Femtobot: Lightweight Automation for Resource-Constrained Environments

A new project called Femtobot aims to provide agent automation capabilities on resource-constrained machines. Written in Rust, Femtobot comes as a single binary of approximately 10MB, a significant improvement over alternatives that can easily exceed 350MB once dependencies are installed.

The main goal is to run OpenClaw-style workflows on less performant hardware, such as older Raspberry Pis or low-cost VPS instances. Femtobot supports:

  • Telegram polling
  • Local storage (SQLite + vector storage)
  • Tool execution (shell, filesystem, web) via rig-core

The code was developed quickly with the help of AI, prioritizing simplicity and small size over more complex Rust idioms. Despite some imperfections, it proves effective on resource-constrained hardware.

The project is open source and available on GitHub, open to contributions from the community.