Replies: 2 comments
|
从 rc.2 当前的所有权设计看,比较安全的实现不是让子 Agent 直接占用主 Session 的提问 UI,而是增加一个由父 Agent 持有的 relay。 现在 建议的流程是:
如果是 foreground/background one-shot child,它没有后续对话句柄,父 Agent 得到问题后只能询问用户,再启动一个带答案的 replacement child。因此工作流预计会发生澄清时,应在开始前选择 continuable provider。 还需要限制 clarification 轮数、问题大小、过期时间和总预算,并明确“澄清回答不是权限授权”,不能借普通问答扩大 sandbox、凭据或网络权限。 我整理了完整英文模式、状态机和 12 个验收场景: https://sandbaseai.github.io/deepseek-harness-handbook/subagent-question-relay.html Disclosure: I maintain the SandBase community handbook. |
0 replies
|
业务上遇到类似的问题,顶帖看看后续 |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
当前定义了workflow开启一系列子agent执行流程后,发现子agent信息不足无法询问用户问题补充信息后继续执行。
是否有可能支持子agent运行期间,如果发现问题,回到主session中询问用户,拿到答案后转给子agent继续?
All reactions