OpenClaw Installation & Usage
This chapter walks you through installation, initialization, and your first conversation via the Dashboard and chat apps.
System Requirements
- OS: macOS, Linux, Windows
- Runtime: Node.js 22+
- Hardware: minimum 2GB RAM / 2 cores (basic chat); 4GB+ recommended for browser automation
- Runs on old laptops, Docker containers, or a $5/month VPS
One-Line Install
macOS / Linux:
Windows (PowerShell):
The installer comes from the network — understand what it does before running. For production, try it first on an isolated machine.
Initialize: onboard
After installing, run the onboarding wizard and install the persistent daemon:
The wizard guides you through the model provider and keys, the chat platforms to connect, basic tool policies, and more.
Check the gateway status:
Open the Dashboard
OpenClaw ships a local web console:
Then open in your browser:
The control plane listens on WebSocket
:18789. Do not expose this port to the public internet, and keep it updated (see Security).
To run the gateway on a custom port:
Your First Conversation
After onboarding you can chat with OpenClaw two ways:
1) Dashboard (fastest to verify): type directly into the chat box at http://127.0.0.1:18789/, e.g.:
2) Chat app: message your bot on a connected platform (e.g., Telegram). See Message Channels.
Try a real task:
For risky actions like writes/deletes, OpenClaw asks for your approval per the tool policy before executing.
Common Commands
Commands vary by version; run
openclaw --helpfor the current full list.
Deploy as a Persistent Service
To keep OpenClaw online 24×7 (great for a VPS):
--install-daemoninstalls a system daemon- Or keep the gateway running with
systemd/pm2/ Docker - Ensure auto-restart on crash, and periodically
git-back up~/.openclaw
Updating
OpenClaw iterates fast and has had a critical vulnerability — keep it updated to the latest (at least 2026.1.29 or newer, see Security). Use the update command per the official repo.
Next Steps
- Model Configuration — connect your LLM
- Message Channels — connect Telegram / WhatsApp
- Hands-on Practice — build a self-running assistant