Install OpenClaw in 5 Minutes and Turn It Into Your Personal AI Assistant

What is OpenClaw?

OpenClaw is an open‑source AI‑agent gateway that lets a single AI assistant operate across multiple chat platforms (Telegram, Discord, WhatsApp, Feishu, DingTalk, etc.) and perform local actions such as file management, alarm setting, and command execution. It runs locally, keeping data on the host machine.

System requirements

OS: Windows 10+, macOS 12+, Ubuntu 20.04+ (recommended: Windows 11, macOS 14+, Ubuntu 22.04+)

Node.js v22+ (v24 LTS recommended)

Memory: 2 GB minimum (4 GB+ recommended)

Disk: 500 MB minimum (1 GB+ recommended)

Network: Internet access for AI model APIs and GitHub

Preparation

Obtain an API key from a supported model provider (e.g., DeepSeek, Anthropic Claude, OpenAI GPT, Google Gemini, Moonshot, Ollama). DeepSeek offers a free tier.

Ensure a stable network that can reach GitHub.

Allocate about 15 minutes for the installation process.

Installation methods

One‑click script (recommended)

Run the official install script; it detects Node.js, installs dependencies, and configures OpenClaw automatically.

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

Windows PowerShell:

iwr -useb https://openclaw.ai/install.ps1 | iex

⚠️ Native Windows installation may have compatibility issues; WSL2 is preferred.

npm installation

# Global install
npm install -g openclaw@latest

# Verify
openclaw --version

For faster downloads in China:

npm install -g openclaw@latest --registry=https://registry.npmmirror.com

Build from source

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

# Install dependencies
pnpm install

# Build
pnpm build

# Link the CLI globally
pnpm link --global

💡 If pnpm is missing, first run npm install -g pnpm .

Verify installation

# Show version
openclaw --version

# Run diagnostics
openclaw doctor

All checks should report ✅.

Configuration wizard (onboarding)

Start the wizard: openclaw onboard --install-daemon The interactive prompts guide you through:

Select a setup mode – Quickstart for beginners or Advanced for manual control.

Choose a model provider (Anthropic, OpenAI, Google, Moonshot, DeepSeek, Ollama).

Enter the API key.

Optionally connect a chat channel (e.g., Telegram, WeChat Work).

Optionally install skills (plugins).

After the wizard finishes, start the gateway:

# Run the gateway
openclaw gateway run

# Or open the web dashboard
openclaw dashboard

Basic usage

Terminal UI (TUI)

openclaw tui

Web UI

Open a browser and go to http://127.0.0.1:18789/.

Telegram bot

If linked during onboarding, send messages to the bot.

Common commands

openclaw status

– Show runtime status openclaw tui – Open terminal chat openclaw dashboard – Open web console openclaw logs --follow – Tail live logs openclaw update – Check for updates openclaw models status – Show model connection status

Advanced configuration

Multiple model fallbacks

{
  "agent": {
    "model": "anthropic/claude-sonnet-4-5",
    "fallbackModels": [
      "openai/gpt-4o",
      "deepseek/deepseek-chat"
    ]
  }
}

If the primary model is unavailable, OpenClaw automatically switches to the next one.

Domestic model integration

DeepSeek (recommended for low cost):

# Export the key
export DEEPSEEK_API_KEY="your_key"

# In the config file
{
  "agent": {
    "model": "deepseek/deepseek-chat-v3"
  }
}

GLM (Zhipu):

export ZHIPU_API_KEY="your_key"

Cron jobs

Schedule recurring AI tasks, e.g., a daily briefing at 8 AM:

{
  "cron": {
    "jobs": [
      {
        "name": "daily-briefing",
        "schedule": "0 8 * * *",
        "message": "Generate today’s schedule summary"
      }
    ]
  }
}

Skills marketplace (ClawHub)

Install extensions for calendar sync, email management, task tracking, smart‑home control, etc., from https://clawhub.io.

Common issues and fixes

Windows PATH problem

# Add npm global bin to PATH temporarily
export PATH="$(npm prefix -g)/bin:$PATH"

# Persist the change
echo 'export PATH="$(npm prefix -g)/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

Using WSL2 is recommended.

API key errors

Ensure the key is copied without extra spaces.

Check that the key has not expired or exhausted its quota.

Verify the key belongs to the selected provider.

Model timeouts

Check network connectivity.

Try a different model.

Increase the timeout setting in the configuration.

Port conflict (default 18789)

# Find the process using the port (Windows PowerShell)
netstat -ano | findstr :18789

# macOS/Linux
lsof -i :18789

# Change the gateway port in the config, e.g.:
"gateway": { "port": 3000 }