The impressive performance of large language models (LLMs) on code-related tasks has shown the potential of fully
automated software development.