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

OpenClaw went from zero to 60,000 GitHub stars in 72 hours — and for good reason. It's the first open-source AI agent that genuinely feels like having a personal assistant living inside your messaging apps.

But the viral hype has outpaced the documentation. Getting OpenClaw running smoothly requires navigating a few gotchas that most tutorials skip. This guide covers everything — from bare-metal installation to Docker deployment, with the security steps that too many people are leaving out.

---

Prerequisites

Before you start, you'll need:

• An LLM API key — OpenClaw needs a language model backend. I recommend Claude (Anthropic) for the best results, but GPT-5.2 (OpenAI), DeepSeek, or a local model via Ollama all work. See our OpenClaw + Claude guide for setup details.
• Node.js 20+ or Docker installed on your machine
• A messaging account on at least one supported platform (Telegram is the easiest to start with)

---

Method 1: Quick Install (Mac/Linux)

The fastest path to a running OpenClaw instance:

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

# Install dependencies
npm install

# Run the setup wizard
npm run setup

The setup wizard walks you through:

1. LLM provider selection — Choose Claude, OpenAI, DeepSeek, or Ollama
2. API key configuration — Enter your credentials
3. Gateway setup — Pick your first messaging platform
4. Security defaults — Set API spending limits and action approval rules

Once the wizard completes, start the agent:

npm start

That's it. Your agent is running. Send it a message on your configured platform and watch it respond.

---

Method 2: Docker (Recommended for Production)

Docker is the recommended approach for always-on deployments:

docker run -d \
  --name openclaw \
  --restart unless-stopped \
  -v openclaw-data:/app/data \
  -p 3100:3100 \
  openclaw/openclaw:latest

Then open http://localhost:3100 to complete the web-based setup wizard.

Docker Compose (Full Setup)

For a more robust deployment:

version: '3.8'
services:
  openclaw:
    image: openclaw/openclaw:latest
    restart: unless-stopped
    volumes:
      - openclaw-data:/app/data
    ports:
      - "3100:3100"
    environment:
      - ANTHROPIC_API_KEY=your-key-here
      - OPENCLAW_LOG_LEVEL=info
volumes:
  openclaw-data:

---

Method 3: Raspberry Pi Installation

One of the most popular setups — a dedicated $35 AI assistant running 24/7 on your home network.

Hardware You'll Need

• Raspberry Pi 5 8GB — The 8GB model handles OpenClaw comfortably
• A quality microSD card or USB SSD — SSD recommended for better performance
• A USB-C power supply (27W recommended)
• Ethernet cable (optional but recommended for reliability)

Pi Setup Steps

# Update your Pi
sudo apt update && sudo apt upgrade -y

# Install Node.js 20
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt install -y nodejs

# Clone and install OpenClaw
git clone https://github.com/openclaw/openclaw.git
cd openclaw
npm install
npm run setup

For headless operation, set OpenClaw to start automatically:

# Create a systemd service
sudo tee /etc/systemd/system/openclaw.service << EOF
[Unit]
Description=OpenClaw AI Agent
After=network.target

[Service]
Type=simple
User=pi
WorkingDirectory=/home/pi/openclaw
ExecStart=/usr/bin/npm start
Restart=always

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl enable openclaw
sudo systemctl start openclaw

---

Connecting Your First Messaging Platform

Telegram (Easiest to Start)

1. Open Telegram and search for @BotFather
2. Send /newbot and follow the prompts to create a bot
3. Copy the API token BotFather gives you
4. In OpenClaw's settings, add a Telegram gateway with your token
5. Send a message to your new bot — OpenClaw responds

WhatsApp

WhatsApp requires a few more steps due to Meta's API restrictions. OpenClaw uses the WhatsApp Web bridge approach:

1. Enable the WhatsApp gateway in OpenClaw settings
2. Scan the QR code with your WhatsApp app (just like WhatsApp Web)
3. Your agent can now receive and respond to WhatsApp messages

Discord

1. Create a Discord bot at discord.com/developers
2. Copy the bot token
3. Add the bot to your server with appropriate permissions
4. Configure the Discord gateway in OpenClaw

For the full list of 20+ supported platforms, check the OpenClaw docs.

