A green check from the wrong commit is still the wrong evidence

[daichunghy]

One failure mode keeps coming up in the PatchGate design: a check can be green and still say nothing about the pull request currently waiting for review.

For example, a contributor pushes a new commit after CI finishes. The old check remains visible, but the evidence is now attached to a different SHA. A similar problem appears when a workflow reports against a test merge or merge-group revision and the evaluator assumes it was the branch head.

How do you make this distinction visible to maintainers today? Do you show the tested SHA in the PR UI, rely on required-check behavior, or leave it to people to notice? I am interested in the cases where the current setup is easy to misunderstand.