HAFix is an approach that leverages individual historical heuristics associated with bugs and aggregates the results of these heuristics (HAFix-Agg) to enhance LLMs’ bug-fixing capabilities

HAFix: History-Augmented Large Language Models for Bug Fixing

Yu Shi, Abdul Ali Bangash, Emad Fallahzadeh, Bram Adams, Ahmed E. Hassan

lab on Maintenance, Construction and Intelligence of Software (MCIS)

Software Analysis and Intelligence Lab (SAIL)

School of Computing, Queen's University

If you find this repository useful, please consider giving us a star ⭐ and citation:

@article{shi2025hafix,
  title={HAFix: History-Augmented Large Language Models for Bug Fixing},
  author={Shi, Yu and Bangash, Abdul Ali and Fallahzadeh, Emad and Adams, Bram and Hassan, Ahmed E},
  journal={arXiv preprint arXiv:2501.09135},
  year={2025}
}

Project Structure

• dataset: collecting the data of baseline and historical heuristics
• model_inference: conduct prompt construction and LLM inference
• evaluation: evaluating the model-generated fixed code and calculating the pass@k
• analysis: generate the figure and box plot for 3 RQs