mcp-hetzner
Enables conversational management of Hetzner Cloud, DNS, and Storage Box infrastructure through 137 MCP tools for CRUD operations and actions like server power control, DNS zone management, and storage box snapshots.
README
mcp-hetzner
A Model Context Protocol server that exposes the Hetzner Cloud, Hetzner DNS and Storage Box APIs as MCP tools, so any MCP-aware agent can manage your Hetzner infrastructure conversationally.
- Cloud API — full CRUD for servers, volumes, networks, firewalls, load balancers, floating IPs, primary IPs, SSH keys, images, ISOs, placement groups, TLS certificates, plus server actions (power on/off, reboot, rescue, rebuild, snapshot, resize, …).
- DNS API — full CRUD for zones, records (incl. bulk create/update), primary servers, zone-file import/export and validation.
- Storage Box API — full CRUD for Storage Boxes, sub-accounts, snapshots, snapshot plans, and box/sub-account actions (password reset, access settings, type change, snapshot rollback).
- Transports — both stdio and streamable HTTP are supported.
137 tools in total, registered under three clear namespace prefixes:
cloud_*, dns_*, sb_*.
⚠️ Disclaimer: This is an independent open-source project. It is not affiliated with, endorsed by, or sponsored by Hetzner Online GmbH. "Hetzner" is a trademark of Hetzner Online GmbH and is used here in a purely descriptive sense.
Install
Python 3.10+ required. Install directly from GitHub.
Option A — pipx from GitHub (recommended for end users)
pipx install git+https://github.com/ginkelsoft-development/mcp-hetzner.git
Pipx installs the package in an isolated environment and adds hetzner-mcp
to your $PATH. Upgrade later with:
pipx upgrade hetzner-mcp
To pin a specific tag/branch, append @<ref>:
pipx install "git+https://github.com/ginkelsoft-development/mcp-hetzner.git@main"
Option B — pip in a virtualenv
python -m venv .venv
source .venv/bin/activate
pip install git+https://github.com/ginkelsoft-development/mcp-hetzner.git
Option C — clone for development
git clone https://github.com/ginkelsoft-development/mcp-hetzner.git
cd mcp-hetzner
pip install -e .
Smoke test (no tokens needed)
hetzner-mcp --list-tools | python -c "import json,sys; print(len(json.load(sys.stdin)),'tools')"
# -> 137 tools
Tokens
The three Hetzner APIs each use their own token:
| Variable | Where to create it |
|---|---|
HCLOUD_TOKEN |
https://console.hetzner.cloud → your project → Security → API Tokens |
HDNS_TOKEN |
https://dns.hetzner.com/settings/api-token |
HETZNER_TOKEN |
https://console.hetzner.com → API Tokens (for Storage Boxes) |
Copy .env.example to .env and fill in whichever you need. You can run the
server with only the tokens you have — calls to unconfigured APIs return a
structured 401 error instead of crashing.
Use it in an MCP-compatible host
Every MCP-compatible host application accepts the same stdio-server block. Drop the snippet below into the host's MCP-server config and you're done.
The universal config block
{
"mcpServers": {
"hetzner": {
"command": "hetzner-mcp",
"env": {
"HCLOUD_TOKEN": "...",
"HDNS_TOKEN": "...",
"HETZNER_TOKEN": "..."
}
}
}
}
Where each host stores its config
| Host | Config location | Transport |
|---|---|---|
| Cursor | Settings → Cursor Settings → MCP → "Add new MCP server" | stdio or HTTP |
| Cline (VS Code) | Cline extension → "MCP Servers" panel | stdio |
| Continue (VS Code / JetBrains) | ~/.continue/config.json → experimental.modelContextProtocolServer |
stdio |
| Zed | ~/.config/zed/settings.json → context_servers |
stdio |
| Cody | Settings → Cody → MCP servers | stdio |
| Other desktop AI chat apps (macOS/Windows) | Their MCP / connector settings — see the app's own docs | stdio |
| Web-based AI chat apps | Their "custom connectors" / remote MCP settings | HTTP only |
| Custom Python/TypeScript agent (MCP SDK) | Whatever your client passes to StdioServerParameters / equivalent |
stdio or HTTP |
After saving the config, restart the host application. Most hosts surface a tool/plug icon in the chat interface when the server has connected successfully. You can confirm tool registration with a simple prompt such as:
"Run
hetzner_healthand tell me which tokens are configured."
Tip: absolute paths if hetzner-mcp isn't on the host's PATH
GUI applications often don't inherit your shell's PATH. If you see a
"command not found" error in the host's logs, replace "hetzner-mcp" with
the full path returned by which hetzner-mcp. Or use the explicit Python
form:
{
"mcpServers": {
"hetzner": {
"command": "/full/path/to/python",
"args": ["-m", "hetzner_mcp"],
"env": { "HCLOUD_TOKEN": "...", "HDNS_TOKEN": "...", "HETZNER_TOKEN": "..." }
}
}
}
Run as HTTP (remote / shared)
Use this when you want a single shared deployment instead of installing the server on each user's machine — for example to plug it into a web-based AI host that only accepts remote MCP servers via custom connectors.
hetzner-mcp --http --host 0.0.0.0 --port 8765
The streamable-HTTP endpoint is mounted at /mcp. Point an MCP-aware client
at http://your-host:8765/mcp.
🔒 Don't expose this endpoint to the public internet without a reverse proxy that enforces authentication and TLS. The MCP server itself has no auth layer — anyone who can reach the port can use your tokens.
A minimal hardening checklist before going public:
- Put the server behind a reverse proxy (Caddy, nginx, Traefik) with TLS.
- Require authentication at the proxy level (Basic auth, OAuth proxy, Cloudflare Access, Tailscale serve, etc.).
- Allow-list the IP ranges your host actually calls from, where possible.
- Use Hetzner API tokens scoped to read-only or to a specific project.
Tool naming
Tools are prefixed by API so it's always obvious which surface they touch:
cloud_*— Hetzner Cloud (servers, volumes, networks, …) — 91 toolsdns_*— Hetzner DNS (zones, records, …) — 20 toolssb_*— Storage Box (storage boxes, sub-accounts, snapshots, …) — 25 toolshetzner_health— reports which tokens are configured (no network call)
Run hetzner-mcp --list-tools for the full list with one-line descriptions.
Quick examples
"Spin up a cx22 in fsn1 with my SSH key, then add an A record for
staging.example.compointing at it."Calls:
cloud_list_ssh_keys→cloud_create_server→dns_list_zones→dns_create_record.
"Show me everything that's running, grouped by location."
Calls:
cloud_list_servers,cloud_list_volumes,cloud_list_load_balancers,sb_list_storage_boxes.
"Resize storage-box 12345 from bx11 to bx21 — take a snapshot first."
Calls:
sb_create_snapshot→sb_change_storage_box_type.
Safety notes
This server is fully read/write by default. Destructive tools
(cloud_delete_server, sb_delete_storage_box, dns_delete_zone, …) will
happily wipe resources if asked. For a safer setup, provision tokens with the
narrowest permission scope the API supports (Cloud and DNS tokens both
support read-only scopes).
Branch model
main— stable, what's published as releases.develop— active development; PRs target this branch.
See CONTRIBUTING.md for the full workflow.
Development
pip install -e ".[dev]"
ruff check src
hetzner-mcp --list-tools
License
MIT — © 2026 Ginkelsoft Development.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.