Security self-assessment tool for AI agents. Run this against your own configuration to get a structured threat model report with RED/AMBER/GREEN ratings across six security domains — decision boundaries, audit trail, credential scoping, plane separation, economic accountability, and memory safety.

AI代理的安全自评估工具。针对您的配置运行此工具，即可获得涵盖六大安全领域的结构化威胁建模报告，包括决策边界、审计追踪、凭证范围控制、平面隔离、经济责任和内存安全，并附有红黄绿三色评分。