一句话：构建 75 对漏洞/修复 MCP 工具基准，用配对准确率 MPA 揭示 LLM 漏洞检测方法的真实语义辨识能力。
MCP（Model Context Protocol）服务器为 LLM 连接外部工具、数据源与执行环境，但其特权访问与桥接角色带来显著安全风险。已有大量基于 LLM 的检测方法，然而尚不清楚这些方法能否在语义上区分「漏洞版本」与「修复版本」的 MCP 服务器，还是仅仅识别出有风险的 API 调用。本文提出一个专门用于配对式 MCP 服务器漏洞检测的基准，包含 75 对由 Gemini-1.5-Pro 合成、并经人工交叉验证的漏洞/修复 MCP 工具，并采用配对准确率（Matched Pair Accuracy, MPA）指标评估检测方法。实验显示，Claude Haiku-4.5 取得最佳 MPA 0.853，而 GPT-5-mini 虽达到 0.842 的 F1，MPA 却仅为 0.653，说明传统指标会高估配对检测性能。该基准为系统化、可复现且面向 MCP 的安全评估提供了新视角。
Model Context Protocol (MCP) servers have become essential for connecting large language models (LLMs) with external tools, data sources, and execution environments. However, their privileged access and bridging role introduce significant security risks. Despite many LLM-based detection methods, it remains unclear whether they can semantically differentiate paired vulnerable and fixed MCP servers rather than just identifying risky API calls. In this work, we present a dedicated benchmark for pairing MCP server vulnerability detection, comprising 75 paired vulnerable and fixed MCP tools synthesized by Gemini-1.5-Pro and manually cross-validated. We adopt the Matched Pair Accuracy (MPA) metric to evaluate detection methods. Claude Haiku-4.5 achieves the best MPA of 0.853, while GPT-5-mini attains 0.842 F1 but only 0.653 MPA, showing that conventional metrics can overestimate paired detection performance. Our benchmark provides a new perspective for systematic, reproducible, and MCP-specific security evaluation.