How to Use OpenClaw on Mac: Complete Setup & Voice Wake Guide

OpenClaw is a personal AI assistant that's particularly powerful on macOS thanks to a dedicated menu bar companion app. Unlike other AI tools, OpenClaw gives you a native macOS experience with voice wake activation, screen recording, Canvas integration, and tight system integration. This comprehensive guide walks you through installation, setup, and advanced features specific to Mac users.

Why OpenClaw on Mac is Special

macOS is where OpenClaw truly shines. The native companion app sits in your menu bar, always listening for your wake word, and can:

• Listen for voice wake words (like "Hey Molty") and respond with push-to-talk
• Record your screen and solve visual problems
• Control applications via Canvas (visual interface for browser control and UI automation)
• Send notifications directly from your AI assistant
• Execute system commands with your approval (with granular security controls)
• Integrate with your calendar, email, reminders, and more
• Switch between local and remote gateways seamlessly
• Manage notifications and system access through a native interface

If you're on Intel or Apple Silicon Mac (M1/M2/M3/M4), OpenClaw works beautifully. Let's get it running.

Prerequisites

Hardware Requirements:

• Mac with Intel or Apple Silicon (M1/M2/M3/M4 or later)
• macOS 12 (Monterey) or later (newer versions recommended)
• 4GB RAM minimum (8GB+ recommended for comfortable operation)
• 2GB free disk space

Software Requirements:

• Node.js version 22 or higher
• Xcode Command Line Tools (for building the macOS app; optional if CLI-only)
• Anthropic API key (recommended for best performance)
• Optional: Brave Search API key for web search

What You Don't Need:

• WSL2 (this is Windows only—skip that entirely on Mac)
• Docker or virtualization tools
• Enterprise installation tools

Installation Steps

Step 1: Install Node.js 22+

First, verify you have Node 22 or higher installed:

node --version

If you need to install or upgrade Node.js, use Homebrew:

brew install node@22

Or use Node Version Manager (nvm) for more flexibility:

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.0/install.sh | bash
nvm install 22
nvm use 22

Verify the installation:

node --version  # Should show v22.x.x or higher
npm --version

Step 2: Install Xcode Command Line Tools (Optional but Recommended)

If you plan to build the macOS companion app or develop features, install Xcode CLT:

xcode-select --install

For CLI + gateway only (no app building), you can skip this step—Node is sufficient.

Step 3: Install the OpenClaw CLI

Run the official installer:

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

This downloads and installs the OpenClaw command-line interface globally. Once complete, verify installation:

openclaw --version
openclaw --help

Step 4: Run the Onboarding Wizard

The onboarding wizard is the recommended setup path. It guides you through authentication, channel setup, and daemon installation interactively:

openclaw onboard --install-daemon

During onboarding, you'll configure:

A) Gateway Mode

• Choose "Local" (default) to run the gateway on your Mac
• Choose "Remote" if you want to connect to a gateway on another machine

B) AI Provider & Authentication

• Recommended: Anthropic API key (from console.anthropic.com)
• Alternative: OpenAI API key
• Tip: Claude Opus 4.5 provides the best performance for complex tasks

C) Chat Channels

Select which messaging platforms to connect:

• WhatsApp (scan QR code)
• Telegram (bot token)
• Discord (bot token)
• Slack, Signal, iMessage, Google Chat, Microsoft Teams, etc.

Start with one channel (WhatsApp or Telegram recommended) and add more later.

D) Install as Daemon

Enable "yes" to install OpenClaw as a launchd service so it starts automatically when you log in.

E) Gateway Token

The wizard generates a secure token automatically. This authenticates access to your OpenClaw instance.

Step 5: Verify Installation

Check that everything is working:

openclaw status
openclaw health

Ensure the daemon is running:

openclaw gateway status

Step 6: Download the macOS Menu Bar App

The native macOS companion app provides voice wake, notifications, and system integration. Download it from the official site or build from source.

