How to write good prompts? Basics of prompt engineering

A good prompt says what you want, for whom, in what form and based on what, the way you would brief a capable colleague who knows nothing about your project. Most weak answers come from vague requests, not from weak models.

• Task and goal: "Write a 120-word product description for our newsletter" works better than "write something about our product".
• Context: who will read it, what it is for, what you have already tried.
• Material: paste the text the answer should be based on and separate it clearly from the instructions, for example with tags such as <document>.
• Examples: one or two samples of the result you want.
• Format: length, structure, tone, language.
• Steps: for complex jobs, let the model think step by step or split the work into a chain of smaller prompts.
• Testing: decide what a good result looks like, try the prompt on a few real cases and change one thing at a time.

Giving the model a role ("you are an experienced editor") can help with tone, but it does not replace clear instructions and real material.