SWE-agent
To enable language models to autonomously solve real-world software engineering problems through effective Agent-Computer Interfaces (ACI).
At a Glance
- AI Researchers
- Software Developers
- Enterprise Engineering Teams
AI Tools by SWE-agent
(1)
mini-SWE-agent
Lightweight CLI Coding Agent
Discussions
No discussions yet
Be the first to start a discussion about SWE-agent
Latest News
SWE-agent 1.0 + Claude 3.7 achieves SoTA on SWE-bench Full
Releasing SWE-agent 1.0: SoTA on SWE-bench Light & tons of new features
Mini-SWE-agent achieves 65% on SWE-bench Verified in 100 lines of Python
SWE-agent-LM-32b achieves open-weights SOTA on SWE-bench
Products & Services
A framework that turns language models into autonomous software engineers capable of fixing bugs and issues in real GitHub repositories.
A comprehensive evaluation benchmark for software engineering agents, consisting of real-world issues from popular Python repositories.
A simplified, 100-line Python implementation of SWE-agent designed for hackability and rapid experimentation.
Market Position
Leading open-source framework for software engineering agents; sets the standard for research in autonomous code repair.
Leadership
Founders
John Yang
PhD student at Stanford University (previously Princeton). Research focuses on machine learning for software engineering and autonomous agents.
Carlos E. Jimenez
PhD student at Princeton University. Expertise in Natural Language Processing and creating benchmarks for software engineering agents.
Karthik Narasimhan
Associate Professor of Computer Science at Princeton University. Co-director of the Princeton NLP group. Former research scientist at OpenAI and co-author of the original GPT paper.
Ofir Press
Postdoctoral researcher at Princeton University and researcher at Meta AI. Known for contributions to efficient transformer architectures (e.g., ALiBi).
Alexander Wettig
PhD student at Princeton University. Research interests include language modeling and self-supervised learning.
Kilian Lieret
Research Software Engineer at Princeton University. Focuses on developing robust software infrastructure for research projects.
Shunyu Yao
PhD student at Princeton University. Creator of influential agent frameworks like ReAct and Tree of Thoughts.
Executive Team
John Yang
Lead Researcher / Project Co-Lead
Stanford PhD candidate, ex-Princeton.
Carlos E. Jimenez
Lead Researcher / Project Co-Lead
Princeton PhD candidate.
Board of Directors
Founding Story
Started by researchers at Princeton to address the lack of realistic benchmarks for AI coding agents. After creating SWE-bench, the team developed SWE-agent to provide an optimized interface for models to interact with repositories, shells, and file editors.
Business Model
Revenue Model
Open-source project; freely available for research and community use. No commercial revenue model reported.
Pricing Tiers
Available under Apache 2.0 license via GitHub.
Target Markets
- AI Researchers
- Software Developers
- Enterprise Engineering Teams
- Autonomous bug fixing in GitHub repositories
- Automated software maintenance
- Cybersecurity vulnerability detection
- Benchmarking large language models
- Hugging Face
- Cognition AI
- Microsoft/GitHub