The study introduces WIRE, a new pipeline for diagnosing conflicts within natural-language prompt policies in large language model (LLM) agents, revealing that a significant portion of rule pairs lead to compliance violations. This research highlights the complexities of rule interactions in AI systems, emphasizing the need for better governance mechanisms in LLM applications.