[Findings of EMNLP'25] A comprehensive framework designed to safeguard system prompts for Large Language Models (LLMs) against various types of attacks. - View it on GitHub