Clawbot turns your computer into a 24/7 personal agent accessible from any chat app. Control your browser, execute shell commands, manage files, and automate workflows via WhatsApp or Telegram. Features persistent memory, full system access, local privacy, and 50+ integrations.
Hi Product Hunt!
Most AI assistants are trapped in a browser tab. They can generate text, but they can't do anything. I found Clawbot really interesting - it is gonna change that.
Clawbot is a self-hosted agent that lives on your machine (Mac, Windows, or Linux). It bridges the gap between LLM intelligence and your operating system's capabilities.
What makes it different?
- 🏠 Local & Private: Your data stays on your machine.
- 📱 Chat Anywhere: It texts you back on WhatsApp, Telegram, Signal, or Slack.
- ⚡ Real Action: It doesn't just talk. It can open browsers, fill forms, run terminal commands, and edit files.
- 🧠 Memory: It remembers you. No more starting from scratch every session.
It's fully hackable and open to plugins. I'd love to see what skills you build for it.
Maybe AI moves on to next stage.
Such a great tool, I’ve been playing with it for the past few days and it’s been doing a lot of work for me from managing my calendar to getting daily briefs and developing apps for me
Great project! Two questions: 1. Is there voice control? 2. Can it control a mobile phone?
Persistent memory is key. Are the memories stored as vector embeddings locally, or are you using the Markdown-based "user.md" approach for long-term context?
Why does it say Clawbot some places, Clawdbot other places? I heard it's Moltbot now due to Anthropic but now sure why the discrepancy with the old name?
Either way, I got mine setup and been using it! It's been delightful, though I do feel if you could solve the setup complexity to be more user friendly, this thing could absolutely explode. I have to wonder if you're keeping it intentionally a bit more advanced user-centric for now
Best of luck and thanks for creating this
The product is great, and I'd love to use it, but the installation and configuration process is relatively complicated for people without a technical background. It feels like there's a barrier to entry. Is there a simpler way to use it?
Full system access via a chat app sounds incredibly powerful, but also a bit scary if it's not secure. I love that they mentioned 'local privacy,' though. I wonder if the data stays encrypted on the machine, or if there's a risk of someone else hopping into the chat and running commands.
It’s pretty crazy that you can control your entire computer through WhatsApp or Telegram. It sounds like you'd never have to worry about leaving a file at home again! I’m really curious how smooth the browser control actually is when you're just typing commands from your phone.
This is really exciting. I'd love to get in touch with the team to integrate with our OSS lightweight protocol for making agent actions safer by requiring verifiable intent + provenance + evidence before high-impact tool calls (fail-closed by default). It fits super well with Moltbot's agent/tool workflow (ie. wrapping tool calls in Pi sessions with a quick verification step via SDK/JS port). Could add proactive security for things like browser, nodes, camera, or system exec without much overhead. Would love to explore an integration and happy to prototype a skill/wrapper or submit a PR if you're open. Either way, supported and wish you all the best.
Self-hosted agents with shell access are tricky... once you elevate permissions for one task, containing the blast radius gets hard. Curious if Clawdbot has per-session sandboxing or if elevated permissions persist across the whole session.
This is a very “finally” product - assistants that can’t actually do things are basically autocomplete in a tab. The “chat anywhere” idea is especially compelling too: WhatsApp/Telegram as the UI feels like the most natural interface for real life.
I’m curious how you’re thinking about trust when the agent can run commands and touch files - what does the default “safe” experience look like, and how do users review or undo actions when something goes a bit off?
Congrats on the launch - if you nail safety without killing the flow, this becomes something people keep running quietly in the background.
Congrats on the launch! @steipete - Will be tinkering with Clawdbot over the weekend.
I suspect operating systems will eventually become headless, driven by context and intent rather than windows and clicks. This feels like a step in that direction. Congrats.
this looks really great @steipete. excited to try it out :)
giving a bot "full system access" is both awesome and terrifying. does it run in a sandbox/container so it can't accidentally rm -rf my home folder if i give a vague command?
starting using it a week ago, after I saw someone run it on Raspberry Pi and damn! :)
This is a powerful tool. I installed it on a separate computer that I was not using for personal used. The tool is very powerful and could delete or modify your files.
