- 官方设计顺序主轴：Anthropic
Building Effective AI Agents、OpenAIAgents SDK Overview/Safety in building agents/Context management/Sessions/Handoffs - 状态与工作流主轴：LangGraph
Workflows and agents/Persistence - 业务建模主轴：Microsoft
Domain Analysis to Model Microservices/Designing a microservice domain model、Martin FowlerUbiquitous Language/Bounded Context - 安全与权限主轴：Anthropic
Securely deploying AI agents - 评测与运行经济学主轴：OpenAI
Evaluating model performance/Latency optimization - 本章落点主轴：生产级 Agent 的正确设计顺序是目标、对象、动作、状态、控制流、治理、评测，最后才是模型与成本优化
附录 G：分章资料清单