Local Agent Can Fine-Tune Itself While You Use It

sqashTomato · reddit · 2026-07-19

A developer built a local Agent Harness that runs entirely offline on a Mac without API dependencies. It saves notes, browses the web, executes tasks, and "learns while you use it" by fine-tuning itself after users correct its mistakes. The key innovation is embedding **fine-tuning directly into the error-correction workflow**. When web elements change and previous automation steps fail, the Agent attempts self-correction to improve future stability. The project is open-source, currently utilizing Apple's MLX framework and LoRA for local training to leverage unified memory. The creator is seeking help for a CUDA port and further project polish.

Original post →

More from coding & agent

coding & agent channel →