A local agent harness: a small, fast Rust runtime that turns any open-weights
language model into a persistent agent — one that learns in real time, builds a
library of verified skills, thinks fast or deep depending on the question, and
would rather say "I don't know yet" than invent an answer.
The model handles reasoning. The harness gives it a body, a memory, and judgment.
Most local LLM setups are stateless. Close the terminal and everything the… See the full description on the dataset page:
https://huggingface.co/datasets/geraldaton20/ae_engine.
