By default, the tool uses the KUBECONFIG environment variable to access the cluster. In a multi-cluster setup, it treats the configured cluster as the hub cluster, accessing others through it.

Related MCP server: GKE Hub API MCP Server