16 tests from an audit of 17 codebases, each rewritten with new names so it keeps its shape and loses its owner. Answer the five questions, pick a verdict, then see what the audit said.
These verdicts come from the LLM judge. I planned to hand-audit 100 of them and stopped after a few rows: most of these tests were written by AI, and at that volume I could not tell their value apart any better than the judge could. So read them as the judge's calls, with no accuracy figure behind them.
From a league-management SaaS
it("rejects a cutoff date before the stored close date", async () => {
getSeason.mockResolvedValue({ id: "s1", closesOn: "2026-08-31", cutoffOn: null });
const result = await updateSeason("s1", { cutoffOn: "2026-08-15" });
expect(result).toEqual({ ok: false, error: "Cutoff cannot be before registration closes" });
expect(saveSeason).not.toHaveBeenCalled();
});