awevolve
Point an agent at a file and a command that scores it, and let it improve.
The Aither World is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. awevolve is one of its 71 bricks — each installs on its own, runs offline, and needs no account. 5 more are in development and 15 are planned.
The problem
Automated improvement loops fail silently and look identical while doing it. A run that explored honestly and found nothing produces the same logs, the same records and the same stop reason as a run that changed nothing at all -- there is no exception to catch and no failing request, so the loop keeps running and a human periodically concludes the search space is just hard. And the loops that DO work are usually one-shot generators: the model is asked for a candidate, handed no history, and never allowed to test its own idea before committing it.
that your optimisation loop is finding anything
every version it kept, the score that version earned, and the edit that produced it
Adopt it
pip install awevolvePoint it at a file and a command that scores that file, and watch an agent improve it -- keeping every version and the score it earned.
Pairs with
Pairing is composition, never dependency — awevolve installs and runs on its own.
The rest of the Aither World
Every brick below is live, drawn from that repository's own published manifest.