How to Complete OpenClaw Onboarding (Step-by-Step Guide)

New to OpenClaw? This beginner-friendly guide explains how to complete OpenClaw onboarding from terminal setup to model configuration and channel connection, so your AI agent is ready to run.

Do not index
notion image
If you’re setting up OpenClaw for the first time inside your workspace at Agent37, the onboarding process might look slightly technical at first. There’s a terminal. There are model providers. There are channel selections. It can feel like you’re configuring something much bigger than a simple chat tool.
But once you understand the flow, it’s actually structured and logical.
In this guide, we’ll walk step-by-step through how to complete OpenClaw onboarding, following the exact onboarding sequence shown in the walkthrough — from launching the setup wizard to verifying your agent is fully running.
You can access your OpenClaw dashboard at:
Let’s begin.

Step 1: Log Into Your Agent37 Dashboard

After logging into your account at https://www.agent37.com/, navigate to:
OpenClaw → Instances
You’ll see:
  • Instance ID
  • Tier (Basic, etc.)
  • Status (Running)
  • Starter credits
  • Actions: Chat, Terminal, Restart
Before starting onboarding, confirm your instance shows:
Status: Running
If it isn’t running, start or restart the instance before proceeding.

Step 2: Open the Terminal

On your instance row, click:
Terminal
This opens the CLI interface.
To begin onboarding, type:
openclaw onboard
Press Enter.
This launches the interactive onboarding wizard — and this is the true starting point of how to complete OpenClaw onboarding correctly.

Step 3: QuickStart Configuration (Gateway Settings)

The first section you’ll see is:
Config Handling → QuickStart
OpenClaw shows your current gateway configuration:
  • Gateway port
  • Gateway bind (Loopback 127.0.0.1)
  • Gateway auth token
  • Tailscale exposure (usually Off)
For most first-time users, keeping the default settings is perfectly fine.
Unless you’re deploying in a custom network environment, you can proceed without changes.
This step ensures your instance can safely communicate internally before external integrations are connected.

Step 4: Select Your Model / Auth Provider

Now you’ll be prompted to choose your AI model provider.
You’ll see options such as:
  • OpenAI (Codex OAuth + API key)
  • Anthropic
  • MiniMax
  • Moonshot AI (Kimi K2.5)
  • Google
  • XAI (Grok)
  • OpenRouter
  • Qwen
  • GLM 4.7
  • Copilot
  • Vercel AI Gateway
  • Venice AI
  • And others
Choose the provider you intend to use.
Most beginners start with:
  • OpenAI
    • or
  • Anthropic
You’ll then be asked to:
  • Enter your API key
  • Confirm authentication
This is one of the most critical parts of how to complete OpenClaw onboarding, because without a configured model provider, your agent cannot generate responses.
If you skip this or enter an invalid key, the agent will not function properly.

Step 5: Select Your Messaging Channel

Next comes:
Select channel (QuickStart)
You’ll see a list of supported integrations:
  • Telegram (Bot API)
  • WhatsApp (QR link)
  • Discord
  • Google Chat
  • Slack
  • Signal
  • iMessage
  • Feishu/Lark
  • Microsoft Teams
  • Matrix
  • LINE
  • Zalo
  • And more
In the walkthrough example, WhatsApp was selected.
You can:
  • Select a messaging platform now
  • Or choose “Skip for now”
If you select WhatsApp, you’ll connect using a QR link.
If you select Telegram, you’ll configure via Bot API.
This step defines how users will communicate with your OpenClaw agent.

Step 6: Skills Status & Configuration

After channel selection, you’ll see:
Skills Status
This section shows:
  • Eligible skills
  • Missing requirements
  • Allowlist status
Then you’ll be asked:
Configure skills now? (recommended)
For a basic setup, you can safely choose:
No (Skip for now)
Skills can be configured later once the agent is running.
It’s better to get your agent live first, then enhance functionality.

Step 7: Hooks Configuration

Next, you’ll see:
Hooks
Hooks allow automated actions when agent commands are issued.
Examples include:
  • boot-md
  • command-logger
  • session-memory
Hooks are useful for advanced automation and logging.
For beginners, skipping hooks during initial onboarding is completely fine.
You can enable them later once you’re comfortable with the system.

Step 8: Final Initialization

Once all steps are completed:
  • openclaw.json updates
  • Workspace directories initialize
  • Sessions are created
You’ll then return to your dashboard.
Click:
Chat
If WhatsApp was selected, you may see:
System: WhatsApp gateway connected.
That confirms the integration is active.
Now type a simple message like:
“Hi”
If the agent responds — onboarding was successful.

How to Verify Everything Is Working

You’ve successfully completed how to complete OpenClaw onboarding if:
✔ Instance shows Running
✔ Model provider is configured
✔ No authentication errors appear
✔ Channel shows connected
✔ Chat interface responds
✔ Terminal shows no fatal errors
If something doesn’t work, you can re-run:
openclaw onboard
The wizard allows you to safely reconfigure.

Common First-Time Mistakes

Here are issues beginners often encounter:
  • Forgetting to add API key
• Selecting a model without available credits
• Choosing WhatsApp but not scanning QR
• Skipping model provider step
• Confusing gateway settings with channel settings
If the agent doesn’t respond, revisit the provider configuration first.

What to Do After Onboarding

Once you’ve learned how to complete OpenClaw onboarding, your next steps may include:
  • Fully configuring Telegram
  • Optimizing WhatsApp integration
  • Enabling skills
  • Activating memory hooks
  • Customizing prompts
  • Setting cron jobs
  • Monitoring usage
Onboarding is the foundation — optimization comes next.

Final Words

Completing your first setup can feel technical, especially when a terminal is involved. But OpenClaw’s onboarding wizard is structured intentionally — guiding you through provider selection, channel setup, and optional enhancements in a logical order.
If you follow the sequence carefully:
  1. Run onboarding
  1. Select model
  1. Add API credentials
  1. Choose channel
  1. Confirm configuration
  1. Test in Chat
You’ll have a fully functioning AI agent inside your Agent37 workspace.
Understanding how to complete OpenClaw onboarding properly ensures your agent is secure, authenticated, and ready to interact across messaging platforms.
Once your agent responds to that first “Hi,” you’re officially live.
From there, it’s no longer setup — it’s building.