Unraid MCP Server: Give AI Agents Access to Your VMs, Docker Containers, and Array Status

Why Give AI Access to Unraid?

Unraid is the operating system for self-hosted everything. It manages your storage array with mix-and-match drives, runs Docker containers and VMs, and serves as the backbone for home labs and media servers worldwide.

But when you're coding scripts or managing infrastructure from Claude or Cursor, your AI assistant can't see your Unraid server. It can't check if a container crashed, verify array health, or look at VM resource allocation. You end up opening the Unraid WebGUI in another tab.

With DataFaucet, you browse your Unraid dashboard once and it captures the API calls as MCP tools. Now your AI coding assistant can monitor array status, manage containers, and check system health directly.

What AI Agents Can Do With Unraid Access

Once DataFaucet creates an MCP server from your Unraid session, AI agents can:

• Check array status including disk health, temperatures, and parity check progress
• List Docker containers with status, resource usage, and port mappings
• Monitor VMs with CPU, memory allocation, and running state
• Browse shares and see disk allocation across the array
• View system stats including CPU, RAM, UPS status, and network throughput
• Check notifications for hardware warnings or failed operations
• List installed plugins and their update status

Setting Up Unraid as an MCP Server with DataFaucet

Step 1: Browse Your Unraid Dashboard

Open your Unraid WebGUI (typically at your-server.local or an IP address) and navigate through the pages you want AI access to: Main (array), Docker, VMs, Shares, and Settings.

Step 2: DataFaucet Captures the API

DataFaucet's browser extension records the API calls your browser makes to Unraid. Each call becomes a callable MCP tool with parameters and response types.

Step 3: Connect to Your AI Client

Add the generated MCP server to Claude Desktop, Cursor, or Windsurf. Your AI assistant can now query your Unraid server whenever it needs infrastructure context.

Example: AI Checks If a Container is Running

User: Is my Plex container running? How much RAM is it using?

AI (via DataFaucet MCP): Plex is running. Container stats:
- Status: Up 14 days
- Memory: 2.1 GB / 4 GB limit
- CPU: 3.2% average
- Network: 450 MB received today

Example: AI Monitors Array Health

User: Any disk issues on my Unraid array?

AI (via DataFaucet MCP): Array is healthy. 8 data disks + 1 parity.
- All disks report PASSED SMART status
- Highest temp: Disk 3 at 38°C
- Parity check: Last completed 3 days ago, 0 errors
- Array utilization: 67% (24.2 TB / 36 TB)

Why Not Just Use the Unraid API Directly?

Unraid's WebGUI communicates via a mix of PHP endpoints, WebSocket connections, and custom APIs that vary by plugin. Building a dedicated MCP server would mean reverse-engineering each endpoint, handling auth tokens, and maintaining compatibility across Unraid versions.

DataFaucet skips all of that. It watches what your browser does and replicates those exact calls as structured MCP tools. If Unraid updates its API, you just re-capture.

Works With Any MCP Client

The Unraid MCP server created by DataFaucet works with:

• Claude Desktop and Claude Code
• Cursor (AI-powered editor)
• Windsurf (Codeium's IDE)
• Any MCP-compatible client

Your AI gets real-time data from your Unraid server regardless of which tool you use.

Get Started

Create your Unraid MCP server with DataFaucet — browse your Unraid WebGUI once, get AI access to your entire server infrastructure in under two minutes.