Via Homebrew (easiest):

brew tap openclaw/apps
brew install openclaw-macos

Or download from openclaw.ai

Once installed, launch the app from Applications or Spotlight:

open /Applications/OpenClaw.app

Step 7: Configure the macOS App

When you first launch OpenClaw.app, it will:

1. Ask for permissions (Microphone, Screen Recording, Accessibility, Notifications)
2. Automatically detect your running local gateway
3. Connect to your gateway

Grant all permissions for full functionality:

• Microphone: Required for voice wake and voice input
• Screen Recording: For Canvas (visual automation) and taking screenshots
• Accessibility: For system.run and UI automation
• Notifications: To receive messages and alerts from your assistant
• AppleScript/Automation: For advanced macOS automation

Step 8: Test Your Setup

Send a test message from your connected channel:

openclaw message send --target "+1234567890" --message "Hello from OpenClaw"

(Replace with your actual phone number for WhatsApp, or Telegram/Discord ID)

Check the menu bar icon—it should show a cute lobster that animates when working.

macOS-Specific Features You Should Know About

Voice Wake (Always-On Listening)

OpenClaw's voice wake lets you say a wake word (default: "Hey Molty") to trigger your assistant without opening an app.

To Enable Voice Wake:

1. Open OpenClaw.app → Settings
2. Go to Voice Wake tab
3. Enable "Listen for wake word"
4. Customize your wake word (default is "Hey Molty")
5. Set microphone sensitivity if needed

How It Works:

• The wake word detector runs locally on your Mac (no cloud recording of wake words)
• When triggered, your audio is transcribed and sent to the AI model
• The response is delivered back to your last-used chat channel (WhatsApp/Telegram/Discord/WebChat)
• Optional chimes play when wake word is detected and when response is sent

Voice Wake Settings:

• Wake word: Customize to whatever you like
• Microphone input: Choose which microphone to use
• Sounds: Enable/disable chimes for trigger and send events
• Auto-send delay: Customize how long to wait before auto-sending (adjustable)
• Fallback provider: Which channel to deliver responses if primary is unavailable

Push-to-Talk (PTT)

Don't want to wait for wake-word detection? Use push-to-talk:

1. Press and hold Cmd+Fn on your keyboard
2. Speak your message
3. Release to send

The overlay will show your speech in real-time and send automatically when you release.

Menu Bar Icon States

The lobster icon in your menu bar tells you what's happening:

• Idle: Cute lobster blinking and occasionally wiggling
• Listening: Ears enlarge (1.9x) when wake word is detected
• Working: Fast leg scurry motion while the agent is processing
• Paused: Icon appears dimmed and still
• Error: Icon shows a different state if something's wrong

Click the icon for quick access to:

• Status
• Open dashboard
• Toggle voice wake on/off
• Access settings
• View logs

Canvas & Visual Automation

Canvas is OpenClaw's visual interface for browser control and UI automation. The agent can:

• Take screenshots of your screen or browser
• Click elements based on visual context
• Fill forms and interact with web apps
• Solve visual problems by seeing what's on screen

To enable Canvas:

1. Grant Screen Recording permission
2. Grant Accessibility permission
3. The agent can now use canvas.* tools

macOS Integrations

The macOS app exposes native tools to your agent:

System Commands

system.run: Execute terminal commands with approval
system.notify: Send macOS notifications

File System Access

• Read/write files in allowed directories
• Browse folder structures
• List processes running on your Mac

UI Automation

• Control applications via AppleScript
• Trigger Shortcuts workflows
• Interact with the system

Exec Approvals (Security)

For maximum security, configure exec approvals. This lets you control which commands the agent can run without asking:

Settings → Exec Approvals:

• Default: Deny (no commands without approval)
• Ask on miss: Ask user when unknown command is attempted
• Allowlist: Pre-approve specific commands (e.g., /usr/local/bin/curl)

Smart: You can set different approval rules for different agents.

Remote Mode

Want to run OpenClaw on a different Mac or server but control it from your laptop? Use remote mode:

On the Server/Always-On Mac:

1. Set up OpenClaw gateway normally
2. Note the hostname/IP and port

On Your Client Mac:

1. Open OpenClaw.app → Settings
2. Go to "General" → "OpenClaw runs" → Select "Remote"
3. Enter server hostname/IP and port
4. The app automatically opens an SSH tunnel

Voice Wake and WebChat work seamlessly over the remote connection.

