One of the most underused prompt engineering techniques: telling Claude what to avoid, not just what to produce.
AI models have learned from enormous amounts of text that includes all the clichés, filler phrases, and overused constructions you want to avoid. Without explicit instructions, they default to those patterns.
When you name the specific patterns you don't want, Claude avoids them much more reliably than when you only describe what you do want.
Keep a running list of phrases and patterns that consistently appear in AI outputs and that you always delete. Paste this list into every prompt as a 'Do NOT use' instruction block. The output quality improvement is immediate.