问HN：你们是如何在代理工作流中管理提示的？
我正在研究人工智能代理，并且有6个代理正在进行对话。在我构建架构的过程中，我注意到管理提示之间的相互依赖关系非常麻烦。<p>有没有我还不知道的最佳实践？
查看原文
I am working on AI Agents and have 6 agents that are engaging in a conversation. As I am working on the architecture, I noticed that managing the interdependencies between the prompts is annoying.<p>Are there best practices I don't know yet?