The Fastest Setup Path

Do not overcomplicate this on day one. Follow this order and you will be running in under an hour.

1. Install Node.js
2. Install OpenClaw
3. Add your API key
4. Start the gateway
5. Open the interface
6. Customize your first agent

Step 1 — Install Node.js

First check if Node.js is already installed on your machine:

node -v
npm -v

If that returns a version number you are good to go. If it fails, install the LTS version from nodejs.org — LTS means Long Term Support, which is the stable version recommended for most users.

Step 2 — Install OpenClaw

npm install -g openclaw
openclaw --help

The -g flag installs OpenClaw globally so you can run it from anywhere in your terminal. If the --help command returns a list of options, the install worked.

Step 3 — Add Your API Key

OpenClaw needs a model to power your agent. If you are using an OpenAI-style model, set your key like this:

export OPENAI_API_KEY="your_key_here"

To make this permanent so you do not have to re-enter it every session, add it to your shell profile:

~/.bashrc   # for Bash users
~/.zshrc    # for Zsh users (default on Mac)

Then reload your shell:

source ~/.bashrc

If you want to run Claude inside OpenClaw for free using your GitHub Copilot subscription, we documented the exact setup here:

Step 4 — Start the Gateway

OpenClaw runs a local gateway service that handles all communication between your agent and the model. You need to start this before anything works.

openclaw gateway start
openclaw gateway status
openclaw status

If anything looks stuck or unresponsive, restart it:

openclaw gateway restart

Step 5 — Open the Interface

Once OpenClaw is running, it will either print a local URL in your terminal or open the interface automatically. If you see a URL like http://localhost:PORT, paste that into your browser.

If you are ever unsure what state the system is in, this command gives you a clear status readout:

openclaw status

Step 6 — Make Your First Real Agent

This is where it starts feeling alive. A raw OpenClaw install gives you an agent, but it behaves generically until you give it character and context. At minimum, create these four files inside your agent workspace:

• SOUL.md — personality, tone, and identity
• USER.md — who the agent serves and what they care about
• MEMORY.md — long-term memory the agent carries forward
• AGENTS.md — house rules and behavioral guidelines

Simple Beginner Use Cases

Once installed, you can put your agent to work immediately on any of these:

• Content writing and blog drafting
• Research and topic summarization
• Coding help and debugging
• Telegram assistant bot
• Reminders and scheduled heartbeat check-ins
• Business planning and strategy
• Daily alpha reports via cron jobs

If You Want Telegram Too

Telegram integration takes a little extra setup but is one of the most powerful things you can do with OpenClaw — your agent becomes reachable anywhere from your phone. Here is the rough path:

1. Create a bot with BotFather on Telegram
2. Get your bot token from BotFather
3. Add the bot to your group or chat
4. Put the token in your OpenClaw config
5. Restart the gateway
6. Test by messaging the bot

Beginner Mistakes to Avoid

• Forgetting to install Node.js before trying to install OpenClaw
• Forgetting to set your API key — the agent will not respond without it
• Installing OpenClaw but never starting the gateway
• Expecting the agent to feel smart before you configure memory and personality files
• Trying to build the perfect setup on day one — start simple, expand after it works

Ultra-Short Cheat Sheet

node -v
npm -v
npm install -g openclaw
export OPENAI_API_KEY="your_key_here"
openclaw gateway start
openclaw status

Next Upgrade Paths

Once your first agent is running cleanly, here is where to go next:

• OS-specific deep-dive setup guide for Mac, Windows, or Ubuntu
• Full Telegram bot setup tutorial
• Starter templates for SOUL.md, USER.md, MEMORY.md, and AGENTS.md
• Adding skills via ClawHub or building your own
• Setting up cron jobs for daily automated reports

We have already documented most of this in the AI Engineering category. Start with the full security-tested Mac setup guide below.