AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
翻译准备中
Apache Airflow software factory: GitHub work orders become verified pull requests through isolated agent work cells and human release gates.
翻译准备中
取自仓库 README
正在加载完整 README。 以下为已保存的摘录。
Durable intent. Disposable execution. Deterministic convergence.
Turn GitHub work orders into reviewable pull requests through a defined process: specification,
plan, human approval, implementation, tests, review, and retained evidence. Apache Airflow owns the
managed lifecycle; isolated coding workers perform the work; the trusted factory publishes the result.
People decide what reaches main.
Install Git, Python 3.12, and uv,
then run the bundled replay:
The demo rebuilds source bytecode before tests so rapid fixes are tested against current code.查看完整 README ↗目前无法加载完整 README。请通过原文链接阅读。 以下为已保存的摘录。
Durable intent. Disposable execution. Deterministic convergence.
Turn GitHub work orders into reviewable pull requests through a defined process: specification,
plan, human approval, implementation, tests, review, and retained evidence. Apache Airflow owns the
managed lifecycle; isolated coding workers perform the work; the trusted factory publishes the result.
People decide what reaches main.
Install Git, Python 3.12, and uv,
then run the bundled replay:
The demo rebuilds source bytecode before tests so rapid fixes are tested against current code.查看完整 README ↗