Step-by-Step Tutorial — Updated February 2026

How to Set Up OpenClaw
Complete Installation Guide

A detailed step-by-step walkthrough of installing, configuring, and securing OpenClaw on your hardware. From zero to running in 30 minutes.

~30 minutes
8 Steps
Beginner Friendly
OpenClaw Setup Process — Step-by-step installation and configuration guide

Before You Start

A computer with 4GB+ RAM (8GB+ for local models)
macOS, Linux, or Windows 11 with WSL2
An AI API key (Claude, OpenAI, or Gemini) — or Ollama for local
A messaging app account (WhatsApp, Telegram, Slack)
Basic comfort with the command line / terminal
30 minutes of uninterrupted time

Not comfortable with the command line? No problem — we offer professional OpenClaw setup services where our team handles everything for you.

1

Choose Your Hardware

The first decision is what hardware to run OpenClaw on. Your choice depends on your budget, whether you need always-on availability, and whether you plan to run local AI models.

Mac Mini M4

16GB RAM, M4 chip

Best for local AI models & performance

~$899 AUD

Cloud VPS

4-8GB RAM, 2+ vCPUs

Best for always-on & remote access

~$35/mo AUD

Raspberry Pi 5 (8GB)

8GB RAM, ARM CPU

Best budget dedicated device

~$120 AUD

Existing Desktop/Laptop

4GB+ RAM, any modern CPU

Best for trying it out

Free (use what you have)

Tip: If you're just testing OpenClaw, use your existing machine. For production use, a dedicated Mac Mini M4 or VPS is recommended.

2

Install Prerequisites

OpenClaw requires Node.js version 22.12.0 or later. This specific version is critical for security patches. Here's how to install it on each platform:

macOS / Linux (using nvm)
# Install nvm if you don't have it
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash

# Install Node.js 22
nvm install 22
nvm use 22

# Verify version (must be 22.12.0+)
node --version
Windows 11 (WSL2 required)
# First, ensure WSL2 is installed
wsl --install

# Then inside WSL2 terminal:
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash
nvm install 22
node --version

Important: PowerShell is NOT supported. Windows users must use WSL2 (Windows Subsystem for Linux).

3

Install OpenClaw

OpenClaw can be installed with a single command. The installer will guide you through the initial configuration.

Run the OpenClaw installer
npx openclaw@latest

The installer will present several options. For your first setup, we recommend:

Installation type:QuickStart (safe defaults)
AI Provider:Claude API (recommended) or OpenAI
Messaging platform:WhatsApp or Telegram (easiest to set up)

QuickStart sets up OpenClaw with safe defaults. It's the recommended option for beginners and configures basic security automatically.

4

Configure Your AI Provider

OpenClaw needs an AI model to power its intelligence. You can use cloud APIs or run models locally with Ollama.

Claude API (Anthropic)

Best overall performance. Excellent at following instructions and taking actions safely.

Get API key from console.anthropic.com

OpenAI GPT

Strong general performance. Good for coding and creative tasks.

Get API key from platform.openai.com

Google Gemini

Good performance with generous free tier.

Get API key from ai.google.dev

Ollama (Local)

100% offline. No API costs. Requires 8GB+ RAM.

Install from ollama.com, then pull a model

Configure provider in OpenClaw config
# OpenClaw config is typically at ~/.openclaw/config.yaml
# The installer sets this up, but you can edit it:

provider: claude
api_key: sk-ant-your-key-here

# Or for local Ollama:
# provider: ollama
# model: llama3.2
5

Connect Your Messaging Platform

One of OpenClaw's best features is that you control it from whatever messaging app you already use. Here's how to set up the most popular options:

WhatsApp

  1. Enable WhatsApp connector in OpenClaw config
  2. Scan the QR code shown in your terminal with WhatsApp
  3. Send a test message to verify the connection

Telegram

  1. Create a new bot via @BotFather on Telegram
  2. Copy the bot token to your OpenClaw config
  3. Start a chat with your bot and send a test message

Slack

  1. Create a new Slack App in your workspace
  2. Configure the OAuth scopes and bot tokens
  3. Add the credentials to your OpenClaw config
6

Harden Security

Critical: OpenClaw is experimental software. Do NOT skip security hardening. The ClawHavoc campaign exposed 341 malicious skills — take security seriously.

Essential security steps for any OpenClaw installation:

Enable Token Authentication

Generate a 48-character hex token for all non-loopback connections

Run in Isolated Environment

Use a dedicated user account or Docker container. Never run as root.

Configure Firewall Rules

Only expose required ports. Block all unnecessary inbound traffic.

Encrypt All Connections

Use SSH tunneling or Tailscale for remote access. Never expose plain HTTP.

Vet ClawHub Skills

Always review source code and community reviews before installing any skill.

Generate auth token
# Generate a secure random token
openssl rand -hex 24

# Add to your OpenClaw config:
# auth_token: your-generated-token-here
7

Install ClawHub Skills

ClawHub hosts over 3,000 community-built skills. Here are the most popular and useful skills to install first:

Email Manager— Triage, draft, and send emails
Calendar— Schedule, update, and manage events
Web Browser— Browse and extract web content
File Manager— Organise, search, and manage files
Git Manager— Git operations and repo management
Note Taker— Create and organise notes

Security reminder: Always vet skills before installing. Check the source code, review count, and author reputation on ClawHub. Never install skills from unknown or unverified sources.

8

Set Up Remote Access

To access OpenClaw from outside your local network, you need to configure secure remote access. Here are the recommended methods (from simplest to most advanced):

Method 1: SSH Tunneling (Simplest & Most Secure)

# On your local machine, create an SSH tunnel:
ssh -L 18789:localhost:18789 user@your-server-ip

# Then access OpenClaw at:
# http://localhost:18789

Method 2: Tailscale VPN (Recommended for Teams)

Install Tailscale on both your server and client device. OpenClaw will be accessible via your Tailscale IP — no port forwarding or firewall changes needed.

Method 3: Nginx Reverse Proxy (Advanced)

For production deployments, use Nginx as a reverse proxy with SSL termination via Let's Encrypt. This provides HTTPS access with proper certificate management.

Congratulations! OpenClaw is Running

Your AI personal assistant is now installed and configured. Here are some next steps to get the most out of OpenClaw:

Explore ClawHub

Browse 3,000+ skills to extend OpenClaw's capabilities

Set Up Routines

Create morning briefings, email summaries, and scheduled tasks

Connect More Tools

Add n8n workflows, CRM integrations, and custom APIs

More OpenClaw Resources

Complete OpenClaw Guide

Everything you need to know about OpenClaw — history, features, security, and business use cases.

Read Guide

Professional Setup Service

Let our team handle the installation, security hardening, and configuration. Packages from $490.

View Packages

AI Automation

Combine OpenClaw with n8n and custom AI workflows for enterprise-grade automation.

Learn More

Need Help With Your Setup?

Our Australia team can set up OpenClaw on your hardware in one business day — fully secured, integrated, and ready to use.