Advanced Setup: Building from Source

If you want to contribute to OpenClaw or build the latest development version:

git clone https://github.com/openclaw/openclaw.git
cd openclaw
pnpm install
pnpm ui:build  # Auto-installs UI dependencies
pnpm build
openclaw onboard --install-daemon

Then build the macOS app:

cd apps/macos
xcode-build -scheme OpenClaw -configuration Release

(Requires Xcode and Xcode CLT)

Troubleshooting Common macOS Issues

"OpenClaw.app won't start" or crashes on launch:

• Try restarting your Mac
• Check that Node 22+ is installed: node --version
• Check logs: openclaw logs gateway
• Reinstall: brew reinstall openclaw-macos

Voice Wake not working:

• Grant Microphone permission: System Settings → Privacy & Security → Microphone
• Check microphone works: System Settings → Sound
• Restart the app
• Try adjusting microphone sensitivity in Settings

Screen Recording not working (Canvas/Screenshots):

• Grant Screen Recording permission: System Settings → Privacy & Security → Screen Recording
• Restart OpenClaw.app
• Verify with: openclaw doctor

"Permission denied" for Accessibility:

• Grant Accessibility access: System Settings → Privacy & Security → Accessibility
• Check that OpenClaw.app is listed and enabled
• Restart the app

Gateway won't start or keeps crashing:

• Check port availability: lsof -i :18789
• Try a different port: openclaw configure --section gateway.port
• Check system resources: top (is your Mac low on memory?)
• Restart launchd: openclaw gateway restart

App freezes during heavy tasks:

• This is usually temporary while the agent is thinking
• Check Activity Monitor to see CPU/memory usage
• Consider allocating more resources or using a remote gateway

SSH tunnel not connecting (Remote mode):

• Verify SSH access to remote host: ssh user@hostname
• Check firewall rules on remote host
• Verify gateway is running on remote: openclaw gateway status --url <remote_url>
• Try restarting the tunnel: Close and reopen the app

Security Best Practices for macOS

1. Run security audits regularly:

   openclaw security audit --deep

2. Set granular exec approvals: Don't grant blanket command execution permission. Use allowlists.
3. Keep permissions minimal: Only grant what the agent actually needs.
4. Monitor system activity: The agent has broad system access. Review logs periodically.
5. Use pairing approvals for DMs: By default, unknown contacts need approval before the agent responds.
6. Isolate agents by workspace: Create separate agents for different purposes (work vs personal).
7. Enable firewall: Use macOS firewall to restrict inbound connections.
8. Use SSH tunnels for remote: Avoid exposing your gateway port directly; use SSH or Tailscale.

Next Steps

After getting OpenClaw running:

1. Configure web search: Get a Brave Search API key for enhanced search capabilities

   openclaw configure --section web

2. Add more channels: Integrate Slack, Discord, or Signal for team collaboration
3. Install skills from ClawHub: Extend capabilities with custom skills

   clawhub sync --all

4. Set up automations: Use cron jobs and webhooks to trigger tasks
5. Connect iOS/Android nodes: Control your iPhone or iPad remotely
6. Explore Canvas: Let the agent take screenshots and solve visual problems
7. Enable voice calls: Use the voice-call plugin for phone-based interactions

FAQ: Your macOS Questions Answered

cp -r ~/.openclaw ~/.openclaw.backup

openclaw update
# or: npm install -g openclaw@latest

brew upgrade openclaw-macos

Final Thoughts

OpenClaw on macOS is one of the most polished personal AI assistant experiences available. The native menu bar integration, voice wake, and Canvas support make it feel like a natural part of your Mac.

Start simple: get the CLI running, set up one chat channel, and explore the dashboard. Once you're comfortable, add voice wake, grant system permissions gradually, and build your automation from there.

The macOS community is actively developing OpenClaw, and new features are rolling out regularly. Happy automating! 🦞

Version: Based on OpenClaw documentation as of February 2026
macOS Version: Compatible with macOS 12 (Monterey) and later; optimized for macOS 13+
Hardware: Intel and Apple Silicon (M1/M2/M3/M4+)
Need Help? See docs.openclaw.ai or the GitHub repository