跳到正文
RadarAI· X:Addy Osmani (@addyosmani)·· 4 小时前AI 评分52

Addy Osmani:何时该把子智能体收回主循环

AI 导读

Addy Osmani 提出判断子智能体的经验法则:只有当任务可并行、需要全新上下文或单个智能体确实无法选对工具时才保留子智能体,否则应将其收回主循环。他指出系统提示词臃肿往往是问题根源,可用 Skills 和工具搜索解决,而不必把上下文拆分到多个智能体。做法是把子智能体的提示词改写成 skill,跑完评测后保留效果更好的方案。

正文 · 原文

Addy Osmani· @addyosmani · X·2026-10-27 14:30· 27 分钟前AI 评分49

AI 导读

Addy Osmani 提出判断子智能体的经验法则:只有当任务可并行、需要全新上下文或单个智能体确实无法选对工具时才保留子智能体,否则应将其收回主循环。他指出系统提示词臃肿往往是问题根源,可用 Skills 和工具搜索解决,而不必把上下文拆分到多个智能体。做法是把子智能体的提示词改写成 skill,跑完评测后保留效果更好的方案。

正文 · 原文

Before adding another subagent ask what problem it's solving. A bloated system prompt is often the issue.

Skills and tool search can fix that without splitting context across agents. My rule of thumb is to keep a subagent when the work is parallel, needs fresh context or one agent really can't pick the right tools.

Otherwise, fold it into the main loop, turn its prompt into a skill, run your evals and keep whatever wins.

Good write-up:
https://claude.com/resources/articles/building-agents-when-to-fold-subagents-into-your-main-agent-loop

来源:Addy Osmani · x.com

来源:RadarAI · aihot.news