⚡ 10 Minute Setup

Quick Start Guide

From zero to a working AI agent. Follow these steps and you'll be chatting with your personal assistant in minutes.

1
Requirements
2
Install
3
Configure
4
Connect
5
Launch
1

Check Requirements

Before you start, make sure you have:

💻

macOS, Linux, or Windows

OpenClaw runs on macOS, Linux, native Windows, and WSL2. WSL2 is still the steadier Windows path for headless/server setups.

📦

Node 24 recommended

The installer handles Node for you. If you manage Node yourself, use Node 24 recommended or Node 22.19+ minimum.

🔑

Model access

Use an API key, OAuth-backed provider, Claude CLI, OpenAI Code/Codex subscription, Ollama/LM Studio, or another OpenAI-compatible endpoint.

📱

Messaging App (Optional)

Telegram is fastest; WhatsApp, Discord, Slack, Matrix, Signal, iMessage, QQ Bot, and more are also supported.

💡
Don't have an API key yet?

Start with whichever access you already have: Anthropic API key, OpenAI/Codex OAuth, Google Gemini, OpenRouter, or a local Ollama/LM Studio model. Onboarding walks you through the safe path.

2

Install OpenClaw

Use the official installer. It detects your OS, handles Node if needed, installs OpenClaw, and can launch onboarding:

curl -fsSL https://openclaw.ai/install.sh | bash

This installs the openclaw command. Verify it worked:

openclaw --version
Expected Output
OpenClaw 2026.6.6
📦 Alternative: npm/pnpm/bun or Docker

If you manage Node yourself, npm is still supported; containers are available for headless deployments:

npm install -g openclaw@latest
openclaw onboard --install-daemon

# or see the Docker guide for container deployments
3

Run the Setup Wizard

The onboard wizard guides you through configuration:

openclaw onboard

The wizard will ask you for:

a
Your Name

How your agent should refer to you

b
Model/Auth

Choose API key, OAuth, Claude CLI, Codex subscription, custom provider, Ollama, or LM Studio

c
Agent Personality

Pick a preset or create custom (you can change this later)

d
Channels, daemon, and skills

Connect a chat app, install the per-user service, and pick useful skills

📄 What Gets Created

The wizard creates these files in ~/.openclaw/:

📁 ~/.openclaw/
📄 openclaw.json — Main configuration
📁 workspace/
📄 AGENTS.md — Operating manual
📄 SOUL.md — Agent personality
📄 USER.md — About you
📄 MEMORY.md — Long-term memory
4

Connect a Messaging Channel

Chat with your agent through your favorite messaging app. Pick one:

📱

Telegram

Recommended ▼
  1. Create a bot with @BotFather:

    Open Telegram, search for @BotFather, send /newbot, follow prompts

  2. Copy the bot token (looks like 123456789:ABC-DEF...)
  3. Configure Telegram: 
    openclaw configure --section channels
  4. Get your user ID: Message @userinfobot on Telegram
🎮

Discord

▼
  1. Create a Discord Application:

    Go to Discord Developer Portal → New Application

  2. Create a Bot: Go to Bot tab → Add Bot → Copy token
  3. Enable Intents: Enable MESSAGE CONTENT INTENT in Bot settings
  4. Invite to server: OAuth2 → URL Generator → bot + applications.commands scopes → Send Messages permission → Copy URL → Open in browser
  5. Configure Discord: 
    openclaw configure --section channels
🔐

Signal

Privacy-focused ▼
  1. Install signal-cli: 
    brew install signal-cli  # macOS
# or download from GitHub
  2. Register a phone number: Follow signal-cli setup docs
  3. Configure Signal: 
    openclaw configure --section channels
⌨️

CLI Only (No Messaging)

▼

Don't want to set up messaging yet? You can use the terminal:

openclaw chat

This opens an interactive chat in your terminal. Perfect for testing!

5

Launch Your Agent

Onboarding can install and start the daemon for you. Check status, then open the local dashboard or terminal chat:

openclaw gateway status
openclaw dashboard
Expected Output
OpenClaw Gateway: running
Control UI: http://127.0.0.1:18789
Open the dashboard or send a message from your configured channel.
🔄 Run in Background

The gateway runs as a daemon by default. Check status with:

openclaw gateway status
📋 View Logs

See what your agent is doing:

openclaw logs
🛑 Stop the Gateway
openclaw gateway stop
🎉

You're Live!

Send a message to your bot and watch the magic happen. Your agent is now ready to help you.

Try saying:

"Hey! What can you do?"
"What's the weather like today?"
"Help me brainstorm ideas for..."
"Remember that I prefer..."

What's Next?

📁

Customize Personality

Edit SOUL.md and IDENTITY.md to give your agent a unique personality.
🎭

Explore Personas

Browse pre-built personalities from helpful assistant to creative partner.
🛠️

Add Skills

Install skills for calendar, weather, email, and more from ClawHub.
⚙️

Set Up Automation

Create cron jobs for daily briefings, reminders, and background tasks.
📱

Connect Devices

Pair your phone or other devices for camera, location, and more.
📖

Read the Docs

Full documentation for all features, configuration, and advanced usage.

Troubleshooting

❌ "Command not found: openclaw"

Run openclaw doctor. If the CLI is missing, rerun the installer or install with npm install -g openclaw@latest if you manage Node yourself.

❌ "API key invalid" or "401 Unauthorized"

Run openclaw models status or openclaw configure --section model. Config now lives in ~/.openclaw/openclaw.json, with credentials in auth profiles/secret refs where possible.

❌ Bot not responding to messages

  • Check the gateway is running: openclaw gateway status
  • Check logs for errors: openclaw logs
  • Make sure your user ID is in the allowlist
  • For Telegram: make sure you started a chat with the bot first

❌ "Permission denied" errors

Avoid sudo npm install -g when possible. Use the installer script or a local npm/pnpm/bun prefix, then run openclaw doctor.

Still stuck? Get help:

💬 Discord Community 🐛 GitHub Issues