Getting started with OpenClaw doesn't require deep technical knowledge, but understanding the setup process makes all the difference. Whether you're running AI automation, building custom agents, or deploying automation workflows, a clean installation sets you up for success.

This guide walks you through installing OpenClaw on your system, configuring it properly, and running your first automation in under an hour.

What Is OpenClaw?

OpenClaw provides a unified environment for AI automation and agent building

OpenClaw is a personal AI assistant platform that runs on your machine or server. Unlike cloud-only AI tools, OpenClaw gives you:

• Local execution with zero cloud dependency
• Full control over your automation workflows
• Custom agent building without coding expertise
• Integration with Claude API for advanced reasoning
• Browser automation, file processing, and task scheduling

When to use OpenClaw:

• You need repeatable automation that runs on schedule
• You want AI agents handling multi-step tasks
• Your workflows involve sensitive data (stays local)
• You're building custom tools for your team

System Requirements

Ensure your system meets OpenClaw requirements before installing

Minimum specs:

• OS: Ubuntu 20.04+ (Linux), macOS 11+, Windows 10 Pro
• RAM: 4GB (8GB recommended)
• Disk space: 2GB free
• Processor: Intel i5/AMD Ryzen 5 or equivalent
• Network: Stable internet connection (for Claude API calls)

Optional but recommended:

• Node.js 18+ (for running custom scripts)
• Docker (for containerized deployments)
• Git (for version control of workflows)

Step 1: Download and Install OpenClaw

Visit the OpenClaw releases page to download the latest version

Visit the official OpenClaw releases page and download the installer for your operating system.

On Linux/macOS: Open your terminal and run:

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

On Windows: Download the installer executable and run it. Accept the default installation directory.

Step 2: Verify Installation

Verify OpenClaw installed correctly by checking the version

After installation completes, verify OpenClaw is working:

openclaw --version

You should see output like: OpenClaw 2026.2.3-1

If you see "command not found," restart your terminal or add OpenClaw to your PATH.

Step 3: Configure Claude API Access

Configure your Claude API credentials for AI-powered automation

OpenClaw needs access to Claude's API for intelligence. Create a free account at Anthropic's console:

1. Visit console.anthropic.com
2. Create an account or log in
3. Navigate to API keys section
4. Create a new API key
5. Copy the key (save it securely)

In your terminal, set the API key:

export ANTHROPIC_API_KEY="sk-ant-your-key-here"

To persist it across sessions, add this line to your ~/.bashrc or ~/.zshrc file.

Step 4: Initialize Your First Workspace

Initialize your workspace with OpenClaw to create the directory structure

Create a new workspace for your automation projects:

openclaw init my-workspace
cd my-workspace

This creates a folder structure with:

• agents/ — your AI agents
• workflows/ — automation sequences
• skills/ — custom functions and tools
• .openclaw/ — configuration files

Step 5: Test Your Setup with a Simple Task

Run your first OpenClaw task to verify everything is configured correctly

Create a simple test file test-task.js in your workspace:

module.exports = {
  name: "hello-world",
  description: "Test task to verify OpenClaw setup",
  run: async () => {
    console.log("✓ OpenClaw is working!");
    console.log("✓ Ready to build automation workflows");
  }
};

Run it:

openclaw run hello-world

You should see the success message. Congratulations—your installation is complete!

Troubleshooting Common Issues

Use OpenClaw's diagnostic tools to identify and fix setup issues

"Command not found: openclaw"

• Linux/macOS: Restart your terminal or run source ~/.bashrc
• Windows: Restart your command prompt after installation

"API key not found"

• Verify you've set ANTHROPIC_API_KEY correctly
• Check with echo $ANTHROPIC_API_KEY (should print your key)

"Port already in use"

• By default, OpenClaw uses port 3000
• Change it: openclaw gateway start --port 3001

Installation fails on macOS

• You may need Xcode Command Line Tools: xcode-select --install
• Or use Homebrew: brew install openclaw

Next Steps: What to Build

OpenClaw enables complex multi-step workflows with AI agents

Now that OpenClaw is installed and verified, you're ready to:

1. Build your first automation — Process files, summarize reports, send alerts
2. Create a custom skill — Add tools that your agents can use
3. Deploy a scheduled workflow — Run tasks on a schedule (every hour, daily, etc.)
4. Connect integrations — Link to Slack, email, databases, or APIs

Check out the tutorials on claw.ist for step-by-step guides on building OpenClaw automations and creating custom AI agents.

Conclusion

Your OpenClaw installation is complete and ready for building intelligent automation

You now have a fully functional OpenClaw installation ready for building AI-powered automation. The setup process is straightforward, and the platform gives you incredible power once configured.

Start small—build a simple automation, test it, then expand to more complex workflows. Each skill and agent you create becomes a reusable building block for future projects.

Your AI assistant is ready to work. What will you automate first?

FAQ

Q: Is OpenClaw free? A: OpenClaw itself is free and open-source. You pay only for Claude API usage, which starts at $0.03 per 1M input tokens.

Q: Can I run OpenClaw on a cloud server? A: Yes. SSH into your server and follow the Linux installation steps. Many users run it on AWS, DigitalOcean, or Linode.

Q: Does OpenClaw require an internet connection? A: Your workflows run locally, but you need internet to call Claude's API for intelligence.

Q: What if I need help after installation? A: The OpenClaw community is active on Discord. Visit the official Discord for support.

Q: Can I use OpenClaw without knowing programming? A: Absolutely. Many automations are pure workflow configuration. When you do need code, start simple—add complexity as needed.