The agent kept widening the change set and stalling for the next ask. Handbook already named minimize before polish. When your coding agent expands scope, does it already know the rung that says stop?

Someone deleted the FAIL so the log looked clean. Ranex named it removed-observation. If a bad observation vanishes, does your gate still know it was there?

The model remembered the library. The lockfile disagreed. Leitir refused the checksum. Did your agent read the pinned bytes, or its memory?

4 tests removed from the task. Bare CI stayed green. Ranex failed on missing test IDs. If those tests vanish, does your gate still go green?

The agent wrote the patch and the test that passed. That is self-grading, not a verdict. Who wrote the test that just passed, the same agent that wrote the patch, or someone else?

Empty evidence looks like not ready yet. Ranex fails the required claim. Never a skip. When your gate has no evidence, does it block or default to pass?

The check almost ran the PR to grade it. Ranex judges signed evidence instead. No contributor code runs.

The argv still looked right. The probe bytes moved. Ranex refused the bundle.

Three greens this week. None were verdicts. Missing tests. Wrong library. Stale receipt.