How to Install OpenClaw: Step-by-Step Guide 2026

OpenClaw has become the fastest-growing open-source project in GitHub history. In this guide, we’ll walk through the complete installation process — from zero to a working AI agent connected to your messaging app. Every command here is real and tested.

Before we start, a security note: This guide covers installation for development and testing purposes. For business deployment, read our enterprise security guide and consider GDPR-compliant setup on Hetzner before going live.

Install Klawty Instead (Recommended)

If you’re deploying for business use — not just experimenting — we recommend installing Klawty instead of raw OpenClaw. Klawty is d-code’s production-hardened agent OS built on the OpenClaw foundation. Same skills, same messaging integrations, same LLM providers — but with enterprise security, PII protection, and governance built in.

One-command install:

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

Klawty’s installer handles Node.js version checking, security configuration, Docker sandbox setup, and daemon installation automatically. What takes 10 steps with raw OpenClaw takes one command with Klawty.

Why Klawty over raw OpenClaw?

• Deny-by-default security — agents only access what you explicitly allow
• PII router — personal data never reaches cloud LLMs
• Docker sandbox — each agent runs in isolation
• Pre-vetted skills — no malicious packages
• Automatic security updates — no manual cron jobs needed
• Same OpenClaw interface — WhatsApp, Telegram, Discord all work the same

Or skip installation entirely: Use the AI Agent Builder managed platform — deploy agents from a web dashboard, no CLI required.

If you still want to install raw OpenClaw (for experimentation, contributing to the project, or testing), continue with the guide below.

Prerequisites

You’ll need:

1. Node.js 22 or higher — OpenClaw requires Node 22+
2. npm — comes bundled with Node.js
3. An LLM API key — OpenAI, Anthropic (Claude), DeepSeek, or local via Ollama
4. A messaging app — WhatsApp, Telegram, Discord, Slack, Signal, or 20+ others

Step 1: Install Node.js 22+

We recommend using nvm (Node Version Manager) for easy version management.

On macOS / Linux:

# Install nvm
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash

# Reload your shell
source ~/.bashrc   # or source ~/.zshrc on macOS

# Install Node 22
nvm install 22
nvm use 22

# Verify
node --version   # Should show v22.x.x
npm --version

On Windows: Download and run the Node.js 22 LTS installer from nodejs.org. After installation, open a new PowerShell window and verify:

node --version
npm --version

On Windows (WSL2 — recommended for production-like setup):

# In WSL2 Ubuntu terminal
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
source ~/.bashrc
nvm install 22
nvm use 22

Step 2: Install OpenClaw

With Node.js ready, install OpenClaw globally:

npm install -g openclaw@latest

This downloads and installs the latest version of OpenClaw as a global CLI tool. The -g flag makes it available system-wide.

Verify the installation:

openclaw --version

You should see a version string in the format vYYYY.M.D (e.g. v2026.3.15).

Important: Always install the latest version. Versions before v2026.1.30 are affected by CVE-2026-25253 (one-click RCE vulnerability, CVSS 8.8). The patch was released on January 29, 2026.

Step 3: Get Your LLM API Key

OpenClaw needs a language model to power its reasoning. Choose one:

Option A: Anthropic (Claude) — Recommended

1. Go to console.anthropic.com
2. Create an account and add a payment method
3. Navigate to API Keys → Create Key
4. Copy your key: sk-ant-api03-...

Option B: OpenAI (GPT-4)

1. Go to platform.openai.com
2. Create an account and add billing
3. Navigate to API Keys → Create new secret key
4. Copy your key: sk-proj-...

Option C: Ollama (Local — Free, GDPR-friendly)

Run models locally — no API key needed, no data leaves your machine.

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model (choose based on your hardware)
ollama pull llama3          # 8B, ~5GB, good for most tasks
ollama pull mistral         # 7B, ~4GB, fast and capable
ollama pull deepseek-r1:7b  # 7B, strong reasoning

# Verify Ollama is running
ollama list

For businesses with GDPR concerns, Ollama is the most privacy-preserving option — all processing happens on your own hardware.

Step 4: Run the Onboarding Wizard

OpenClaw has a guided setup process:

openclaw onboard --install-daemon

The wizard will ask you several questions. Here’s what to expect:

Choose Your LLM Provider

? Select your AI provider:
❯ Anthropic (Claude)
  OpenAI (GPT-4)
  DeepSeek
  Ollama (local)
  Other (custom endpoint)

Select your provider and enter your API key when prompted. If using Ollama, make sure it’s running first (ollama serve).

Connect a Messaging App

? Which messaging platform would you like to use?
❯ WhatsApp
  Telegram
  Discord
  Slack
  Signal
  (more options...)

Connecting WhatsApp: After selecting WhatsApp, a QR code will appear in your terminal. Open WhatsApp on your phone → Settings → Linked Devices → Link a Device. Scan the QR code. Once linked, OpenClaw will show:

✓ WhatsApp connected successfully

Connecting Telegram: Select Telegram, then follow the instructions to create a bot via BotFather. You’ll paste the bot token when prompted.

