OpenClaw Quick Start Guide
This guide gets you from zero to your first conversation with your self-hosted OpenClaw AI agent in about 5–10 minutes. It uses the official one-line installer on macOS or Linux (and Windows via WSL2), then walks you through the onboarding wizard to choose an AI model and connect a messaging channel such as Telegram, WhatsApp, or Discord.
Prerequisites
Before starting, make sure you have:
- ✅ Operating System: macOS 12+, Ubuntu 20.04+, Debian 11+, or Windows 10+ (with WSL2)
- ✅ Node.js 22+: OpenClaw requires Node.js 22 or newer. Check with
node -v; install from nodejs.org or your package manager. The one-line installer can install it for you on supported systems. - ✅ Internet connection: Required for the installer, LLM API calls, and channel connections
- ✅ Terminal access: Command line / bash shell
- ✅ LLM API Key: At least one from:
- Anthropic (Claude) - Get key (recommended for OpenClaw)
- OpenAI (GPT) - Get key
- DeepSeek - Get key (cost-effective)
- Ollama (local) - no API key; runs models on your machine
- ✅ Messaging app account: For your first channel - Telegram (easiest), WhatsApp, or Discord are popular choices.
Step 1: Install OpenClaw (30 seconds)
Open your terminal and run this one-line installer:
curl -fsSL https://openclaw.ai/install.sh | bash
- Detects your operating system automatically
- Installs Node.js 22+ if not already installed
- Downloads the latest OpenClaw version
- Creates
~/.openclawdirectory structure - Launches the onboarding wizard
For a video walkthrough, see our Video Tutorials.
For Windows users: First install WSL2, then run the command inside your WSL terminal. For a full Windows walkthrough, see Install OpenClaw on Windows (WSL2).
Alternative manual installation (if you prefer cloning the repo):
git clone https://github.com/openclaw/openclaw.git
cd openclaw
npm install
npm run setup
For platform-specific details (Homebrew on macOS, systemd on Linux, Docker, or VPS), see the Installation hub: macOS, Windows, Linux, Docker, VPS.
Step 2: Run Onboarding Wizard (3 minutes)
After installation, the onboarding wizard launches automatically. If it doesn't, run:
openclaw setup
The wizard will guide you through:
2.1 Choose AI Model Provider
Select your preferred LLM:
- Claude (Anthropic) - Most popular among OpenClaw users due to strong instruction-following
- GPT (OpenAI) - ChatGPT models; excellent general capabilities
- DeepSeek - Cost-effective alternative
- Ollama (Local) - Run models locally (no API cost)
2.2 Enter API Key
Paste your API key when prompted. The key is stored securely in ~/.openclaw/credentials/ with 600 permissions.
2.3 Configure Your First Channel
Choose a messaging platform to connect. Step-by-step channel setup guides:
- Telegram - Easiest; no phone number required for the bot
- WhatsApp - Popular; requires QR code scan to link your account
- Discord - Great for communities; create a bot in Discord Developer Portal
2.4 Personality Configuration
The wizard creates your agent's SOUL.md file (personality and behavior instructions). You can customize this later.
Step 3: Start the Gateway (10 seconds)
Start OpenClaw as a background service:
# Start the gateway daemon
openclaw start
# Check status
openclaw status
# View logs
openclaw logs
Enable auto-start on boot:
On macOS:
openclaw service install --launchd
launchctl load ~/Library/LaunchAgents/com.openclaw.gateway.plist
On Linux:
sudo openclaw service install --systemd
sudo systemctl enable openclaw
sudo systemctl start openclaw
For a full post-install checklist (security, updates, backups), see Post-Installation Setup and Security best practices.
Step 4: Send Your First Message!
Open your connected messaging app (Telegram/WhatsApp/Discord) and send a message to your agent:
Try These Commands:
Hello! Who are you?
What can you help me with?
List all your available skills
Tell me a joke
Test Autonomous Actions:
Create a file called test.txt with "Hello World"
Show me my system info
What files are in my home directory?
Check my current time
Next Steps
- Complete post-install: Post-Installation Setup - security checklist, auto-start, and optional tuning.
- Add more channels: Connect WhatsApp, Discord, Slack, Signal, and more.
- Install skills: ClawHub skills marketplace - browse and install community skills to extend your agent.
- Configure memory: Customize persistent context so your agent remembers across sessions.
- Explore use cases: Use cases library - 50+ automation examples for productivity, business, and development.
- Secure your setup: Security best practices - binding, credentials, and hardening.
All installation methods are listed on the Installation hub (macOS, Windows, Linux, Docker, VPS, Raspberry Pi).
Common First-Time Issues
| Issue | Solution |
|---|---|
| Agent not responding | Check openclaw status and openclaw logs |
| API key invalid | Verify key at provider's console, re-run openclaw setup |
| Channel disconnected | QR code expired - restart setup for that channel |
| Node.js version error | OpenClaw requires Node.js 22+. Update with nvm, brew upgrade node (macOS), or install from nodejs.org. |
| Gateway won't start / unclear error | Run openclaw doctor (and openclaw doctor --fix if suggested) to diagnose. See Quick diagnostics and Installation errors. |
Other Installation Methods
If you need a different path (e.g. native macOS with Homebrew, Windows WSL2 step-by-step, Docker, or VPS), use these guides from the Installation hub:
- Install OpenClaw on macOS - Apple Silicon & Intel, ~10 min
- Install OpenClaw on Windows (WSL2) - ~15 min
- Install OpenClaw on Linux - Ubuntu, Debian, Fedora, Arch
- Deploy OpenClaw with Docker - Portable, VPS/homelab
- VPS Deployment - AWS, DigitalOcean, Linode, Hetzner
- Run OpenClaw on Raspberry Pi - Low-cost always-on