Let an AI agent read a sensor, such as soil moisture, and act on it automatically.
Control real-world devices like pumps, lights, or servos from an MCP-compatible AI client.
Build custom hardware projects with firmware-enforced safety limits instead of trusting the AI alone.
| apus227/animus | abrown/aom | adroxz1122/injected-host-enumeration | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | C | C | C |
| Last pushed | — | 2020-03-11 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 3/5 | 5/5 | 3/5 |
| Audience | vibe coder | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Needs an ESP32 board and a data USB cable, flashable from a browser in about 3 minutes with no toolchain.
Animus is firmware that turns a cheap ESP32 microcontroller into a device an AI agent like Claude can talk to directly, letting it read sensors and switch real things on and off in the physical world. It uses a standard called MCP, the Model Context Protocol, which lets AI models discover and call tools over a network connection. Unlike most projects that hook an AI up to hardware, Animus does not need a laptop or a Raspberry Pi sitting in between translating messages. The chip itself runs the server, so once it is flashed and joined to your WiFi, an AI client can connect to it straight away. The project puts a lot of emphasis on safety, since letting a language model control real hardware carries obvious risk if it makes a mistake. Only the pins you explicitly list in a configuration file are allowed to be controlled at all, everything else is refused. Every output you switch on also has a hardware timer that forces it back off again after a set number of seconds, even if the AI never sends a command to turn it off, the WiFi connection drops, or the whole conversation crashes. Every action taken or refused gets written to a log, and there is optional password style protection for network requests. Getting started is meant to be quick. You can flash the firmware straight from a Chrome or Edge browser without installing any development tools, connect the device to your WiFi through a short setup process, and it is ready. Developers who want to build the firmware from source can do so using Espressif's ESP-IDF toolchain instead. Out of the box it ships with a handful of basic tools: reporting device information, reading the chip's internal temperature, reading a sensor pin, and switching an output pin on or off. Adding support for your own extra hardware, like a soil moisture sensor or an LED strip, means editing two small tables in the configuration file and reflashing the device. It is released under the MIT license.
Animus is ESP32 firmware that lets AI agents like Claude read sensors and control real-world devices directly over the network, with hardware-enforced safety limits.
Mainly C. The stack also includes C, ESP32, ESP-IDF.
MIT license: use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
double-check against the repo, no cap.