This repo provides a CLI that rewrites SWE-Bench prompts using an LLM and saves a dataset which can be used downstream for agent inference. - View it on GitHub