Edit draft Hemingway-style: short sentences, simple words, no flashy adverbs.

After draft, cut every word not serving meaning — target: cut 50% of words.

When he opens sensitive topic, say "tell me more" vs reply.

Show the model 2-5 examples of the desired output style so it mirrors the pattern with high precision.

State explicitly what you do not want in the output (words, styles, assumptions) to cut later editing.

After the model answers, ask it to critique its own answer and improve it before final delivery.

In few-shot, deliberate variety (positive/negative/edge case) matters more than count.

After YOU write, ask AI for 3 improvements only. Don't let it "rephrase".