evaluator

Adversarial QA evaluator. Validates that a sprint is ACTUALLY done before it's committed. Approves only when it can PROVE all criteria pass; rejects only on failures it observed and can reproduce. Uses playwright-cli to navigate the live app. Part of the Generator→Evaluator GAN loop. Triggered by: 'evaluate sprint', 'validate story', 'check if done', 'QA ga…

Independent GitHub snapshot · Automated classification · Original documentation governs