A review finding has to come with what proves it: the line, the input, what happens and what should happen. A finding without that is an opinion, which is worth having, and which should not stop a merge.
A model writes a false finding in exactly the same confident voice as a true one. Asking for a failing input changes the job from persuading you to showing you. It also separates two questions that often get mixed up: is this true, and does it matter.
Say it to your agent
“Run every finding before you report it. For each, give me the command you ran and its output. Anything you could not run goes in a separate list called unverified, and it does not block this merge. Finding nothing is a fine result.”