SWE-agent/mini-swe-agent
Minimal 100-line Python coding AI agent
- Chart
- Not on a chart right now
- Stars
- 9.9k
What it is
An AI software engineering agent written in around 100 lines of Python that solves GitHub issues and assists in the command line. It operates using only bash and a linear message history without custom tool-calling interfaces.
Who it is for
Developers and researchers looking for a simple local CLI coding agent, sandboxed benchmark runner, or baseline for fine-tuning and reinforcement learning.
Why it is trending
It gained 57 stars today, reaching 9,816 total stars alongside its v2.4.6 release. Its minimal 100-line Python design still achieves over 74% on SWE-bench verified.
Main features
- Bash-only execution with independent subprocess.run actions
- Model compatibility via litellm, openrouter, and portkey
- Supports local, Docker, Podman, and Singularity environments
- Includes CLI, batch inference, trajectory browser, and Python bindings
Summarized on November 19, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary