Advantech GAIS MCP Server

A Model Context Protocol (MCP) server implementation for monitoring system status and hardware resources through Grafana and Prometheus.

Features

• Real-time system monitoring:
  • CPU usage
  • Memory usage
  • Disk usage
  • GPU monitoring (if available):
    • GPU power consumption
    • GPU temperature
    • GPU VRAM usage

Installation

npm install advantech-gais-mcp

For one-time usage, you can also use npx:

npx -y advantech-gais-mcp@latest

MCP Configuration

Add this to your MCP config:

"GAISMCP": {
  "command": "npx",
  "args": ["-y", "advantech-gais-mcp@latest"],
  "env": {
    "GRAFANA_URL": "your-grafana-url",
    "GRAFANA_API_KEY": "your-grafana-api-key"
  }
}

Usage

1. Set up the required environment variables:

export GRAFANA_URL="http://your-grafana-server:3000"
export GRAFANA_API_KEY="your-grafana-api-key"

2. The server exposes the following MCP tool:

• Tool Name: GAISMCP.getSystemStatus
• Description: Retrieves current system status and hardware resource usage
• Returns: JSON object containing:
  • CPU usage percentage
  • Memory usage percentage
  • Disk usage percentage
  • GPU metrics (if available)

Prerequisites

• Node.js 16 or higher
• Running Grafana instance
• Prometheus data source configured in Grafana
• Node Exporter for system metrics
• DCGM Exporter (for NVIDIA GPU metrics, if needed)

Environment Variables

| Variable | Description | Required | Default | |----------|-------------|----------|---------| | GRAFANA_URL | Your Grafana server URL | Yes | http://localhost:3000 | | GRAFANA_API_KEY | Grafana API key with read permissions | Yes | - |

License

ISC

Author

Advantech GAIS Team