Overview
Princeton’s open-source coding agent that achieves strong SWE-Bench scores, designed for autonomous software engineering tasks.
Details
SWE-agent is an open-source autonomous coding agent from Princeton NLP, designed to fix bugs and complete software engineering tasks autonomously. It achieved leading scores on SWE-Bench (a benchmark of real GitHub issues) when introduced and continues as a research and reference implementation for autonomous coding. SWE-agent uses a custom Agent-Computer Interface for navigating files, running tests, and editing code reliably. SWE-agent is influential in academic AI coding agent research.
Tags
coding, autonomous-agent, open-source, research