Install OpenClaw on Mac

Want to run your own AI agent on your Mac? OpenClaw is an open-source autonomous AI agent that functions as a personal assistant — and it integrates with messaging apps like WhatsApp, Telegram, and Discord.

In this guide, I'll walk you through installing OpenClaw on your Mac and configuring it with Ollama to run local AI models for free.

What is OpenClaw?

OpenClaw is an open-source autonomous AI agent that can:

• Run system commands on your computer
• Integrate with messaging apps (WhatsApp, Telegram, Discord)
• Execute scheduled tasks via cron jobs
• Read and write files on your filesystem
• Use local AI models via Ollama (no API costs)

Think of it as your own AI employee that lives on your machine and can do real work.

Prerequisites

Before we start, make sure you have:

• A Mac (macOS)
• Terminal access
• Internet connection
• Git installed (the installer will check for this)

Step 1: Install OpenClaw Using the One-Liner

Open your Terminal app and run this command:

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

The installer will:

1. Detect your operating system
2. Check for NodeJS (install if needed)
3. Verify Git is installed
4. Install the OpenClaw npm package
5. Start the onboarding process

Step 2: Complete the Onboarding Process

After installation, OpenClaw will start the onboarding process automatically.

1Read the Security Warning

Use your arrow keys to select Yes and press Enter to continue.

2Select Quick Start Mode

Choose Quick Start for the fastest setup. This handles most configuration automatically.

3Skip Model Selection (For Now)

Since we want to use Ollama for local models, select Skip for now when asked about models. We'll configure Ollama in the next steps.

If you don't see Ollama in the list, just enter any placeholder name (like /demo) to complete onboarding — we'll replace it later.

4Skip Channels (Optional)

You can skip channel configuration for now. We'll use the native web chat first. You can always add WhatsApp, Telegram, or Discord later.

5Skip Skills and Hooks

Select Skip for now for skills and hooks. These can be configured anytime later.

Important: Your Gateway Token

Note the token shown at the end of onboarding. You'll need it to access the web dashboard:

http://127.0.0.1:18789/token/YOUR_TOKEN_HERE

The token is also stored in your config file:

cat ~/.openclaw/openclaw.json

Step 3: Install Ollama

Now let's install Ollama so we can run local AI models for free.

First, stop the OpenClaw gateway:

openclaw gateway stop

Then install Ollama. You can either:

• Download from ollama.com — Get the Mac installer app
• Or use the command line:

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

Wait for the installation to complete. You might see a message about GPU detection — Ollama works on both Apple Silicon and Intel Macs.

Verify Ollama is running by visiting http://localhost:11434 in your browser. You should see "Ollama is running".

Step 4: Configure OpenClaw with Ollama

Now let's connect OpenClaw to Ollama and download a model.

Run the Ollama launch command for OpenClaw:

ollama launch openclaw

You'll see a list of recommended models. I suggest MiniMax M2.5 cloud — it's a solid choice for general use.

Select your model and press Enter. When asked to download, select Yes.

Once connected, the model will download and OpenClaw will start with it configured.

Step 5: Test Your Setup

Open your browser and go to the OpenClaw dashboard:

http://127.0.0.1:18789

Go to the Agents section to verify your model is configured (e.g., ollama/minimax-m2.5:cloud).

Now try the chat! Ask OpenClaw something that requires system access:

What are the specs of this Mac?

OpenClaw will run system commands to gather information about your CPU, RAM, storage, and macOS version — then report back with the results.

Another Test: Find Files

Try something more advanced:

Find the folder in my Documents directory where I have saved a Python script and tell me what it does.

OpenClaw will use the find command to locate your file, then read its contents and explain what the program does. It's not just giving answers from its training — it's actually running commands on your system.

Exploring the Dashboard

The OpenClaw web dashboard gives you full control over your AI agent:

• Chat — Talk to your agent directly
• Agents — View and configure AI models
• Sessions — See past conversations
• Usage — Track token usage and costs (helpful for paid models)
• Scheduler — Set up automated tasks (cron jobs)
• Skills — Add AGENTS.md files for custom behaviors
• Channels — Connect WhatsApp, Telegram, Discord, etc.
• Config — View environment and settings
• Logs — Debug issues and see what's happening

Gateway Commands Reference

Here are the essential commands for managing your OpenClaw gateway:

# Check gateway status
openclaw gateway status

# Start the gateway
openclaw gateway start

# Stop the gateway
openclaw gateway stop

# Restart the gateway
openclaw gateway restart

Troubleshooting

Token Not Working?

If you get an error about your token, go to the Overview section in the dashboard, enter your gateway token, and reload the page.

No Model Response?

If you see no answer in the chat, check the Agents section to confirm your model is set correctly. If you skipped model setup with a dummy name, you'll need to reconfigure it with Ollama.

Ollama Not Detected?

Make sure Ollama is running. Check with:

curl http://localhost:11434

You should see "Ollama is running".

Next Steps

Now that OpenClaw is running, you can:

• Connect WhatsApp — Chat with your agent from your phone
• Add skills — Create AGENTS.md files for custom behaviors
• Set up cron jobs — Automate recurring tasks
• Try different models — Experiment with other Ollama models

Key Takeaways

• One command installs everything — NodeJS, dependencies, and OpenClaw
• Ollama provides free local models — No API costs
• OpenClaw runs real commands — It's not just answering from training data
• Your config is at ~/.openclaw/openclaw.json — Contains settings and your gateway token
• The dashboard runs at 127.0.0.1:18789 — Full control from your browser