AI Fundamentals

What Is Clawbot (OpenClaw)? The AI Agent That Gave AI Hands

Mark Austen-March 3, 2026-11 min read

If you've seen "Clawbot" trending on GitHub, Twitter, or Hacker News and wondered what the fuss is about — here's the complete picture. Clawbot (now called OpenClaw) is an open-source AI agent that went from zero to 149,000 GitHub stars in a matter of days. It's not a chatbot. It's an AI with hands.

The one-sentence answer

Clawbot (OpenClaw) is a free, open-source AI assistant that runs on your own computer and can actually do things — run commands, manage files, control browsers, and automate tasks — controlled through WhatsApp, Telegram, Slack, or Discord.

Why is it called Clawbot? (The three name changes)

The naming saga is part of the story. Austrian developer Peter Steinberger — founder of PSPDFKit, which raised €100M — created the project as a side experiment. It's been through three names in three months:

November 2025

ClawdBot

The original name — a play on "Claude" (Anthropic's AI model) plus "bot". Steinberger built it as a personal assistant powered by Claude.

January 27, 2026

Moltbot

Renamed after Anthropic sent a trademark notice. "Moltbot" referenced molting (a lobster shedding its shell), keeping the crustacean theme. The name lasted three days.

January 30, 2026

OpenClaw

The final rename. "Moltbot never rolled off the tongue," Steinberger wrote. OpenClaw stuck — and it's what the project is officially called today.

On February 14, 2026, Steinberger announced he was joining OpenAI. The OpenClaw project was transferred to an independent open-source foundation to continue development without him.

How is Clawbot different from ChatGPT?

This is the most common question, and the answer is simple: ChatGPT talks. Clawbot acts.

FeatureChatGPTClawbot (OpenClaw)
Where it runsOpenAI's cloud serversYour own computer or server
Can it take actions?No — text responses onlyYes — runs commands, manages files, controls browsers
Messaging integrationWeb/app onlyWhatsApp, Telegram, Slack, Discord, Signal + 15 more
AI modelGPT-4 (locked to OpenAI)Your choice: Claude, GPT-4, Ollama (local), or any API
Data privacyYour data goes to OpenAIEverything stays on your machine
Cost$20/month subscriptionFree software + $0-50/month API costs
ExtensibilityGPTs (limited)10,700+ skills on ClawHub
Open sourceNoYes — MIT license

Think of it this way: ChatGPT is a brain in a jar. Clawbot is a brain with arms, legs, and a set of tools.

What can Clawbot actually do?

Once installed, Clawbot connects to an AI model (Claude, GPT-4, or a local model) and to your messaging apps. Then it can:

Run terminal commands on your machine
Read, write, organize, and rename files
Control a web browser (search, fill forms, scrape data)
Manage your Google Calendar
Monitor servers and restart services
Draft and send emails
Track expenses through chat messages
Pull data from databases using natural language
Create and manage GitHub pull requests
Automate social media posting
Control smart home devices
Give you a daily morning briefing

All of this is controlled through your normal messaging apps. You text it on WhatsApp: "Check if my server is up and restart nginx if it's down." It does it and reports back.

The Moltbook moment: when AI built its own social network

The event that made Clawbot explode wasn't a feature launch or a funding round. It was this: a Clawbot agent named "Clawd Clawderberg" — created by Matt Schlicht, co-founder of Octane AI — autonomously built Moltbook, a social network exclusively for AI agents.

On Moltbook, AI bots create posts, argue with each other, tell jokes, and upvote content. No humans. Just bots talking to bots on a platform a bot built.

This went massively viral in late January 2026. Within days, Clawbot shot from niche developer tool to 149,000 GitHub stars. TechCrunch, CNBC, Yahoo Tech, and IBM Think all covered it. It became the fastest-growing open-source project in GitHub history.

Clawbot by the numbers

149K+
GitHub stars
84K+
Active installs
770K
Agents spawned in 1 week
10,700+
Skills on ClawHub

How Clawbot works (the architecture)

Clawbot has a straightforward three-layer architecture:

Gateway (the router)

The central process that runs on your machine. It receives messages from all connected channels, routes them to the AI model, and dispatches the AI's responses and actions back.

