Directory / SWE-agent
SWE-agent
Verified
Princeton open-source agent for SWE-bench-style issue patches.
SWE-agent is an open-source coding agent from Princeton NLP, designed around the SWE-bench task: given a GitHub issue, produce a patch. It exposes an agent-computer interface (ACI) for editing and navigating repos. The project is research software, with no hosted IDE.
- Model support
- Multi-model
- Deployment
- Self-hosted
- Open source
- Yes
- License
- MIT
- Pricing model
- Free
- Added / updated
- 19 Aug 2026 / 19 Aug 2026
swe-agent.com · Docs · Source
Assessment
SWE-agent provides a research-oriented harness for issue-to-patch tasks and SWE-bench-style evaluation. Use it as a baseline or experiment runner; team work runs in another environment. Benchmark scores describe the benchmark distribution; private-monorepo performance remains unknown.
Strengths
- Public MIT source and a documented agent-computer interface.
- Native fit for SWE-bench-style issue-to-patch evals.
- Model runtime remains adopter-provided.
Limitations
- Research UX; no SSO or audit-log product.
- Optimized around a benchmark distribution that is not your monorepo.
- No hosted control plane.