{"id":"swe-agent","name":"SWE-agent","maker":"Princeton and Stanford researchers","url":"https://swe-agent.com/","repo":"SWE-agent/SWE-agent","summary":"A research agent that takes a GitHub issue and tries to fix it automatically in a repository using a language model.","details":"SWE-agent runs from the command line against a repository and an issue, with a configurable agent-computer interface of commands. It was built at Princeton and Stanford and is used in research on agent benchmarks. It is licensed under MIT. The maintainers now recommend mini-swe-agent for new work.","open_source":true,"hosted":false,"also":["benchmarks"],"verified":"2026-11-15","category":"coding-agents","category_name":"Coding agents and agentic IDEs","page":"https://modelmoot.com/tools/swe-agent","reviewed":"2026-11-15","checked":{"at":"2026-11-15T16:10:15.342Z","address_answers":true},"note":"Descriptions are the site's own. Facts under \"checked\" were read automatically from the tool's address and public repository."}