memovai/mimiclaw

MimiClaw: Harness on a $5 chip. No OS(Linux). No Node.js. No Mac mini. No Raspberry Pi. No VPS. Hardware agents OS.

What it solves

It enables the creation of a low-cost, low-power personal AI assistant that runs directly on a tiny ESP32-S3 microcontroller chip without requiring a full operating system like Linux or a heavy runtime like Node.js.

How it works

Written in pure C, the project turns an ESP32-S3 board into an agent loop. It connects to WiFi and interfaces with LLMs from Anthropic (Claude) or OpenAI (GPT) via API. The system uses a ReAct pattern to allow the AI to call tools and manage local memory stored on the chip's flash memory (SPIFFS) as plain text files. Users interact with the assistant primarily through a Telegram bot.

Who it’s for

Hardware enthusiasts and developers who want to build a minimal, always-on AI agent using affordable embedded hardware.

Highlights

  • Ultra-minimal footprint: Runs on a $5 chip with no Linux or Node.js.
  • Local Memory: Stores personality (SOUL.md), user preferences (USER.md), and long-term memory (MEMORY.md) locally on flash.
  • Proactive Capabilities: Features a "heartbeat" service that checks a task list and a cron scheduler that allows the AI to schedule its own recurring or one-shot tasks.
  • Tool Integration: Supports web search (via Tavily or Brave) and system clock management.
  • Flexible Configuration: Includes a serial CLI for runtime configuration changes without needing to recompile the code.

Related

  • Dispatch
    Running a 28.9M Parameter LLM on an $8 ESP32-S3

    The esp32-ai project runs a 28.9‑million‑parameter language model on an $8 ESP32‑S3 microcontroller by storing most weights in flash and using per‑layer embeddings, achieving roughly 9.5 tokens per second without any network connection.

  • Project
    nullclaw/nullclaw

    A tiny, high-performance AI assistant infrastructure written in Zig that enables autonomous agents to run on minimal hardware with near-zero overhead.

  • Project
    espressif/esp-claw

    An AI agent framework for ESP32-series IoT devices that enables users to define device behavior through chat and execute decision-making locally on the edge.

  • Project
    xinnan-tech/xiaozhi-esp32-server

    A backend server for the xiaozhi-esp32 project that enables ESP32 devices to become AI assistants with voice, vision, and smart home control capabilities.

  • Project
    SWE-agent/mini-swe-agent

    A minimal AI software engineering agent that uses only bash commands to solve coding tasks, providing a high-performance, hackable alternative to complex agent frameworks.