An assistant suggests this file-cleanup plan: remove the old folder, copy the files to an archive, then check the archive. Every action sounds familiar. The order makes the plan unusable: it may remove what the next step needs.
A tidy numbered list is still a draft. NIST describes confidently incorrect AI output as a known risk. The file example below is a practical way to review a procedure, not a file-management method prescribed by NIST.
Give each step an entry condition
Ask the assistant to add three things beside every action: what must already exist, how success will be checked, and what should happen on failure. Then inspect those conditions yourself.
For a small folder of workshop notes, a workable sequence could be:
1. Identify the source folder and destination; leave the source intact. 2. Copy the files and wait for completion without reported errors. 3. Compare the expected file list with the destination and open the copied documents there. 4. Keep the source while you resolve missing files or unreadable copies.
For important archives, use a suitable backup process and stronger verification. Opening a few documents does not establish that every byte in a large collection survived.
Make stopping part of the procedure
Try the plan first with disposable sample files. Ask what happens if the destination is full or one file cannot be copied. “Continue with cleanup” is not an acceptable response to an incomplete copy.
Do not include deletion as an automatic final step. Treat it as a separate decision after verification and any required retention checks. The word “copied” in an assistant’s response is not evidence that the files exist.