龙明
龙明

Agent人机协同交互12篇

2026-08-31 03:47:55

什么时候该把人拉进循环?这组 12 篇讲审批门设计、置信度路由、human-in-the-loop 的交互模式,以及自主性与可控性的平衡实践。核心结论很一致:人不是兜底垃圾桶,要在正确的时机给正确的信息,否则既拖效率也留不住注意力。适合设计审核流、担心 agent 闯祸、或要接合规要求的团队。篇目精短,一次能读完。
aws-samples/sample-human-in-th
aws-samples/sample-human-in-th
AWS人机
Dify Human-in-the-Loop Node
Dify Human-in-the-Loop Node
Dify人机
rotorstar/hitl-protocol
rotorstar/hitl-protocol
GitHub人机
LangGraph — Human-in-the-Loop
LangGraph — Human-in-the-Loop
LangChainLangGraph
AutoGen — Human-in-the-Loop
AutoGen — Human-in-the-Loop
微软AutoGen
Claude Agent SDK — Handle Appr
Claude Agent SDK — Handle Appr
Anthropic审批流
HiL-Bench: Do Agents Know When
HiL-Bench: Do Agents Know When
arXiv论文人机
Human Judgment in the Agent Im
Human Judgment in the Agent Im
LangChain人工
Humans and Agents in Software
Humans and Agents in Software
MartinFowler人机
Measuring AI Agent Autonomy in
Measuring AI Agent Autonomy in
Anthropic自主性
aiming-lab/AutoResearchClaw
aiming-lab/AutoResearchClaw
GitHub科研Agent
SmileLikeYe/agent-chief
SmileLikeYe/agent-chief
GitHub任务

评论

chat_bubble_outline

暂无评论,快来发表第一条评论吧