Back to directory
About
This agent reads GitHub issues and writes code patches to fix them. It's built for developers who want to automate routine bug fixes. The agent navigates codebases, runs tests to verify changes work, and submits solutions without manual intervention.
Key Features
- GitHub issue to pull request pipeline
- Agent-Computer Interface (ACI) design
- SWE-bench benchmark leader
- Configurable LLM backend
- Detailed execution traces
Pros
- Strong benchmark results
- Excellent research documentation
- Fully open source
Cons
- Research-grade — not production-hardened
- Setup requires Python environment
- Best results need GPT-4/Claude
Pricing
FreeOpen source (MIT). You supply your own LLM API key.
Categories
CodingDevOps