Prerequisites

Before installing OpenClaw, make sure your system meets the following requirements. OpenClaw is designed to run on a wide range of hardware, from powerful desktop machines to modest VPS instances and even a Raspberry Pi.

Requirement	Details
Operating System	macOS 12+, Ubuntu 20.04+ / Debian 11+, Windows 10+ (via WSL2)
Node.js	Version 20 or later (auto-installed by the setup script)
RAM	Minimum 2 GB (4 GB recommended)
Storage	At least 1 GB free disk space
Internet	Required for LLM API calls and messaging integrations
LLM API Key	At least one key from Anthropic (Claude), OpenAI (GPT), or DeepSeek

You will also need an account with at least one messaging platform if you want to interact with OpenClaw through chat (WhatsApp, Telegram, Discord, Slack, or Signal).

Step 1: Install OpenClaw

The fastest way to install OpenClaw is with the official one-liner. Open your terminal and run:

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

This script will detect your operating system, install any missing dependencies (including Node.js if needed), download the latest version of OpenClaw, and set up the basic directory structure at ~/.openclaw.

For Windows users: You will need to install WSL2 (Windows Subsystem for Linux) first, then run the command inside your WSL terminal. Microsoft provides excellent documentation on setting up WSL2 if you have not done so already.

Alternative installation methods:

If you prefer more control over the installation process, you can also install OpenClaw manually:

# Clone the repository
git clone https://github.com/nicepkg/openclaw.git
cd openclaw

Install dependencies
npm install

Start the setup wizard
npm run setup

Step 2: Run the Onboarding Wizard

After installation, OpenClaw will automatically launch its onboarding wizard. If it does not start automatically, you can run it manually:

openclaw setup

The wizard will walk you through the following configuration steps:

Choose Your AI Model — Select which LLM provider you want to use. Claude (Anthropic) is the most popular choice among OpenClaw users due to its strong instruction-following capabilities, but GPT-4 and DeepSeek are also excellent options. You can configure multiple providers and switch between them later.

Enter Your API Key — Paste your API key for the chosen provider. OpenClaw stores this securely in your local configuration file at ~/.openclaw/config.yaml. Your key never leaves your machine except when making API calls to the provider.

Connect a Messaging Platform — Choose how you want to communicate with OpenClaw. The wizard will guide you through the pairing process for your chosen platform. WhatsApp and Telegram are the most commonly used options.

Set Your Preferences — Configure basic preferences like your name, timezone, language, and how proactive you want OpenClaw to be.

Step 3: Verify the Installation

Once the wizard completes, verify that OpenClaw is running correctly:

openclaw status

You should see output similar to:

OpenClaw v2.4.1
Status: Running
Model: Claude 3.5 Sonnet (Anthropic)
Messaging: WhatsApp (connected)
Memory: 0 conversations stored
Skills: 3 built-in skills active
Uptime: 2 minutes

Try sending a simple message through your connected messaging platform: "Hello, are you there?" OpenClaw should respond within a few seconds, confirming that everything is working.

Step 4: Configure Your Environment

OpenClaw's configuration lives in ~/.openclaw/config.yaml. Here are the most important settings you may want to customize:

# Core settings
ai:
  provider: anthropic          # anthropic, openai, deepseek, or local
  model: claude-3-5-sonnet     # specific model to use
  temperature: 0.7             # creativity level (0.0 - 1.0)

Memory settings
memory:
  enabled: true
  retention_days: 30           # how long to keep conversation history
  max_context_tokens: 100000   # context window size

Messaging
messaging:
  platform: whatsapp
  allowed_contacts: []         # empty = allow all, or list specific numbers
  group_mention_required: true # require @mention in group chats

Proactive behavior
proactive:
  enabled: true
  quiet_hours:
    start: "22:00"
    end: "08:00"
  max_daily_messages: 10       # limit unprompted messages per day

Security
security:
  sandbox_mode: false          # enable Docker-based sandboxing
  file_access:
    allowed_paths:
      
"~/Documents"
      
"~/Downloads"
    blocked_paths:
      
"~/.ssh"
      
"~/.gnupg"

Step 5: Install Your First Skills

Skills extend what OpenClaw can do. Browse available skills and install a few to get started:

# List available skills
openclaw skills search

Install some popular skills
openclaw skills install browser-control
openclaw skills install calendar-sync
openclaw skills install smart-home

You can also browse the ClawhHub at clawhhub.com for community-created skills.

Step 6: Running OpenClaw as a Service

For the best experience, you will want OpenClaw running continuously in the background. The recommended approach depends on your operating system:

On macOS:

openclaw service install
openclaw service start

This creates a LaunchAgent that starts OpenClaw automatically when you log in.

On Linux:

sudo openclaw service install --systemd
sudo systemctl enable openclaw
sudo systemctl start openclaw

Using Docker (recommended for servers):

docker run -d \
  --name openclaw \
  --restart unless-stopped \
  -v ~/.openclaw:/root/.openclaw \
  openclaw/openclaw:latest

Common First Tasks

Once OpenClaw is up and running, here are some things to try:

Send it a message like "Summarize the top 5 news stories today" and watch it browse the web, gather information, and send you a concise summary. Ask it to "Monitor this webpage for changes and let me know when the price drops below $50" for ongoing automated monitoring. Try "Create a new folder called 'Project Notes' on my desktop and save a summary of our conversation there" to see its file management capabilities.

The more you interact with OpenClaw, the better it understands your preferences and workflow. Give it a few days, and you will start to see why users describe it as an "AI with hands."

Troubleshooting

If you encounter issues during setup, here are the most common solutions:

OpenClaw will not start: Make sure Node.js 20+ is installed (node --version). If the install script did not set it up correctly, install it manually from nodejs.org.

Cannot connect to messaging platform: Double-check that you completed the pairing process. For WhatsApp, you may need to scan the QR code again. For Telegram, verify your bot token is correct.

Slow responses: This is usually related to your LLM provider. Check your API key has sufficient credits and that you are not hitting rate limits. Consider switching to a faster model if response time is critical.

High memory usage: If OpenClaw is using too much RAM, reduce the max_context_tokens setting in your config file. You can also disable memory retention for less important conversations.

For more detailed troubleshooting, visit the official documentation at docs.openclaw.ai or join the community Discord server.