Arguing with my AI showed me what I was missing

I asked my AI to move an old file into the history folder. It told me Living Projects didn't have a button for that. I built Living Projects.

It was right, there's no button. Living Projects keeps the record in your own storage, and the AI already has a connector for that storage, so it can move the file directly. I just hadn't described that in HOME, the file that explains how Living Projects works.

I handle old information two ways. Supersede is for a fact that changed: the old entry stays, marked. Moving to history is for wording I want out of what gets read. Only the first was written down.

So I added a short paragraph and ran a small test, one run each: the same short request, to take one word out of an old entry, in two test projects, one with the paragraph and one without.

Without it, the AI superseded the entry and reported that it "has moved to history." The word was still in the list.

With it, the AI asked what the new wording should be, moved the file whole and put a clean copy in its place. The word was gone from the list.

What part of your work do you keep explaining to your AI because you've never written down how it works?