Connecting Discord: You’ll need to create a Discord application and bot at discord.com/developers. Paste the bot token when prompted.

Install as a System Daemon

The --install-daemon flag sets OpenClaw up to start automatically when your system boots:

? Install OpenClaw as a system service? (Recommended)
❯ Yes — runs in background, starts on boot
  No — I'll start it manually

Select Yes for a persistent setup. On Linux, this creates a systemd service. On macOS, it creates a LaunchAgent.

Step 5: Verify the Installation

Once onboarding completes, test your agent by sending a message through your connected app:

You: Hello, are you running?
OpenClaw: Yes! I'm your OpenClaw agent. I'm ready to help. What would you like me to do?

Try a simple task:

You: What's the current time and date?
OpenClaw: It's currently 14:32 on Wednesday, March 15, 2026.

Then something more interesting:

You: Search the web for the latest news about AI agents
OpenClaw: [Browsing the web...]
Here are the top stories about AI agents from the past 24 hours: ...

Step 6: Install Your First Skills

Skills extend what OpenClaw can do. Browse them at github.com/openclaw/clawhub.

Install a skill via OpenClaw:

You: Install the Gmail skill
OpenClaw: Installing Gmail skill from ClawHub...
✓ Installed. To connect Gmail, go to Settings → Skills → Gmail and authorize access.

Or via CLI:

openclaw skills install gmail
openclaw skills install calendar
openclaw skills install seo-research

Security warning: Before installing any skill, check its GitHub repository. Only install skills with recent commits, visible source code, and trusted publishers. Avoid skills with generic names and no documentation — these are common patterns in the 1,184 malicious skills identified by security researchers.

Configuration File

OpenClaw stores its configuration in ~/.openclaw/config.json. Key settings you may want to adjust:

{
  "llm": {
    "provider": "anthropic",
    "model": "claude-3-5-sonnet-20241022",
    "maxTokens": 4096
  },
  "security": {
    "requireConfirmation": true,
    "allowedPaths": ["/home/user/work"],
    "blockedCommands": ["rm -rf", "format", "dd"]
  },
  "messaging": {
    "platform": "whatsapp",
    "autoReply": true
  }
}

Always set requireConfirmation: true for destructive operations — this adds a confirmation step before OpenClaw executes file deletions, emails, or external API calls.

Updating OpenClaw

Stay current — OpenClaw releases updates frequently, often with security patches:

npm update -g openclaw
openclaw --version

Set up a weekly reminder to run this command. Better yet, for a VPS deployment, use a cron job:

# Add to crontab (crontab -e)
0 3 * * 1 npm update -g openclaw  # Every Monday at 3am

Running OpenClaw on a VPS (Business Setup)

For business use, never rely solely on a laptop installation. A VPS ensures your OpenClaw agent is always available, even when your computer is off.

Recommended: Hetzner Cloud (EU servers, GDPR-compliant, excellent price/performance):

1. Create a Hetzner account
2. Create a new server: Ubuntu 22.04, CX21 (2 vCPU, 4GB RAM — sufficient for most setups)
3. SSH into your server
4. Follow the Node.js and OpenClaw installation steps above
5. Configure a firewall: only allow SSH (port 22) and your VPN IP — never expose OpenClaw ports publicly

# Hetzner firewall basics
ufw allow 22/tcp
ufw enable
ufw status

We cover the full GDPR-compliant VPS setup in our Hetzner deployment guide.

Troubleshooting Common Issues

”Command not found: openclaw”

Node global binaries aren’t in your PATH. Fix:

export PATH="$PATH:$(npm bin -g)"
# Add the above to your ~/.bashrc or ~/.zshrc

WhatsApp QR code expires before scanning

The QR code has a 60-second window. Run openclaw onboard again and scan faster. Make sure your phone has a stable internet connection.

”LLM API key invalid”

Double-check your key. Anthropic keys start with sk-ant-, OpenAI keys start with sk-proj- or sk-. Check your provider’s dashboard for any billing issues.

OpenClaw daemon not starting on boot

# Check systemd service status (Linux)
systemctl status openclaw
journalctl -u openclaw -n 50

# Restart the service
systemctl restart openclaw

Node version mismatch

node --version   # Must be 22+
nvm use 22       # Switch to Node 22
npm install -g openclaw@latest  # Reinstall

What’s Next

Now that OpenClaw is running, here’s what to explore:

• OpenClaw for SMEs — real workflow automation examples
• OpenClaw & GDPR: Hetzner Setup — for business-grade, GDPR-compliant deployment
• OpenClaw for Enterprise — security hardening before production use
• Best OpenClaw Skills for Agencies — the most useful skills for digital businesses

Ready to go beyond experimentation? When you’re ready to deploy agents for real business use, upgrade to Klawty — the production-ready agent OS built on OpenClaw. All your existing skills and configurations carry over.

Want it managed? The AI Agent Builder platform handles hosting, security, updates, and monitoring — so you can focus on building workflows, not maintaining infrastructure.

Need help with your deployment? d-code provides full implementation, security hardening, and ongoing support. Book a free consultation.