In recent years, the rapid development of IT systems in fields such as cloud computing and 5G networks has significantly increased system complexity, driving the transition from automated to intelligent IT operations. While large language models have shown remarkable success in natural language processing tasks, their application in IT operations knowledge Q&A remains underexplored, particularly in multi-agent systems. This paper proposes a multi-agent collaboration framework to enhance the accuracy of LLMs in IT operations-related Q&A without domain-specific fine-tuning. Evaluated on a test set of nearly 2,000 multiple-choice questions, the framework demonstrates an accuracy improvement of up to 7.68% compared to direct LLM inference. These results highlight the potential of multi-agent collaboration in enhancing the practical value of LLMs for IT operations intelligence, offering a promising direction for future research and application.