Flaky Test Detector: Sequential Probability Ratio Test

Mobile CI pipelines rerun failing tests to tell "flaky" from "broken," but a fixed rerun count either wastes runner time on obviously-reliable tests or gives up too early on genuinely intermittent ones. This simulator renders Wald's Sequential Probability Ratio Test as a literal 3D walk: each simulated rerun nudges a log-likelihood-ratio marker up (fail) or down (pass) between a "flaky" ceiling and a "reliable" floor computed from your chosen false-positive and false-negative risk budgets, and the walk stops the instant either boundary is crossed — not after a fixed number of tries. Adjust the hidden true fail probability to see how a genuinely reliable test resolves in a handful of reruns while a borderline-flaky one can wander for dozens, and watch the running average of reruns-to-verdict across repeated trials converge to the theory's expected sample size.