Your First Agent
This guide walks you through the entire process of getting a working AI agent, from creating your account to sending your first message. By the end, you'll have an agent running 24/7 that you can connect to any messaging platform.
Step 1: Create an Account
- Go to openclawhost.co and click Get Started.
- Sign up with email and password, or use Continue with Google.
- If you used email, check your inbox and click the verification link.
For more details, see Create Your Account.
Step 2: Choose a Plan and Pay
- After signing in, select a plan. Starter ($10/mo) is a good starting point.
- Complete the Stripe checkout with your payment details.
- You'll be redirected to the dashboard.
For help choosing, see Choose a Plan.
Step 3: Wait for Provisioning
After payment, your instance begins provisioning automatically. This typically takes 30 to 60 seconds. You'll see a progress indicator on the dashboard. Once it's done, your instance status will show Running.
If provisioning takes longer than 2 minutes, try refreshing the page. If the status still shows provisioning, check the Troubleshooting section.
Step 4: Connect an AI Provider
Your agent needs an AI model to generate responses. OpenClaw Host supports three providers:
| Provider | Cost | Setup Difficulty |
|---|---|---|
| Gemini CLI (Google) | Free tier available | Easiest — OAuth login, no API key needed |
| Claude Code (Anthropic) | Paid (per token) | Easy — paste a setup token |
| Codex (OpenAI) | Paid (per token) | Easy — OAuth login |
Start with Gemini CLI. It's free (Google offers a generous free tier), requires no credit card on the AI provider side, and connects with a simple Google OAuth flow. You can always add Claude or Codex later.
Quick Setup: Gemini CLI (Fastest Path)
- Go to Instance > Integrations.
- Click Gemini CLI.
- Click Connect with Google and complete the OAuth flow.
- Once connected, the status will show a green indicator.
Quick Setup: Claude Code
- Go to Instance > Integrations.
- Click Claude Code.
- Follow the instructions to generate a setup token from your Anthropic account.
- Paste the token and click Save.
Quick Setup: Codex
- Go to Instance > Integrations.
- Click Codex.
- Click Connect with OpenAI and complete the OAuth flow.
For detailed step-by-step instructions with screenshots, see the AI Providers section.
Step 5: Test Your Agent
Once a provider is connected, your agent is live. Test it:
- Go to Instance > Web Terminal.
- Run the following command to send a test message through the agent:
openclaw agent --message "Hello, are you there?" --deliver - You should see the agent process the message and return a response.
Alternatively, if you've already connected a messaging channel, send a message to your agent through that platform.
Step 6: Connect a Messaging Channel (Optional)
Channels let your agent communicate through messaging platforms. This step is optional — your agent works through the web terminal and API without any channels.
- Go to Instance > Channels.
- Pick a platform (e.g., WhatsApp, Telegram, Discord).
- Follow the on-screen instructions to connect.
Some channels (like WhatsApp) require scanning a QR code from your phone. Others (like Telegram) need a bot token from BotFather. Each channel has its own setup flow — see the Channels section for platform-specific guides.
What's Next?
Your agent is now running. Here are some things you can do next:
- Add more AI providers — Set up model fallback so your agent switches providers automatically if one goes down (Pro plan and above). See AI Providers.
- Connect messaging channels — Link WhatsApp, Telegram, Discord, or any supported platform. See Channels.
- Install skills — Extend your agent with 700+ community skills from ClawdHub. Use the web terminal to browse and install them:
openclaw skills list --eligible
npx clawdhub@latest install <skill-slug> - Configure your agent — Customize personality, behavior, and tools through the OpenClaw configuration files. See Instance Management.