---

Critical Security Steps (Don't Skip These)

After the viral WebSocket vulnerability (CVE-2026-25253), security is non-negotiable. Read our full OpenClaw Security Guide, but at minimum:

1. Update immediately: git pull && npm install — make sure you're on version 2026.1.29+
2. Set API spending limits at your LLM provider (e.g., $50/month cap on Anthropic)
3. Enable action approval for all irreversible actions (sending messages, file deletion, payments)
4. Don't expose port 3100 to the public internet without proper auth
5. Audit every skill before installing — read the source code

---

Recommended Skills to Install First

Once your agent is running, add these essential skills:

Skill	Purpose	Trust Level
Web Search	Search the internet and summarize results	Official ✅
Calendar	Manage Google/Outlook calendar	Official ✅
File Manager	Read and organize local files	Official ✅
Weather	Get weather forecasts	Community ⚠️
News Summary	Daily news digest	Community ⚠️

Install skills via the CLI:

openclaw skill install web-search
openclaw skill install calendar

For a deep dive on the best skills, see our 10 Best OpenClaw Skills article.

---

Troubleshooting Common Issues

"Connection refused" when messaging the bot

Your OpenClaw instance isn't running or the gateway isn't configured correctly. Check:

openclaw status
openclaw logs --tail 50

High API costs

OpenClaw can burn through API credits fast if not configured properly. Set the MAX_TOKENS_PER_MESSAGE and DAILY_SPENDING_LIMIT in your config:

llm:
  max_tokens_per_message: 4096
  daily_spending_limit: 5.00

Agent responding slowly

If using a Raspberry Pi, slow responses are usually due to network latency to the LLM API, not local processing. Ensure you're using a wired Ethernet connection for the best experience.

---

Hardware Recommendations

Raspberry Pi 5 8GB

The sweet spot for a dedicated OpenClaw server. Low power consumption, silent operation, and enough RAM to handle the agent framework plus several concurrent skills.

Samsung T7 Portable SSD 1TB

A significant performance upgrade over microSD for your Pi. OpenClaw's persistent memory reads and writes constantly — an SSD makes everything snappier.

Vilros Raspberry Pi 5 Starter Kit

If you don't have a Pi yet, this kit includes everything: the Pi 5 8GB, aluminum case with turbo cooler, 32GB microSD, power supply, and HDMI cables.

LG 27UN850-W 4K Monitor

If you're monitoring your agent's dashboard and doing development work, a 4K monitor makes reading logs and config files much more comfortable.

Prompt Engineering for LLMs

OpenClaw's usefulness is directly proportional to how well you write its system prompts. This book covers the techniques that make AI agents follow instructions reliably.

---

What's Next?

Now that you have OpenClaw running, explore these WikiWayne guides:

• All About OpenClaw — The complete overview
• OpenClaw Security Guide — Essential security hardening
• 10 Best OpenClaw Skills — Must-have plugins
• OpenClaw + Claude Integration — Connecting the best LLM
• OpenClaw for Content Creators — Automation workflows for creators

---

Got your OpenClaw running? Share your setup and messaging platform of choice on X (@wikiwayne). I'm always curious to see what hardware people are running it on.

Recommended Gear

These are products I personally recommend. Click to view on Amazon.

Raspberry Pi 5 8GB — Great pick for anyone following this guide.

Samsung T7 Portable SSD 1TB — Great pick for anyone following this guide.

Vilros Raspberry Pi 5 Starter Kit — Great pick for anyone following this guide.

Prompt Engineering for LLMs — Great pick for anyone following this guide.

Logitech MX Keys S Wireless — Great pick for anyone following this guide.

LG 27UN850-W 4K UHD — Great pick for anyone following this guide.

---

This article contains affiliate links. As an Amazon Associate I earn from qualifying purchases. See our full disclosure.