Agentless

What it is

Agentless is an agentless approach to automatically solve software development problems. It follows a three phase process: localization, repair, and patch validation.

How it works

• Localization: localizes the fault to files, then to classes/functions, then to edit locations.
• Repair: samples multiple candidate patches in a simple diff format.
• Patch Validation: selects regression tests to run and generates additional reproduction tests, then re-ranks patches to select one to submit.

Getting started

git clone https://github.com/OpenAutoCoder/Agentless.git
cd Agentless

conda create -n agentless python=3.11 
conda activate agentless
pip install -r requirements.txt
export PYTHONPATH=$PYTHONPATH:$(pwd)

# for contribution, please install the pre-commit hook.
pre-commit install  # this allows a more standardized code style

export OPENAI_API_KEY={key_here}

Now you are ready to run Agentless on the problems in SWE-bench!

## Recent releases
- v1.5.0 Agentless v1.5.0 (2024-10-29)
- v0.1.0 Agentless v0.1.0 (2024-07-02)

## Traction
- Stars: 2092
- Forks: 236
- Open issues: 54

## Artifacts
You can download the complete artifacts of Agentless in our v1.5.0 release:
- 🐈‍⬛ agentless_swebench_lite: complete Agentless run on SWE-bench Lite
- 🐈‍⬛ agentless_swebench_verified: complete Agentless run on SWE-bench Verified
- 🐈‍⬛ swebench_repo_structure: preprocessed structure information for each SWE-Bench problem

## Citations

bibtex @article{agentless, author = {Xia, Chunqiu Steven and Deng, Yinlin and Dunn, Soren and Zhang, Lingming}, title = {Agentless: Demystifying LLM-based Software Engineering Agents}, year = {2024}, journal = {arXiv preprint}, }