Vigil: LLM Prompt Injection Detection and Defense Toolkit

DeadBits

Abstract

Open-source scanner for detecting prompt injections using vector similarity, YARA rules, text classifiers, and canary tokens.

Categories

Tags

Framework Mappings

Cite This Resource

@article{llmsec202400056,
  title = {Vigil: LLM Prompt Injection Detection and Defense Toolkit},
  author = {DeadBits},
  year = {2024},
  journal = {GitHub},
  url = {https://github.com/deadbits/vigil-llm},
}