Graduate to Production

You've outgrown the sandbox. Here's how to run OpenClaw for real.

LobsterSandbox was your training wheels. Now you're ready to set up a full OpenClaw instance with your real accounts, real infrastructure, and real workflows. Here's how to make the transition safely.

Before you graduate, make sure you've:

Completed at least a few Sandbox Missions

Understand how OpenClaw agents, channels, and skills work

Know which AI model you want to use and roughly what it costs

Decided which channels you want to connect (WhatsApp, Discord, Telegram, etc.)

Set a real budget for API costs

Not there yet? No rush. Your sandbox isn't going anywhere.

← Back to Sandbox Missions

Step 1: Export Your Sandbox Config

Download your current sandbox configuration so you can import it into a fresh OpenClaw installation. This includes your model preferences, safe mode settings, and channel configurations. API keys are NOT included in the export for security.

Log in to export your config. You need to be logged in to download your sandbox configuration.

→ Go to Setup

API keys are never included in exports. You'll re-enter them in your new setup.

Step 2: Install OpenClaw

Follow the official setup on your own machine or cloud server.

1 Install OpenClaw

curl -fsSL https://www.openclaw.com/install.sh | bash
openclaw onboard --install-daemon

Requires Node.js 22+. The installer handles dependencies.

2 Open the Control UI

Once installed, open http://127.0.0.1:18789 in your browser. This is your OpenClaw dashboard — same interface you used in the sandbox, but now it's yours.

3 Add your API key

Go to Settings → Model Configuration and enter your API key. If you exported your config, your model preference is noted in the export file.

4 Connect your real channels

Now you can connect your actual WhatsApp, Discord, Telegram, or other accounts. The sandbox taught you how each channel works — now do it with your real accounts.

⚠️ This is the point of no return for account safety. You're connecting real accounts now. LobsterSandbox safety controls (Kill Switch, Wipe) only work inside the sandbox — they won't protect your production instance.

5 Set up your own cost controls

OpenClaw doesn't have built-in budget limits. Set spending caps directly with your API provider:

• Anthropic: console.anthropic.com → Settings → Spend Limits
• OpenAI: platform.openai.com → Settings → Limits
• OpenRouter: openrouter.ai → Settings → Credit Limit

6 Run a health check

openclaw doctor

This verifies your installation. Fix any issues it flags before going further.

Step 3: Deploy to the Cloud (Optional)

If you want your agent running 24/7, deploy to a cloud provider.

🚂

Railway

One-click deploy. $5/mo.

🌊

DigitalOcean

1-Click Droplet. $24/mo.

🐳

Docker/VPS

Any Linux server with Docker.

For detailed deployment instructions, see the official OpenClaw docs at docs.openclaw.com

You don't have to choose

Many users keep their sandbox running alongside their production setup. Use the sandbox to test new skills, try risky configurations, or experiment with new channels before connecting them to your real accounts. That's what it's there for.

← Back to Your Sandbox

How was your LobsterSandbox experience?

We'd love to hear how the sandbox helped you get started with OpenClaw.

You can also open issues or contribute at github.com/StartIQ-LLC/lobstersandbox

← Back to homepage