# Prompt Framework

Task

• Persona: what expertise you want LLM to have
• Format: how you want the output

Context: Be precise when telling context, for example, instead of use `some`, `under`, we should give the accurate number and range.

References: Give examples of the similar output you had in the past for LLM to be better understanding on what you need.

Evaluate: Check if the output is what we want by the input we provided.

Iterate: If output is not expected, then we can add more information and try again.