OpenClaw AI Agent Setup Guide on Contabo VPS
Part 1: What Is OpenClaw?
OpenClaw is an open-source AI agent that runs on your own server. In addition to answering questions, OpenClaw executes actions: automating tasks, interacting with AI models, and building workflows.
Why Self-Host?
| Feature | Cloud AI Services | OpenClaw (Self-Hosted) |
|---|---|---|
| Data control | On someone else's servers | ✅ Entirely on your infrastructure |
| Integration control | Limited by provider | ✅ You choose the integrations |
| Cost control | Provider-set pricing | ✅ You manage API usage & costs |
| Privacy | Shared infrastructure | ✅ Fully private |
OpenClaw is the agent framework. The AI intelligence comes from the provider you connect (OpenAI, Anthropic, etc.). You will need an API key from your chosen provider, which incurs usage-based costs per token.
Part 2: Choosing a Contabo VPS Plan
Step 1: Visit the OpenClaw Hosting Page
- Go to the OpenClaw hosting page on Contabo.
- Review the recommended configurations for running OpenClaw.
Step 2: Select a Plan
The Cloud VPS 10 is a great starting point, but pick the tier that suits your needs.
For light usage and experimentation, the Cloud VPS 10 is more than enough. If you plan to run additional services alongside OpenClaw or handle heavier workloads, consider a higher tier.
Step 3: Configure Your Server
- Click Configure on your chosen plan.
- Choose your server location: pick the region closest to you for lower latency (e.g., Europe, US, Asia).
- Set your admin password (this is critical):
| Used For | Details |
|---|---|
| SSH access | You'll type this when connecting via SSH |
| Initial admin credential | Your first login password for the server |
Choose a strong, unique password and save it somewhere secure. You'll need it repeatedly throughout this guide.
- Click Next.
Part 3: Ordering & Setting Up Your Server
For Existing Contabo Customers
- Log in with your existing credentials.
- You'll go straight to the order confirmation page.
For New Customers
Before reaching the order confirmation, you'll need to:
- Enter your personal information.
- Add a billing address.
- Set up a payment method.
This is a one-time setup. Future orders skip this step entirely.
Confirm Your Order
- Double-check your server specs, location, and that OpenClaw is listed as your selected application.
- Complete your order.
After Ordering
- You'll receive a confirmation email with:
- Your server's IP address
- A reminder about the password you set
- Log into the Contabo Customer Control Panel at new.contabo.com.
- Navigate to Servers and Hosting → VPS.
- Your new server will appear here.
The OpenClaw installation can take up to 30 minutes to complete in the background, but usually takes much less. You'll see the server status update while it's installing. Once it's ready, you can connect via SSH.
Part 4: Adding OpenClaw to an Existing VPS
If you already have a Contabo VPS, you can add OpenClaw without ordering a new server.
Reinstalling wipes ALL existing data on the server. Make sure you've backed up everything you need before proceeding!
Steps
- In your Customer Control Panel, find the server you want to use.
- Click the three dots (⋮) in the More section.
- Select Reinstall from the menu.
- In the installation options, go to Application Installation.
- From the dropdown, select OpenClaw.
- Set your admin password (same rules as before: this is your SSH and server access password).
- Note the warning, then click Install.
The status will change to Installing. Once complete, you're ready for SSH configuration.
Part 5: Connecting via SSH
OpenClaw doesn't have a public web interface by default. All initial configuration is done via SSH.
Step 1: Get Your Server IP
Copy your server's IP address from the Contabo Customer Control Panel.
Step 2: Open Your Terminal
| Operating System | Application to Use |
|---|---|
| macOS | Terminal |
| Linux | Terminal |
| Windows 10/11 | PowerShell |
Step 3: Connect
bashssh root@YOUR_SERVER_IP
- Press Enter.
- When prompted, type your admin password and press Enter.
- Once connected, the command prompt indicates you are logged into your server.
On your first connection, you may be asked to trust the server fingerprint. Type yes and press Enter.
Part 6: Running the Onboarding Wizard
After logging in, OpenClaw may start the onboarding wizard automatically. If it doesn't, run:
bashopenclaw onboard
Step 1: Security Acknowledgment
OpenClaw reads files on your server and executes actions if those capabilities are enabled. Confirm that you understand these implications.
OpenClaw can interact with your server at a deep level. Only enable capabilities you trust and understand.
Action: Confirm that you understand.
Step 2: Choose Onboarding Mode
| Mode | Best For |
|---|---|
| Quick Start (recommended) | Configures basic settings you can adjust later |
| Manual | Full control over every setting during setup |
Action: Select Quick Start.
Step 3: Select Your AI Provider
OpenClaw supports multiple AI providers:
- OpenAI
- Anthropic
- Others
Action: Choose your preferred provider (this guide uses OpenAI as an example).
Step 4: Choose Authentication Method
- API Key (Recommended)
Step 5: Enter Your API Key
This is the most critical step. OpenClaw will not function without a valid API key.
How to Get an OpenAI API Key
- Go to platform.openai.com.
- Navigate to API Keys.
- Click Create new secret key.
- Copy the key immediately (you won't be able to see it again).
Action: Paste your API key into the terminal.
Using AI provider APIs incurs usage-based costs per token, which can add up quickly depending on your usage. You are responsible for managing your API keys, usage limits, and OpenClaw security. Contabo does not control or monitor how OpenClaw uses these services.
Step 6: Choose Your Preferred Model
Select the AI model you want OpenClaw to use (e.g., GPT-4, GPT-3.5-turbo, etc.).
