作者位次 #4/9ACL 2026Findings· Findings of ACL
MCP -Guard: A Multi-Stage Defense-in-Depth Framework for Securing Model Context Protocol in Agentic AI
Wenpeng Xing · Zhonghao Qi · Yupeng Qin · Yilin Li · Caini Chang · Jiahui Yu · Changting Lin · Zhenzhen Xie · Meng Han
摘要
While Large Language Models (LLMs) have achieved remarkable performance, they remain vulnerable to jailbreak. The integration of Large Language Models (LLMs) with external tools via protocols such as the Model Context Protocol (MCP) introduces critical security vulnerabilities, including prompt i...