Skip to content

aaronclong/prompt-manners

Repository files navigation

Prompt Manners

Overview

This library attempt to mitigate the prompt injection vulnerability in LLMs prompts for LLM-Integrated Applications.

Inspiration

It is inspired and based on the Open-Prompt-Injection project, and the research their paper "Prompt Injection Attacks and Defenses in LLM-Integrated Applications".

@misc{liu2023promptattackdefense,
      title={Prompt Injection Attacks and Defenses in LLM-Integrated Applications}, 
      author={Yupei Liu and Yuqi Jia and Runpeng Geng and Jinyuan Jia and Neil Zhenqiang Gong},
      year={2023},
      eprint={2310.12815},
      archivePrefix={arXiv},
      primaryClass={cs.CR}
}