SWE-Agent is an open-source Artificial Intelligence tool designed to self-governingly address software engineering tasks by transforming language models (LMs) like GPT-4 into software engineering agents. It operates on a custom Agent-Computer Interface (ACI) that enhances the LM’s ability to navigate, edit, and execute code commands efficiently. SWE-Agent is capable of taking GitHub issues as input and generating pull requests as solutions, demonstrating its utility in real-world coding environments. It achieves impressive performance on benchmarks like SWE-bench, showcasing its potential in automating complex software development processes.