Loading...
Loading...
Ultra-lightweight personal AI assistant. 99% smaller than OpenClaw at ~4K lines of Python. MCP-based architecture with pluggable servers. Built for people who want a simple, hackable AI companion.
Ensure you have Python 3.11 or newer installed.
# Ubuntu/Debian
sudo apt update && sudo apt install python3 python3-pip python3-venv
# macOS
brew install python@3.11Download the Nanobot source code.
git clone https://github.com/HKUDS/nanobot.git
cd nanobotCreate and activate an isolated Python environment.
python3 -m venv venv
source venv/bin/activateInstall all required Python packages.
pip install -r requirements.txtConfigure your environment variables and start Nanobot.
cp .env.example .env
# Edit .env with your settings
python -m nanobotTime from process start to first response
Average response time after startup
Nanobot fits easily. TPU could accelerate specific ML tasks.
Use an NVMe SSD instead of microSD for much better I/O performance.
Add a heatsink or active cooling to prevent thermal throttling under load.
Configure swap on a fast storage device for extra memory headroom.
Use a quality USB-C power supply rated for the board's requirements.