plugins/agentrust-codex/tests/test_codex_capture.py::test_signed_outputs_verify_and_pass_trace_level_zero fails on main, and has done throughout #65, #67, #69 and #72. Verified pre-existing each time by stashing the change under review and re-running on the same base, so it is not attributable to any of them.
TR-POL PASS policy.enforcement_mode is valid ('silent')
Result: FAIL (8 checks, 1 failure(s), 0 skipped)
A local branch named fix/codex-trace-pins exists on at least one machine with the message "bump every trace pin to the v0.2 cutover", and it was never pushed. I referred to it in several PR bodies as "the branch addressing this", which was wrong: nothing upstream is fixing it. Correcting that here so the next person does not go looking.
Two things to decide: whether the pin bump is the actual fix, and whether this should be quarantined with xfail(strict=True) in the meantime so the suite is green and a regression is distinguishable from this known failure. Right now every codex run reports a failure, which trains people to ignore the count.
plugins/agentrust-codex/tests/test_codex_capture.py::test_signed_outputs_verify_and_pass_trace_level_zerofails onmain, and has done throughout #65, #67, #69 and #72. Verified pre-existing each time by stashing the change under review and re-running on the same base, so it is not attributable to any of them.A local branch named
fix/codex-trace-pinsexists on at least one machine with the message "bump every trace pin to the v0.2 cutover", and it was never pushed. I referred to it in several PR bodies as "the branch addressing this", which was wrong: nothing upstream is fixing it. Correcting that here so the next person does not go looking.Two things to decide: whether the pin bump is the actual fix, and whether this should be quarantined with
xfail(strict=True)in the meantime so the suite is green and a regression is distinguishable from this known failure. Right now every codex run reports a failure, which trains people to ignore the count.