Step 7: Configure Messaging Channels (Optional)
OpenClaw can receive messages from multiple platforms:
| Platform | Setup |
|---|---|
| Telegram | Now or later |
| Discord | Now or later |
| Slack | Now or later |
| Now or later |
Action: Set up now or skip and configure later.
Step 8: Enable Skills and Hooks (Optional)
Skills are capabilities you can enable for OpenClaw:
| Skill Type | Examples | Notes |
|---|---|---|
| Web search | Browse and summarize web pages | May require additional API keys |
| File access | Read/write files on the server | Powerful (enable with caution) |
| Code execution | Run scripts and commands | Requires additional dependencies |
Action: Enable what you need or skip for now.
Onboarding Complete
After a few moments, OpenClaw:
- Installs its gateway service
- Initializes the agents
OpenClaw is now installed and running on your server! 🎉
Part 7: Interacting via the Terminal UI (TUI)
The Terminal UI is the recommended way to interact with your OpenClaw agent directly from SSH.
Launch the TUI
bashopenclaw tui
You can also type tui at the end of the onboarding wizard to jump straight in.
Using the TUI
- Type messages just like you would with any AI assistant.
- OpenClaw responds using the AI model you configured.
- If you enabled system tools, you can ask OpenClaw to perform actions on your server directly.
Example Interactions
You: What's my current disk usage?
OpenClaw: [executes df -h and returns results]
You: Summarize the latest system logs.
OpenClaw: [reads /var/log/syslog and provides a summary]
Exit the TUI
- Press Ctrl + C, or
- Follow the TUI exit instructions displayed on screen.
Part 8: Accessing the Web-Based Control UI
OpenClaw also has a web-based Control UI with a visual, browser-based experience.
Why It's Not Publicly Accessible
The Control UI runs on localhost:18789, making it accessible only from the server itself. This layout keeps the AI agent secure.
Access It with SSH Port Forwarding
SSH port forwarding creates a secure tunnel from your local computer to the server.
Step 1: Open a New Terminal Window
On your local computer, open a new terminal window (keep any existing SSH sessions open).
Step 2: Create the SSH Tunnel
bashssh -L 18789:localhost:18789 root@YOUR_SERVER_IP
- Enter your password when prompted.
- The terminal will appear to hang, which is normal. The tunnel is running.
Keep this terminal window open for as long as you need access to the Control UI. Closing it will close the tunnel.
Step 3: Generate a Tokenized Link
In a separate terminal window, SSH into your server:
bashssh root@YOUR_SERVER_IP
Then run:
bashopenclaw dashboard
After a few moments, OpenClaw will output a tokenized link.
Step 4: Open the Control UI
- Copy the tokenized link.
- Paste it into your browser.
- The OpenClaw Control UI appears! 🎉
What You Can Do in the Control UI
- 💬 Chat with your agent
- ⚙️ Adjust settings
- 🔌 Enable or disable skills
- 📊 Monitor activity
Closing the Control UI
- When you're done, press Ctrl + C in the terminal running the SSH tunnel.
- The Control UI will become inaccessible until you create the tunnel again.
This is a security feature, not a bug. Your agent is only accessible when you explicitly create the tunnel.
Part 9: Next Steps & Advanced Configuration
Now that OpenClaw is installed, here's what to explore next:
Connect Messaging Platforms
Link OpenClaw to your preferred messaging apps so you can interact with your agent from anywhere:
- Telegram (useful for mobile interaction)
- Discord (useful if you are already active in Discord communities)
- Slack (designed for team and work environments)
- WhatsApp (suited for personal use)
Enable Advanced Skills
Extend OpenClaw's capabilities through skills:
- Web browsing and research
- File management
- Code execution and scripting
- Custom integrations
Build Custom Workflows
Create AI-powered automation workflows tailored to your specific needs:
- Task automation
- Data processing pipelines
- Scheduled operations
- Multi-step AI chains
Part 10: Security Best Practices
OpenClaw has access to your server and can execute actions. Treat security seriously.
Essential Security Measures
| Measure | Why It Matters |
|---|---|
| Use strong passwords | Prevents brute-force attacks on SSH and admin access |
| Keep software updated | Patches known vulnerabilities |
| Use SSH keys | More secure than password-based SSH authentication |
| Monitor AI provider usage | API costs can escalate quickly with heavy usage |
| Limit enabled skills | Only enable capabilities you actually need |
| Review agent activity | Regularly check what actions your agent is performing |
Monitor API Costs
OpenClaw's power comes with API costs that vary based on usage:
| Factor | Impact on Cost |
|---|---|
| Model choice | GPT-4 is more expensive than GPT-3.5 |
| Message length | Longer prompts and responses = more tokens |
| Frequency of use | More interactions = higher bills |
| Skills with API calls | Web search and similar skills may use additional API credits |
Set usage limits in your AI provider's dashboard (e.g., OpenAI's usage page) to avoid unexpected bills. Start with conservative limits and adjust as you understand your usage patterns.
Quick Reference: All Commands
bash# --- Connect to Your Server ---
ssh root@YOUR_SERVER_IP
# --- Run the Onboarding Wizard ---
openclaw onboard
# --- Launch the Terminal UI ---
openclaw tui
# --- Generate the Control UI Dashboard Link ---
openclaw dashboard
# --- SSH Port Forwarding (run on your LOCAL machine) ---
ssh -L 18789:localhost:18789 root@YOUR_SERVER_IP
# --- Open the Dashboard in Browser (after port forwarding) ---
# Copy the tokenized link from 'openclaw dashboard' and paste into your browser