$ paw-test judge compare.json --suite suite.yaml --out verdicts.json
exit code: 0
--- stdout ---
Verdict differs between A and B (1/3):

  Input: today
    A: 2026-09-11 -> True (matches the ISO-8601 shape the spec asks for)
    B: tomorrow -> False (not an ISO-8601 date)

judge disagrees with assertions (A) (1/3):
  Input: February 30th
    Output: INVALID
    rule_passed=True judge=False (not an ISO-8601 date)

judge disagrees with assertions (B) (1/3):
  Input: February 30th
    Output: INVALID
    rule_passed=True judge=False (not an ISO-8601 date)

Summary: A pass rate 66.7% (2/3), unparseable A 0, errored A 0, B pass rate 33.3% (1/3), unparseable B 0, errored B 0
Wrote verdicts to verdicts.json
