Unverified70% confidenceFactTime unknown
在Agent技能路由场景中，通过RAG向量检索可以将上下文占用从数万token压缩到几百token，同时保持极高的召回准确率
1
Sources
70%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
AI Agent Skill架构拆解：决策层+封装层+执行层实战指南
bilibili智能AI大模型
Related Entities
Related Claims
Unverified工作流文档配合提示词后追机制本质上是一种检索增强生成（RAG）的简化实现，用于在上下文重置后快速恢复Agent的任务认知70% similarUnverifiedBilibili UP host Peng Yu conducted a full-pipeline breakdown of RAG recall optimization in his large model interview series videos.67% similarUnverified父子分块策略用小块做检索、用大块喂给模型，在检索精度和上下文完整性之间取得平衡64% similarUnverifiedReAct的思考—行动—观察循环机制使Agent在处理多步骤复杂任务时表现出更高可靠性，显著减少幻觉和错误累积64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/20736API
curl https://kongchang.com/api/v1/knowledge/claims/20736MCP
get_claim(id=20736)