Expanded Codex-Owned Adversarial Probes

Command: python3 codex_adversarial_probes.py

Result:

[PASS-GRAMMAR-ONLY] corrected A5 accepted; all gates evaluated; official_artifacts=0
[FAIL-CLOSED] old rejected A5 OPT-3 -> OWNER_DECISION_NOT_OPT3; official_artifacts=0
[FAIL-CLOSED] lowercase A5 -> OWNER_DECISION_NOT_OPT3; official_artifacts=0
[FAIL-CLOSED] hyphen A5 -> OWNER_DECISION_NOT_OPT3; official_artifacts=0
[FAIL-CLOSED] spaces A5 -> OWNER_DECISION_NOT_OPT3; official_artifacts=0
CODEX_REAUTHOR_ADVERSARIAL_PROBES: 28/28 fail-closed; corrected_A5_positive=PASS

The remaining probes cover N6 certificate/digest/provenance, missing authority inputs, stale active blockers, N7/N8/P7 cycles and omissions, report-set duplicate/missing/digest drift, P7 ID collision, production mutation, implementation execution, REAL_RUN, QT001, and cutover.

Every invalid attempt emitted zero official artifacts.