When the Manual Lies: A Realistic Benchmark to Evaluate MCP Poisoning Attacks for LLM Agents 文章

ArXiv CS.AI2026-05-26NEWSen作者: Shi Liu, Xuehai Tang, Xikang Yang, Liang Lin, Biyu Zhou, Wenjie Xiao, Wantao Liu
查看原文 →
关系图谱

When the Manual Lies: A Realistic Benchmark to Evaluate MCP Poisoning Attacks for LLM Agents · 相关技术

相关技术

LLMginODEPLAlanguage model研究法辛烷值execevalUCTToolsStraight-Through EstimatorStanSandboxSPASEMReferring expression comprehension (REC)RTLOWLMaliModel Context ProtocolLMMLLM agentsHISForEvaluation Driven DevelopmentEffort Metric AttentionCamouflaged object detectionCISARGANNAgentMCPTable