Victor, the outgoing request is logged again; I cannot establish a second physical cycle from what we retained. Beth, pending currently has no distinction for possibly sent.

Beth's hold is sensible, but don't label every uncertain job failed. Ruth's first execution might have finished perfectly while the application was having its own bad afternoon.

Mark, the screen needs to say what is uncertain too, because an operator seeing failed will reasonably expect retry to be the next action.

I want that job to remain visible. A separate state is useful. Moving it out of sight would only exchange a repeat request for forgotten work.

Lin, yes, held and visible. I meant remove it from automatic dispatch, not remove the entry. The review itself also needs to survive another application restart.

Sam, our cell lead will own the physical review; Lin and Beth, I'll keep the entry on the main screen with automatic dispatch disabled while its status is uncertain.

Does the controller side retain anything tied to this particular request, rather than just a current finished flag?

A finished flag alone won't identify Ruth's job. We needed identity to stop repeated observations becoming repeated counts.

Hana, same warning for a new number after restart. Giving the retry a fresh identity doesn't make it new physical work or prove the first attempt did nothing.

Victor, not enough retained detail to tie the current flag to this request. The cell lead has the bracket held for review; I haven't declared either execution finished.

That wording is useful, Ruth; could the review screen also show the last known send and why the result is missing, without presenting the flag as an answer?

For the software check, interrupt it around each save and send boundary with dispatch mocked. Then reopen during review. I would want to see no automatic request for every uncertain case.

Ruth, has the startup change been exercised yet? I am interested in what the operator sees on the second reopen, not just the first.

And who can release that hold after the cell lead's review?

The mocked restart checks now leave possibly sent work held, including a second reopen, and the screen shows the last recorded send; physical review and the release responsibility are still open.

Useful software progress. Keep the unresolved release ownership beside it in the handover; the mock proving no resend doesn't settle what happens to this bracket.