Most AI assistants today are little more than chat interfaces. They answer questions but cannot access your tools, remember your preferences, or perform real work. OpenClaw takes a different approach.

Instead of functioning purely as a chatbot, OpenClaw acts as a programmable cognitive partner. It can:

  • remember long‑term context
  • connect to external tools
  • automate workflows
  • run entirely on infrastructure you control

This guide explains how to deploy OpenClaw from scratch and configure it on Windows, macOS, or Linux.

By the end of this tutorial you will have a working OpenClaw instance running locally.

Why Use OpenClaw

Private Deployment

All conversations and memory stay on your own machine or server.

Tool Integration

OpenClaw can interact with tools such as:

  • GitHub
  • calendars
  • APIs
  • automation workflows

Persistent Memory

OpenClaw includes both short‑term and long‑term memory, allowing it to remember:

  • user preferences
  • project details
  • conversation context

Open Source

OpenClaw is released under the MIT license, meaning the entire system is fully auditable and customizable.

System Requirements

Minimum environment:

plaintext
Node.js 18+
npm or pnpm
8 GB RAM recommended
2 GB disk space

Supported platforms:

  • macOS
  • Linux
  • Windows (PowerShell or WSL2)

Verify Your Environment

macOS / Linux

bash
node --version
npm --version

If Node is missing, install it from:

https://nodejs.org

Installation Method 1: One‑Step Installer

macOS / Linux

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

After installation open:

http://localhost:18789

Windows (PowerShell)

Clone the repository and start the gateway manually:

bash
git clone https://github.com/openclawai/openclaw.git
cd openclaw
npm install
npm run start:gateway

Install WSL:

wsl --install

Then inside WSL:

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

Access the interface via:

http://localhost:18789

Installation Method 2: Manual Installation

Clone Repository

bash
git clone https://github.com/openclawai/openclaw.git
cd openclaw

Install Dependencies

bash
npm install -g pnpm
pnpm install

or

npm install

Configure Workspace

macOS / Linux

plaintext
export OPENCLAW_WORKSPACE=~/.openclaw/workspace
mkdir -p $OPENCLAW_WORKSPACE

Windows PowerShell

plaintext
setx OPENCLAW_WORKSPACE "%USERPROFILE%\.openclaw\workspace"
mkdir %USERPROFILE%\.openclaw\workspace

Copy Configuration

cp config.example.yaml config.yaml

Windows

copy config.example.yaml config.yaml

Start Gateway

pnpm start:gateway

Docker Deployment

Create docker-compose.yml

yaml
version: "3.8"

    services:
      openclaw:
        image: ghcr.io/openclaw/openclaw:latest
        container_name: openclaw
        ports:
          - "18789:18789"
          - "18792:18792"
        environment:
          - OPENCLAW_WORKSPACE=/workspace
          - NODE_ENV=production
        volumes:
          - ./workspace:/workspace
          - ./config.yaml:/app/config.yaml
          - ./logs:/var/log/openclaw
        restart: unless-stopped

Start the container:

docker-compose up -d

View logs:

docker-compose logs -f

Configuring AI Models

Example configuration in config.yaml:

yaml
agents:
      defaults:
        model: "siliconflow/deepseek-ai/DeepSeek-V3.2"

Alternative models:

plaintext
openai/gpt-4
anthropic/claude-3-opus
google/gemini-2.0

Enabling Skills

Skills allow OpenClaw to interact with external services.

Example configuration:

plaintext
skills:
      enabled:
        - github
        - weather

Install additional skills:

openclaw skills install github

Creating Your First Custom Skill

Create directory:

mkdir ~/.openclaw/workspace/skills/my-skill

Create SKILL.md:

md
# Custom Skill

## Description
Automation skill for internal APIs.

## Usage
/my-skill [arguments]

## Capabilities
- API access
- file reading
- workflow automation

Troubleshooting

Port Already In Use

macOS / Linux

lsof -i :18789

Windows

netstat -ano | findstr 18789

Kill process:

Linux/macOS

pkill -f openclaw

Windows

taskkill /PID <pid> /F

Conclusion

OpenClaw is more than a chatbot. It is a programmable AI platform capable of integrating with real workflows, remembering context, and evolving through custom skills.

Once deployed, it can become a powerful personal AI infrastructure layer that helps automate tasks, analyze code, and manage complex workflows.

For developers interested in building their own AI ecosystem, OpenClaw provides a strong foundation for creating a true digital second brain.