Here be dragons
Running a LLM based AI agent with shell access is dangerous. There is a risk of prompt injection or adversarial input getting sent to the bot via the enabled communications channels or other text it finds online. The bot could run malicious software, or expose API tokens and other private information that it has access to if proper safeguards are not implemented and maintained. Read and follow the advice on the OpenClaw docs security page and be warned that this is cutting edge technology that is rapidly changing. LLMs are notorious for hallucinating and behaving unexpectedly, even if you follow all of the best practices there is still some risk with this type of tool.
ladyada here! openclaw basically lets an LLM become the 'brains' of a computer, in essence like Claude Code --super-dangerous-living. we're using this to experiment with giving LLMs access to hardware & firmware development without needing a lot of babysitting, but also not giving it full run of a computer. we recommend NOT giving it any keys or tokens or access, for example we made ours it's own github account so it can submit PRs or comment on issues but not with elevated privs. your mileage may vary, this is very cutting-edge and interesting but not necessarily useful. also, it looooves to burn through tokens, so do not give it full leash on billing!
Clawdbot err Moltbot, or maybe OpenClaw is a project that aims to make LLM based AI Agents do things. It allows you to hook up Claude Opus or your preferred model to have free reign over a computer. Like Claude code with memory and intended to be a more generalized personal assistant rather than only to help you write software. It can run commands, write and run programs, and install modules to expand its own capabilities. It hooks into 3rd party communications channels like Telegram, Discord, email and more. It integrates with other systems like Home Assistant, Trello, Apple Notes & Reminders, Notion, GitHub and others. You can talk to it through your preferred chat app and ask it to do things and get back to you when it's done or encounters an issue and needs more direction. Scheduled tasks can be set up to carry out specified prompts repeatedly.
Its recently release (January 2026) been going viral and gaining lots of attention. It supports integrations into several different Apple programs, many people have reported running it inside its own Mac Mini. With the "thinking" being sent off remotely to Opus, we thought surely this could be run on lower powered hardware, so we started experimenting with the Pi 5. Running on a Pi 5 also makes it easy to connect more interesting bits of hardware to give it the ability to sense and interact with the physical world in different ways.
This guide will document the setup of OpenClaw on a Rasperry Pi 5 w/ 8GB RAM as well as how to give it access to a TFT display, temperature/pressure sensor, USB camera, and NeoPixels. On the software side, it will use eSpeak text-to-speech and Whisper Small speech-to-text to enable speech based communication with the bot.
Page last edited February 04, 2026
Text editor powered by tinymce.
