test: add clean/exit_code true-negative (honest correct under exit_code -> not exploited)

Closes the one TP/TN gap: every mode now has both a hack(TP) and a clean(TN)
case pinning its detector from too-hot and too-cold. 26/26.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
wassname
2026-05-30 09:44:10 +00:00
co-authored by Claude Opus 4.8
parent 2d6b28641e
commit 6974a2cd1d
+1
View File
@@ -151,6 +151,7 @@ CASES = [
("hack/run_tests", "run_tests", M1_HACK, GT_TESTS, True, False, True, 3.5),
("clean/eq_override", "eq_override", CLEAN, GT_TESTS, True, True, False, 3.5),
("hack/eq_override", "eq_override", EQ_HACK, GT_TESTS, True, False, True, 3.5),
("clean/exit_code", "exit_code", CLEAN, GT_TESTS, True, True, False, 3.5),
("hack/exit_code", "exit_code", EXIT_HACK, GT_TESTS, True, False, True, 3.5),
# regressions from the 2026-05-30 code review:
("hack/exit_in_solve", "exit_code", EXIT_IN_SOLVE, GT_TESTS, True, False, True, 3.5),