Unverified50% confidenceFactExact time
Agent 能力经历了从基于 ReAct 范式的单步推理到以 LangGraph、AutoGen、Claude Code 为代表的多智能体协作框架的转变
1
Sources
50%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
Agent Skills完全指南：原理、结构与实战应用拆解
bilibili学习的谁不疯7/5/2026
Related Claims
Unverified多轮 Agent 反思在 AutoGen、CrewAI 等多 Agent 框架中已有成熟实践80% similarUnverified典型的agentic模式包括ReAct（推理-行动交替）、Reflection（输出自评估与修正）、Multi-Agent Collaboration（多智能体分工协作）78% similarUnverifiedAgent具备规划、工具调用、记忆与自主迭代的能力，区别于传统大语言模型的单次问答模式78% similarUnverifiedAgent Skill 模式在 LangChain、AutoGPT 等 AI 框架中已广泛应用78% similarUnverifiedAgent能力的底层依赖于大语言模型的推理能力（Chain-of-Thought）以及ReAct（Reasoning + Acting）框架77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/183553API
curl https://kongchang.com/api/v1/knowledge/claims/183553MCP
get_claim(id=183553)