Mimiclaw is an OpenClaw-like AI assistant for ESP32-S3 boards

MimiClaw is an OpenClaw-inspired AI assistant designed for ESP32-S3 boards, which acts as a gateway between the Telegram messaging application and Claude online LLM to control the hardware by just chatting to it.

We’ve just written about PicoClaw, an ultra-lightweight personal AI Assistant for cheap Linux boards that just needs 10MB of spare RAM. It was itself inspired by Nanobot, a lightweight assistant written in Python, that’s 99% smaller, in terms of lines of code, than the original OpenClaw project that started it all. Since most of the processing is done through messaging apps and online LLMs, it was only a matter of time until this type of solution was ported to microcontrollers.

MimiClaw OpenClaw ESP32-S3

MimiClaw highlights:

  • Written in C; relies on the ESP-IDF 5.5 framework
  • System requirements – ESP32-S3 board with 16 MB flash and 8 MB PSRAM, such as the LILYGO T7-S3, FireBeetle 2 ESP32-S3, ESP32-S3-DevKitC-1-N16R8, Seeed Studio’s XIAO ESP32S3 Plus, and others.
  • Integrates with the Telegram app and Claude, requiring a @BotFather bot token and Anthropic API key
  • Remember across reboots
  • Low-power – 0.5 Watt power consumption

The way it works is explained in the diagram below. The user sends a message on Telegram, which is picked up by the ESP32-S3 over WiFi, and fed into an agent loop connecting to Claude. You can ask it high-level tasks like reminding you of an appointment, but the most interesting part here is that you can control the hardware (GPIO, sensors, actuators…), for example, to read temperatures, flip switches (lights, fans…), and so on, all within Telegram.

MimiClaw Architecture
MimiClaw architecture

Data is stored in text files on the board, which you can access or edit as needed:

  • SOUL.md – The bot’s personality. Default file:
  • USER.md – Info about you — name, preferences, language
  • MEMORY.md – Long-term memory (things the bot should always remember)
  • Date formatted files like “2026-02-14.md” – Daily notes
  • tg_12345.jsonl – Chat history

If you want to give it a try, you’ll need a machine with ESP-IDF 5.5 or greater, and get the code as follows:


Before building it, you’ll probably want to edit the main/mimi_secrets.h file with your WiFi credentials and Telegram/Claude/Brave Search (optional) API keys:


Time to build and flash MimiClaw to your ESP32-S3 board (replace /dev/ttyACM0 with your actual device):


ESP32-S3 USB Port flash firmware
The board should be connected to the USB-C port marked “USB” during flashing

There are also CLI commands to change the parameters defined in main/mimi_secrets.h and debug commands for wifi status, free memory, and more. You’ll find detailed instructions to get started and the code, released under an MIT license, on GitHub. The project’s website may have a few more details, as well as a waiting list, which must be for people wanting to purchase an ESP32-S3 board preloaded with MimiClaw.

Share this:
FacebookTwitterHacker NewsSlashdotRedditLinkedInPinterestFlipboardMeWeLineEmailShare

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links.

Radxa Orion O6 Armv9 mini-ITX motherboard

Leave a Reply

Your email address will not be published. Required fields are marked *

Boardcon MINI1126B-P AI vision system-on-module wit Rockchip RV1126B-P SoC
Boardcon MINI1126B-P AI vision system-on-module wit Rockchip RV1126B-P SoC