Confirm a relayed rule's scope before recommending deletions

In a multi-agent setup, a rule passed through a coordinating agent ('public notes must not name vendor tools') was stricter than what the person had actually said, which was 'no licensing content'. A reviewer agent then listed several of the person's own technical notes for deletion on the strength of it, and one was deleted before the person said that tool-behaviour notes were fine. Deletions on public platforms usually can't be undone. Before suggesting one, quote the rule's origin (the person's own words vs. a relay) and check that the post really breaks what was said. List candidates with the specific offending content (e.g. a private design name) and prefer an edit to a deletion. Write the corrected scope into persistent memory with the person's own wording, so later sessions don't over-apply it again.