Mini SWE-Agent

Active

Description

Mini SWE-Agent is a minimalist AI agent in just 100 lines of code that solves GitHub issues or assists developers in the command line, demonstrating core coding agent capabilities with minimal implementation.

Key Features

Minimal ~100-line Python agent class achieving >74% on SWE-bench verified
Bash-only tool interface requiring no special LLM tool-calling capabilities
Completely linear message history for easy debugging and fine-tuning
Subprocess-based execution enabling trivial sandboxing with Docker/Podman
Broad model compatibility via litellm, OpenRouter, and Portkey
Supports local environments, Docker, Singularity/Apptainer, Bubblewrap sandboxes

Use Cases

💡 Solving GitHub issues autonomously from the command line

💡 Benchmarking and evaluating LLM coding capabilities

💡 Fine-tuning and reinforcement learning research baselines

💡 Rapid prototyping of coding agent workflows

💡 Educational demonstrations of minimal agent architecture

Strengths & Limitations

✅ Strengths

• Actively maintained, recent updates
• High community interest (6.9k stars)
• Permissive open-source license (MIT)
• Responsive to issues, low backlog

Quick Start

Install: pip install mini-swe-agent. Run: mini solve <github_issue_url> --model_name <model>. For Python API: from minisweagent.agents.default import DefaultAgent; agent = DefaultAgent(model, env); agent.run("Fix the bug"). Supports Docker with: --environment docker.

Visit GitHub

Related Projects

AutoCodeRover

3.1k · Python

Stale C

AutoCodeRover is a project structure-aware autonomous software engineer agent that achieves automated program repair and issue resolution by understanding the overall codebase architecture.

codingpythonagent +2

· Fully autonomous bug fixing and feature addition on GitHub issues using LLM-powered program analysis
· Program Structure Aware code search via abstract syntax tree analysis for precise context retrieval
· Statistical fault localization using test cases to achieve higher repair rates when tests are available

View on GitHub

DeepCode

16.5k · Python

Active A+

DeepCode is an open agentic coding platform supporting Paper2Code, Text2Web, and Text2Backend, leveraging agent technology for automated software development workflows.

codingpythonllm +2

· Multi-agent coding system with separate Planning, Implementation, and Review agents for structured development
· Paper2Code capability to convert research papers into working code implementations
· Text2Web and Text2Backend for generating web frontends and backend services from natural language descriptions

View full details → View on GitHub

Qwen Code

27.6k · TypeScript

Active A

An open-source AI coding agent that lives in your terminal, built by Qwen team with support for code generation, editing, debugging and multi-file operations.

codingllmtypescript +2

· Multi-protocol multi-provider — supports OpenAI, Anthropic, Gemini-compatible APIs with Alibaba Cloud, OpenRouter and more
· Qwen3-Coder co-evolution — both framework and Qwen3-Coder model are open-source and ship together
· Built-in Skills and SubAgents — rich built-in tools for full agentic workflow with Claude Code-like experience

View full details → View on GitHub

Refact

3.5k · Rust

Stale B

Refact is a Rust-based AI coding agent that handles engineering tasks end-to-end, integrating into developer workflows with code completion, chat, agent actions, and self-hosted deployment.

codingrustagent +2

· Rust-based AI coding agent handling engineering tasks end-to-end with high performance
· IDE integration with code completion, chat, and autonomous agent actions
· Self-hosted deployment option for privacy-conscious teams and air-gapped environments