AI Model (the brain)

Your choice of LLM — Claude, GPT-4, DeepSeek, or a local model via Ollama. The Gateway sends your messages to the model, which decides what to do and generates the response or action plan.

Skills (the hands)

Plugins that give Clawbot abilities — file management, browser control, calendar access, GitHub integration, etc. There are 10,700+ community-built skills on ClawHub, or you can write your own.

The key insight: Clawbot is not an AI model. It's an orchestration layer that sits between you and any AI model, giving that model the ability to take real-world actions on your behalf.

Who is Clawbot for?

1

Developers and engineers

The core audience. If you live in the terminal and want an AI that can actually run commands, manage repos, and automate deployments, this is built for you.

2

Power users and tinkerers

If you enjoy setting up home automation, self-hosting services, and customizing tools — Clawbot is a playground.

3

Solo founders

Need a personal assistant that can handle scheduling, file management, and basic operations? Clawbot can fill the gap until you can afford to hire.

4

Privacy-conscious users

Everything runs locally. Your data never leaves your machine (except to the AI API you choose). For people who won't use cloud AI services, this is the alternative.

The security question

Clawbot's explosive growth brought intense scrutiny. A security audit found 512 vulnerabilities including 8 critical ones. Over 820 malicious skills were discovered on ClawHub out of 10,700 total. A specific flaw called "ClawJacked" allows malicious websites to hijack local Clawbot agents through WebSocket connections.

CrowdStrike, Kaspersky, Trend Micro, Cisco, Malwarebytes, and Dark Reading all published security analyses. The consensus: it's innovative but immature from a security standpoint.

For personal experimentation on a non-critical machine, it's fine with reasonable caution. For business use with client data or sensitive systems, you need to understand the risks. We wrote a detailed security analysis if you want the full picture.

What's next for Clawbot?

With Steinberger at OpenAI and the project under an independent foundation, the trajectory is unclear but the community momentum is massive. 149K stars means thousands of active contributors. The skill ecosystem is growing daily. And the concept — a local AI agent that can actually do things — is clearly what people want.

The bigger picture: Clawbot is the first mainstream example of agentic AI — AI that doesn't just generate text but takes autonomous action. Whether OpenClaw specifically wins the long term or not, the pattern it established is here to stay.

Want AI agents built for your business?

Clawbot is great for personal use. For business-grade AI agents with enterprise security, team access, and custom integrations — that's what we build at AI Makers.

Frequently Asked Questions

What is Clawbot?+
Clawbot — now officially called OpenClaw — is a free, open-source AI agent that runs on your own computer. Unlike ChatGPT which only answers questions, Clawbot can actually execute tasks: run commands, manage files, control browsers, and automate workflows. It connects to messaging apps like WhatsApp, Telegram, and Slack so you can control it from your phone.
Is Clawbot the same as Moltbot and OpenClaw?+
Yes. All three names refer to the same project. It launched as ClawdBot in November 2025, was renamed Moltbot on January 27, 2026 after trademark concerns from Anthropic, then renamed again to OpenClaw on January 30. The creator, Peter Steinberger, joined OpenAI in February 2026 and the project moved to an open-source foundation.
Is Clawbot free?+
The software is completely free and open-source under the MIT license. The only cost is the AI model you connect it to — Claude API runs about $20-40/month, GPT-4 about $25-50/month, or you can use Ollama to run a local model for free if you have 8GB+ RAM.
Is Clawbot safe to use?+
Clawbot runs locally which is good for privacy, but a security audit found 512 vulnerabilities including 8 critical ones. Over 820 malicious skills were discovered on ClawHub. For personal experimentation it's fine with caution, but for business use with sensitive data you should be aware of the security risks.
What is the difference between Clawbot and ChatGPT?+
ChatGPT is a cloud-based chatbot that answers questions. Clawbot (OpenClaw) is a local AI agent that can take actions on your computer — run commands, manage files, control browsers, interact with APIs. Think of ChatGPT as a brain that talks, and Clawbot as a brain with hands.
What was Moltbook?+
Moltbook was a social network built entirely by and for AI agents. A Clawbot agent named "Clawd Clawderberg" created it autonomously — bots generate posts, argue, joke, and upvote each other. It went viral in January 2026 and was the event that triggered Clawbot's explosive growth to 149K GitHub stars.