What is Moltbot? The AI Butler That Runs on Your Machine

So you've heard about Moltbot and you're wondering what the fuss is about. Fair enough. Let me break it down.

Moltbot is basically an AI agent platform that runs locally on your own machine. You install it, you run it, and it does stuff for you. Think of it like having a really eager intern sitting inside your terminal, except this one doesn't need coffee breaks and won't accidentally delete your production database. Hopefully.

Wait, Wasn't This Called Clawdbot?

Yeah. If you've been around since the early days, you might remember the project under the name Clawdbot. That name had to go because of an Anthropic trademark issue. The team renamed it to Moltbot and honestly, the new name grew on me pretty fast. Molting -- shedding the old skin, becoming something new. Kinda poetic if you squint.

If you see old forum posts or GitHub issues referencing "Clawdbot," that's the same project. Just a different name.

What Does It Actually Do?

Okay here's the real stuff.

Moltbot runs as a daemon on your machine. Once it's up, you get a web dashboard at http://127.0.0.1:18789/ where you can talk to AI agents, manage tasks, and configure how everything works. The agents can do things like:

• Read and write files on your system
• Run shell commands
• Browse the web
• Chain together multi-step workflows
• Talk to APIs

It's not just a chatbot wrapper. The agents actually do things. That's the whole point.

Why Not Just Use ChatGPT or Claude Directly?

Good question. You totally can. But here's where Moltbot gets interesting:

It runs on your machine. Your data stays local. No uploading sensitive code to some random API endpoint and hoping for the best. The agents have direct access to your filesystem, your terminal, your local dev environment. That's powerful.

It's persistent. The daemon keeps running. You can set up tasks, walk away, come back later. It's not a one-shot conversation that vanishes when you close the tab.

It's extensible. You can configure agent behaviors, sandbox modes, connect different AI backends. It's built for tinkering.

That said -- if you just want to ask a quick question about Python syntax, you don't need Moltbot. Use whatever chat interface you want. Moltbot shines when you need agents that actually interact with your system.

How Big Is This Project?

As of the latest stable release v2026.1.24, Moltbot has over 60,000 GitHub stars. So yeah, it's not some tiny side project. There's a real community behind it, active development, and plenty of people using it in production workflows.

The repo lives at github.com/steipete/moltbot if you want to poke around the source.

System Requirements

Before you get too excited, let's make sure your machine can run it.

• Node.js >= 22 -- yes, 22. Not 18, not 20. You need 22 or newer.
• Mac -- works great. Uses launchd for the background service.
• Linux -- works great. VPS setups are popular too.
• Windows -- needs WSL2. Native Windows is NOT supported. I know, I know. But WSL2 works fine and honestly it's not that painful to set up.

Minimum specs for a VPS deployment: 1 vCPU, 1GB RAM. That's it. Pretty modest.

The Dashboard

This is where you'll spend most of your time once everything is installed. Run:

moltbot dashboard

And open http://127.0.0.1:18789/ in your browser. You get a nice UI where you can chat with agents, see running tasks, check logs, adjust settings. It's not the prettiest thing in the world but it gets the job done.

Architecture (The Short Version)

I'm not going to draw you a 47-box diagram. Here's the gist:

1. The daemon runs in the background on your machine
2. The gateway listens on loopback (127.0.0.1) -- important for security, it's not exposed to the internet by default
3. Agents are the AI workers that execute tasks
4. The dashboard is a web UI that talks to the daemon

Agents can run in a sandbox mode for safety. The Docker setup has a neat sandbox config where you set agents.defaults.sandbox.mode to "non-main" and agents run in isolated containers. Pretty slick if you're paranoid about what they might do. And you should be at least a little paranoid.

Troubleshooting Already?

If something's broken before you even start, there's a built-in diagnostic tool:

moltbot doctor

This checks your Node.js version, daemon status, config files, the works. Run it before you start filing GitHub issues. Saves everyone time.

Quick Install (Just to Tease You)

Full install guides are in the next pages, but if you're impatient:

npm install -g moltbot@latest
moltbot onboard --install-daemon

That's it. Two commands. Then run moltbot dashboard and you're off.

For VPS installs there's a one-liner too:

curl -fsSL https://clawd.bot/install.sh | bash

Yeah, the install script URL still uses the old clawd.bot domain. History is funny like that.

Should You Use Moltbot?

Depends on what you want.

If you want a local AI agent platform that can actually interact with your system, automate workflows, and run persistently in the background -- yes. Absolutely.

If you want a simple chatbot for answering questions -- probably overkill. Just use a regular chat interface.

If you're on Windows without WSL2 -- sorry, not yet. Get WSL2 set up first.

If you like tinkering with tools and don't mind reading docs when things break -- you'll love it.

What's Next?

Head to the Getting Started guide to actually install this thing and get your first agent reply. It takes about 10 minutes if everything goes smoothly. Maybe 30 if you hit a Node.js version snag. Been there.

The platform-specific guides cover Mac, Windows, and Docker setups in detail. Pick the one that matches your situation.

Let's go.