Make Support Boring, On Purpose

You don't need a hundred support prompts. You need five you trust and actually reuse.

Make the output reviewable

Write down what you never want as explicitly as what you do. The exclusions often do more work than the instructions.

• Have the model flag what it could not cover
• Save the prompt that worked so the next person reuses it
• Keep one ask per prompt — chaining beats cramming

Cut before you polish

Lock the parts that never change and leave one slot for the variable. Now every run shares a family resemblance instead of starting from scratch.

• Save the prompt that worked so the next person reuses it
• Name the audience and the single goal before you generate
• Have the model flag what it could not cover

Ask for the reasoning, then the result. When the model shows its steps you can see exactly where to push back.