Prompt Engineering Best Practices

Chapter 1 · Practice 1 of 4

Write the prompt as a brief for a capable colleague with no background

From Clear instructions and context

Why it matters

Anthropic's prompting guide puts it as a golden rule: show your prompt to a colleague with minimal context on the task and ask them to follow it. If they would be confused, the model will be too.

A good brief answers the questions a newcomer would ask:

  • What is the task? What should be produced, and for whom.
  • What does done look like? Length, format, fields, level of detail.
  • What are the constraints? What must be included, what is out of scope.
  • What happens at the edges? Missing data, ambiguous input, off-topic requests.

If you want more than the minimum, ask for it. The same guide notes that "above and beyond" behaviour has to be requested explicitly rather than inferred from a vague prompt.

Do

  • State the audience and the purpose of the output
  • Say what to do when the input is missing, ambiguous or off-topic

Avoid

  • One-line prompts for tasks that need judgement
  • Relying on the model to infer quality bars you never stated