Tools designed for rewriting, refactoring, and optimizing code should prioritize both speed and accuracy. Large language models (LLMs), however, often lack these critical attributes. Despite these limitations, there remains significant potential for leveraging LLMs to enhance code quality. In a new paper Don’t Transform the Code, Code the Transforms: Towards Precise Code Rewriting using LLMs, a Meta research team proposes a novel chain-of-thought strategy