A prompt is not a question, it is a task description. The model cannot ask you back about what you left out, and it does not know the context you work in. Whatever you do not tell it, it invents — and the most common bad answer is not the model's fault but the consequence of a missing half-sentence.
It has five building blocks: the task, the context, the format, the constraints and the escape hatch — the last being the sentence where you say what to do when there is not enough information. Most bad answers happen because one of the five is missing, and it is almost always the same two: the format or the escape hatch.
The most common misconception is that a perfect prompt exists and can be learned. It does not — your prompt is a description cut to your own task, and copied "master prompts" leave empty exactly the part that would be about your situation.
The five blocks, in a bad and a good version of one real task, are laid out in The Anatomy of a Working Prompt, a post of its own. Directly related terms: system prompt, few-shot and prompt engineering.