OpenClaw on a VPS in Frankfurt

Updated 2026-10-30

The OpenClaw app gives you a Frankfurt cloud server with OpenClaw, an AI agent gateway, installed, so your AI agents have an always-on home on a server you control in Germany.

It takes a few minutes after boot, and you keep full root access to the machine.

What is installed

• OpenClaw, an open-source AI agent gateway, set up to run on the server.
• A standard Linux system underneath with full root access.

A gateway like OpenClaw is the part of an agent setup that stays online: it holds the configuration, connects to the model providers you choose, and links agents to the tools and channels you use. Because it runs on your own server, its settings, keys and history stay there.

OpenClaw changes often. Its own documentation is the reference for features and configuration, and we recommend reading it before you connect accounts.

Which plan to start with

The language models run at the providers you connect, so the server does not need much power.

• Personal use or testing: the smallest Standard plan.
• Several agents, browser tools or other software on the same server: 2 to 4 GB of memory.

Billing is hourly from a prepaid balance, capped at the monthly price. Try it for a few hours, and resize or delete it when you know what you need.

Signing in after boot

1. Create a server in the console and choose OpenClaw as the app.
2. When the server is running, give the installation a few minutes.
3. Connect with SSH as root, using your SSH key or the emailed root password. See connect with SSH.
4. Follow OpenClaw's own setup steps to add your model provider keys and connect your first agent or channel.

First steps

Lock the server down first

An agent gateway holds API keys and can act on your behalf, so treat the server as sensitive.

• Use SSH keys and turn off password logins.
• Attach a free firewall that allows SSH and only the ports the gateway really needs. If you can, keep its interfaces off the public internet and reach them through SSH or a VPN.
• Turn on two-factor authentication for your Wolkra account. See teams and security.

Reach it privately

For a private setup, run a WireGuard server in the same private network and reach OpenClaw only through the VPN. The gateway server can even run without a public IP and use a NAT gateway for outbound calls to model providers.

Keep its data safe

Configuration and history live on the server. Turn on daily backups for 20% of the server price, and take a snapshot before upgrading OpenClaw.

Keep it updated

Projects in this space change quickly. Update OpenClaw and the operating system regularly, and read the release notes before major upgrades.

Who it suits

OpenClaw on a cloud server is a good fit if you:

• want an AI agent that stays online around the clock instead of running on your laptop;
• prefer to keep agent settings, keys and history on a server you control;
• want one place where several agents and the services they use come together;
• like to experiment, and want a clean server you can rebuild or delete at any time.

If you only want to try it, start with the smallest plan, test for an evening and delete the server. Billing stops when the server is gone.

Why run OpenClaw on Wolkra

• Always on. A cloud server keeps your agents reachable when your laptop is closed.
• Data in Germany. Settings and history stay on a server in Frankfurt. What you send to a model provider is governed by that provider. See EU VPS and GDPR.
• Unmetered traffic with fair use, on ports up to 10 Gbit/s, shared. See network.
• Cheap to try. Hourly billing means a test costs a few cents.

To connect the gateway to automations, look at the n8n app. For running your own services in general, see self-hosting.

Start your OpenClaw server

Create an account, add balance by card or crypto, and choose the OpenClaw app. Your agent gateway in Frankfurt is ready in minutes.