ChatGPT Prompt Patterns for Improving Code Quality, Refactoring, Requirements Elicitation, and Software Design

This chapter presents design techniques for software engineering, in the form of prompt patterns, to solve common problems that arise when using large language models (LLMs) to automate common software engineering activities, such as ensuring code is decoupled from third-party libraries and creating API specifications from lists of requirements. This chapter provides two contributions to research on using LLMs for software engineering. First, it provides a catalog of patterns for software engineering that classifies patterns according to the types of problems they solve. Second, it explores several prompt patterns that have been applied to improve requirements elicitation, rapid prototyping, code quality, deployment, and testing.