wilson-score
Statistical pass/fail gating for flaky tests and LLM evals: Wilson lower-bound verdicts, paired A/B, pytest plugin, stdlib only
Analyse efficacy of your own confidence interval (CI) methods
