I built an open-source MCP (Model Context Protocol) server for QNAP Virtualization Station that lets you manage your VMs using AI assistants like Claude.
What it does
35 tools for full VM management through natural language:
- VM lifecycle — start, stop, reset, suspend, resume
- VM management — update CPU/memory, clone, export, delete
- Snapshots — create, revert, delete
- Disks & ISOs — resize, mount/unmount
- Monitoring — resource dashboard with host CPU/RAM utilization, per-VM summary with networking
- Safety — all destructive operations require explicit confirmation
How it works
The server talks to the Virtualization Station REST API (the same API the web UI uses). It supports two modes:
- Local (stdio) — runs on your computer, connects to your NAS over HTTPS
- Remote (SSE) — runs on the NAS itself as a Docker container, AI clients connect over the network with Bearer token auth
Install options
# Via pip/uvx
uvx mcp-server-qnap-qvs
# Via Docker
docker pull ghcr.io/arnstarn/mcp-server-qnap-qvs:latest
# Via QPKG (sideload in App Center)
# Download from GitHub Releases
You can also add this App Center repository URL to get updates automatically:
https://raw.githubusercontent.com/arnstarn/mcp-server-qnap-qvs/main/qpkg/repo.xml
(App Center > Settings > App Repository)
Requirements
- QNAP NAS with x86 CPU (Intel/AMD)
- QTS 5.1.0+ or QuTS hero h5.1.0+
- Virtualization Station installed
- Container Station (for Docker/QPKG deployment)
Links
- GitHub: GitHub - arnstarn/mcp-server-qnap-qvs: MCP server for QNAP Virtualization Station (QVS) - manage virtual machines, snapshots, disks, and more via the QVS REST API · GitHub
- PyPI: Client Challenge
Tested on QuTS hero h5.2.8 with Virtualization Station 4.1.x. MIT licensed, contributions welcome.
Happy to answer questions or take feature requests!