A2A系统多跳委托授权传播的安全评估
原标题：Authorization Propagation in Agent2Agent Systems: An Experimental Security Evaluation of Multi-Hop Delegation
AI 摘要
该论文针对Agent2Agent（A2A）多跳委托工作流中的授权传播安全问题，在合成的四方医疗报销流程中评估了一种分层下游验证设计，将当前调用者上下文、传递性授权证据、逐跳Ed25519请求认证以及请求与授权工件的独立重放状态分离。在15种对抗条件下，14种被拒绝；但一项隔离的委托路径测试暴露出缺乏语义权限衰减，另一项授权中间人实验表明，持有共享授权证据HMAC密钥者可篡改所表示的权限并生成有效证据。在固定深度本地基准中，完整配置的中位安全处理延迟为5.11毫秒，序列化载荷为3.87 kB。
正文节选
Authorization Propagation in Agent2Agent Systems: An Experimental Security Evaluation of Multi-Hop Delegation Abstract Delegated Agent2Agent (A2A) workflows create a security problem when the agent sending a request is different from the principal from which the authority originated. This paper evaluates a layered downstream-verification design in a synthetic four-party healthcare reimbursement workflow. The prototype separates current-caller context, transitive authorization evidence, per-hop E