Determinants and Limits of LLM Security-Tool Orchestration: A Study with HexStrike-AI
最新研究揭示LLM安全工具编排的边界:模型能力与客户端架构如何影响效果
arXiv:2607.02873v1 Announce Type: cross Abstract: Large language model agents driving security tool suites over the Model Context Protocol are increas…