The development and integration of Language Model Agents (LLMs) into real-world applications have garnered significant attention due to their remarkable natural language processing capabilities. However, the rise of prompt injection as a potential threat to LLM integrity has raised concerns about their security and reliability. The threat of prompt injection Prompt injection poses a significant