OpenClaw: The Open Source Personal AI Assistant That’s Breaking GitHub (And How to Install It in 15 Minutes)
If you follow GitHub Trending, you’ve already seen the number: over 210,000 stars. In approximately 60 days. For a project that started as a weekend hack called “Clawd” — a play on Claude, until Anthropic’s legal team politely suggested changing the name.
That project is now called OpenClaw, and it might be the most interesting thing happening in personal AI right now.
What is OpenClaw?
OpenClaw is a personal AI assistant that runs on your own machine. It’s not a chat interface — it’s a gateway. You install it once, and your assistant starts responding to you from the messaging apps you already use: WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Teams, and more than 40 others.
The premise is simple: you shouldn’t have to open a browser tab or a separate app to talk to your AI. If you already live in WhatsApp or Telegram, your assistant should live there too — on your hardware, under your control.
No SaaS subscription. No data on someone else’s servers. Just bring your own API key (Claude, GPT, DeepSeek, Gemini, or local models via Ollama) and you’re running.
Why Is It Exploding?
The short version: it feels different.
The testimonials are unusually enthusiastic — developers comparing it to “the first time I saw ChatGPT”, people building personalized meditations with TTS, others running it on a Raspberry Pi from their phone while walking the dog. The project went from 9,000 to 60,000 stars in 72 hours at launch, hit 100,000+ in a week, and now approaches 250,000.
The creator, Peter Steinberger (founder of PSPDFKit), built the first version for himself. Within weeks, Sam Altman announced that Steinberger was joining OpenAI. OpenClaw continues as an open source foundation project.
The reaction makes sense once you use it. It’s not a tool you open — it’s an assistant that’s always active, accessible from wherever you already are.
The Architecture in Simple Terms
OpenClaw runs as a persistent Node.js process called the Gateway. That process:
- Connects to your messaging channels (Telegram, WhatsApp, Discord, etc.)
- Routes your messages to the AI model you chose
- Executes Skills — plugins that give your assistant real capabilities
Skills are where things get interesting. Out of the box you get more than 50. The community published more than 3,000 on ClawHub (clawhub.ai). Skills can read and write files, execute shell commands, control a browser, integrate with GitHub, manage your calendar, post to Slack, query databases, and much more.
Skills can also be written by your own assistant — OpenClaw can create new skills when you ask it to do something it doesn’t know how to do yet.
Setup: From Zero to Running in 15 Minutes
Prerequisites: Node.js 22.16+ (Node 24 recommended) and an API key from any supported provider.
Step 1 — Global installation:
npm install -g openclaw@latest
# or with pnpm:
pnpm add -g openclaw@latest
Step 2 — Run the onboarding wizard:
openclaw onboard --install-daemon
This wizard guides you through setting up your API key, choosing your model, and connecting your first channel. The --install-daemon flag installs the Gateway as a background service (launchd on macOS, systemd on Linux) so it starts automatically and keeps running.
Step 3 — Connect a channel
The onboarding wizard takes care of this. For those starting out, Telegram is the fastest — takes about 2 minutes. WhatsApp requires scanning a QR code. Discord and Slack need a bot token from their developer portals.
Step 4 — Send your first message
Once your channel is connected, open Telegram (or whichever you chose) and say hello. Ask: “What can you do?” — it will list the available skills. Then start exploring.
Step 5 — (Optional) Explore the web UI
OpenClaw includes a control panel in your browser at http://localhost:18789. You can chat, manage skills, configure channels, and monitor sessions from there.
Useful Config
If you’re on WhatsApp and want to restrict who can talk to your assistant:
{
"channels": {
"whatsapp": {
"allowFrom": ["+5491112345678"],
"groups": {
"*": { "requireMention": true }
}
}
}
}
The config lives in ~/.openclaw/openclaw.json. You can also configure your model there:
{
"model": {
"provider": "anthropic",
"id": "claude-opus-4-6"
}
}
What People Are Using It For
From the community:
- Navigate Instagram from Telegram and ask the assistant to create posts
- Connect WHOOP data and receive daily summaries via WhatsApp
- Run Claude Code on your laptop remotely from your phone, via Telegram
- Automate the organization of tax documents
- Content creation and publishing pipelines
- Code review on the go
The pattern: tasks that normally required sitting at your computer are now accessible from your phone, in the apps you’re already using.
Things to Keep in Mind Before You Start
Security matters here. OpenClaw runs with significant local access — files, terminal, browser. Read the security documentation before exposing it outside localhost. The team is actively improving the security model, but prompt injection is an unsolved problem at the industry level, and they themselves say so explicitly.
Model choice affects quality. The documentation recommends the most powerful model available for better results and lower prompt injection risk. For Anthropic users, that means Opus.
It’s in active development. Frequent releases, very active community. Things break and get fixed quickly.
The Big Picture
The SaaS AI assistant market is built on a model where your data, context, and conversations live in someone else’s infrastructure. OpenClaw directly challenges that model — not ideologically, but in practice.
When your assistant runs on your own machine, with your own keys, talking to you through apps you already use, the product experience improves: it’s faster, more private, and more extensible than any hosted alternative.
Over 210,000 developers starred this in 60 days. It’s not just hype — it’s a signal about where personal AI is heading.
Are you already running OpenClaw or evaluating it? What channel did you connect first and which skill did you find most useful? Let us know in the comments 
Links: