How should my FR5 startup treat a request with no saved outcome?

Helper disabled for uncertain entries. We can identify this submitted request, but no saved result.

Use that request identity to reconcile available evidence; failure to save a result is not evidence that execution never began.

Who can resolve the uncertain entry, and what evidence do they need before it leaves review?

Our lab once gave everyone a confirm button and nobody the same definition of confirmed. Decide who owns that judgement before the developer turns it into a click.

Quality owns the disposition with the application maintainer. Matching evidence required; missing evidence remains unresolved.

Does a repeated matching result change the count again? Once you're recovering old results, duplicates are quite likely to turn up at the same party.

Offline duplicate replay did increment twice. Maintainer changed the total to current plate dispositions, keeping attempts underneath.

Test interruption during the reconciliation save as well; an agreed decision should not reappear as an ordinary pending request after another restart.

And make the unresolved reason visible to the covering shift, so they do not interpret the lack of a result as permission to create another request.

Tests pass for duplicate result, interrupted save and second restart. Unknown remains outside automatic submission.

Original request reconciled from matching retained result with quality. Installed startup checks passed. Old helper no longer resubmits uncertain work.

Did the covering shift find the reason and decision without the maintainer explaining the database? That's the part I'd want to see before teaching this as finished.