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