✨AutoPatch

✨AutoPatch: LLM Based Low Level Memory Safety Bug Patching in C with Fuzzing, Address Sanitization, and Code Property Graphs

A Google and NSF funded project of the University of Illinois Chicago in collaboration with Texas A&M University

About The Project

✨AutoPatch leverages the power of Large Language Models (LLMs), such as LLaMa, OpenAI, Gemini, DeepSeek and MistralAI, to automatically generate patches for memory safety bugs found in C code. Given vulnerable code snippet, AutoPatch-LLM aims to produce a correct patch to remediate the issue.

This tool can be used in a wider agentic workflow and designed to assist developers in the often complex and time-consuming task of vulnerability patching.

Key Features

Getting Started

To get started with ✨AutoPatch, refer to the repository's README for detailed instructions:

Please consult the project's README for specific commands, options, and prerequisites.

Source Code Repository

The complete source code, detailed documentation, and issue tracker can be found on GitHub. Contributions and feedback are welcome!

View on GitHub

Research & Citation

If you use ✨AutoPatch in your research, please cite the relevant publication(s). Check the project's README or repository for the correct citation information.