Frequently Asked Questions

What is RunClaw?

RunClaw is a control plane for self-hosted AI agents. It helps you deploy and operate OpenClaw, Agent Zero, or Goose on infrastructure you control, while keeping server ownership and shell access on your side.

You can start with one of three setup paths:

• Provision on Hetzner for the easiest cloud path
• Provision on Vultr as the alternative cloud path
• Connect an existing Ubuntu server over SSH if you already run your own infrastructure

See Choose Your Setup Path for the side-by-side version.

Do I need a Hetzner account?

No. Hetzner is the easiest cloud path, but it is not the only path.

• If you choose Provision on Hetzner, you need a Hetzner Cloud account and API token
• If you choose Provision on Vultr, you need a Vultr account and API key
• If you choose Connect existing Ubuntu server over SSH, you use a server you already control

How much does it cost?

There are three separate costs:

1. RunClaw -- free during the current beta (no platform fee)
2. Infrastructure -- if you provision on Hetzner or Vultr, they bill you directly; if you use SSH, you keep using your existing server
3. AI provider -- pay-as-you-go to your chosen provider (OpenRouter, OpenAI, Anthropic, etc.)

RunClaw does not own your infrastructure account. You keep the provider relationship and pay infrastructure costs directly.

Is my data private?

RunClaw is built so you keep control of the infrastructure account and server. Here is what happens with different types of credentials:

• Cloud provider tokens (Hetzner, Vultr) are encrypted in your browser and never stored on RunClaw's servers. RunClaw stores only one-way hashes for verification.
• Setup-time secrets (access gate token, optional LLM keys) are temporarily sent to RunClaw during provisioning so they can be injected into your server. LLM keys are deleted from our database after setup completes.
• Management features (logs, terminal, workspace, config) relay content through RunClaw's control plane when you use them. RunClaw does not continuously inspect your server.

You own the provider account, the server, and the data on it. RunClaw operates the control-plane layer: setup coordination, URLs, status, updates, and management workflows.

Path-specific setup details are documented in Choose Your Setup Path and the setup guides.

What agents are supported?

RunClaw currently supports three agent types:

• OpenClaw -- A messaging-first AI agent with chat platform integrations (Telegram, Discord, WhatsApp, Slack)
• Agent Zero -- An autonomous coding agent that can browse the web, write code, and execute tasks
• Goose -- A general-purpose AI agent for automation

Each agent is installed as a Docker container on your dedicated VPS with automatic TLS, a management sidecar, and Caddy as a reverse proxy.

Which AI provider should I use?

We recommend OpenRouter for beginners. One API key gives you access to 200+ models from OpenAI, Anthropic, Google, Meta, and more. Pay-as-you-go with no minimums -- you can start with as little as $1.

See Configure LLM Keys for a full comparison of providers and setup instructions.

Can I SSH into my server?

Yes. You keep shell access on every path.

ssh root@YOUR_SERVER_IP

If you provision on Hetzner or Vultr, the server IP is shown on your RunClaw dashboard. If you connect an existing server over SSH, use the host you already manage.

See SSH & Server Access for detailed commands and file locations.

What happens if I cancel RunClaw?

Your infrastructure keeps running under your control. RunClaw's dashboard and management workflows stop, but the server and agent continue to exist on the infrastructure you chose.

If you provisioned on Hetzner or Vultr, you can manage the machine directly in that provider and over SSH. If you connected an existing server, it remains your server before, during, and after RunClaw.

How do I update my agent?

From your RunClaw dashboard:

1. Click on your instance
2. If an update is available, you'll see an Update button with a version badge
3. Click it -- the update runs automatically with progress indicators

Updates are typically zero-downtime. Your agent shows a brief maintenance page while the container is rebuilt with the new version. You can also enable auto-updates in your instance settings.

Can I run multiple agents during beta?

Yes, with the current beta limits. Today the product allows one instance of each supported agent type per account during beta.

Can I use a custom domain?

Yes. You can point your own domain to your agent's VPS. Caddy (the reverse proxy on your server) handles automatic TLS certificate provisioning via Let's Encrypt.

See Custom Domains for setup instructions.

What if my infrastructure provider has an outage?

Your uptime depends on the infrastructure path you chose:

• If you provisioned on Hetzner, a Hetzner outage affects that server
• If you provisioned on Vultr, a Vultr outage affects that server
• If you connected an existing Ubuntu server over SSH, uptime depends on that server and network

RunClaw is in beta with no SLA. Provider outages, instance failures, and network disruptions are outside RunClaw's control.

What is the access gate?

Every RunClaw agent is protected by a gate page that requires a token to access. This prevents anyone from finding your agent's URL and using it without permission.

The token is shown during setup and available on your dashboard. Once you enter it, a cookie is set and you won't need to enter it again on that device. The validation happens entirely in your browser using SHA-256 hashing.

Still stuck?

Discord is the fastest way to get help during beta. Join the community to ask questions, report issues, or share feedback directly with the team.

Join RunClaw Discord

For anything that doesn't fit Discord, email hello@runclaw.run. For abuse reports, email report-abuse@runclaw.run.