SWE-agent
Developed by Princeton and Stanford researchers, SWE-agent is a software tool that allows language models to use tools autonomously. The project is among the open-source projects on SWE-bench and can be configured with a single YAML file. Developers recommend using the simpler mini-swe-agent version for the time being.
SWE-agent
100SWE-agent · AI Communities & Swarms
- A GitHub issue will automatically attempt to fix it with your chosen language model.
- You configure the agent's tools and behavior with a single YAML file.
- You perform comparison and test operations on SWE-bench.
- • NeutralHacker News
Show HN: Anterion – Open-source AI software engineer (SWE-agent and OpenDevin)
“Hi! It seems like the key thing for approaches like SWE-agent is that constraining the interface and then the actions within that interface is crucial for performance. In the case of the SWE-agent approach, they limit the input into the agent based on just bash terminal output,…”
Verified against README · Oct 25, 2026 · source