k8m

k8m is an AI-driven, lightweight Kubernetes dashboard designed to simplify cluster management. It integrates a multi-cluster MCP toolset and an embedded AI assistant to help you interact with Kubernetes resources, run commands, view logs, and manage access across clusters. The server bundles a wide range of built-in MCP-enabled capabilities, including multi-cluster management, APIGateway and OpenKruise support, CRD discovery, Helm integration, and access control integration with large-model workflows. You can access the web UI at the default port and use the built-in tools to manage clusters, run kubectl-like operations, inspect resources, and trigger AI-assisted actions and explanations.

Prerequisites:

• A machine or VM with a supported OS (Linux, macOS, or Windows) and internet access
• A downloaded release of k8m or access to its binary build
• Optional: Docker for container-based deployment via docker-compose

Installation steps (binary release):

1. Download the latest release from GitHub: https://github.com/weibaohui/k8m/releases
2. Make the binary executable: chmod +x k8m
3. Run the server: ./k8m
4. Open your browser and navigate to http://127.0.0.1:3618 (default credentials: username: k8m, password: k8m)

Alternative: Docker compose (recommended for quick local trials):

1. Use the provided docker-compose snippet:

services: k8m: container_name: k8m image: registry.cn-hangzhou.aliyuncs.com/minik8m/k8m restart: always ports: - "3618:3618" environment: TZ: Asia/Shanghai volumes: - ./data:/app/data

2. Start the service: docker-compose up -d
3. Access the UI at http://localhost:3618 with the default credentials mentioned above.

• Default port: 3618. If you run behind a reverse proxy or in a different environment, adjust the port mapping accordingly.
• The server ships with MCP-enabled tooling to perform a wide range of Kubernetes operations through model-assisted workflows. You can log actions and leverage AI to translate YAML attributes, explain resource details, and run recommended commands.
• If you need debugging information about the AI integration, start the server with higher verbosity, e.g. ./k8m -v 6, to capture more detailed logs.
• For multi-cluster deployments, ensure kubeconfig is accessible and that the server has permissions to read the clusters you intend to manage. The UI supports auto-registration of clusters and heartbeat-based reconnection.
• If using Docker, ensure the volume mappings provide persistence for data under ./data to avoid losing configuration on container restarts.