AI AGENTS
by SWE-agent
Open-source coding agent that takes GitHub issues and attempts to automatically fix them using a chosen language model.
SWE-agent is an open-source coding agent for automated software engineering. Its public site and documentation describe it as a tool that takes GitHub issues and attempts to automatically fix them using a chosen language model. The project’s arXiv paper describes SWE-agent as a system that helps language-model agents autonomously use computers to solve software engineering tasks through a custom agent-computer interface. Official setup documentation covers installing locally from source and running in a browser-based VS Code environment.
Use SWE-agent to experiment with language-model agents that attempt to resolve GitHub issues automatically. Use it when evaluating open-source software-engineering agents agent-computer interfaces or coding-agent workflows tied to repositories. Use it for research or trials around autonomous software engineering tasks before considering production reliance.
Last updated Jun 25, 2026
