Unverified50% confidenceFactExact time
现代Agent通常以大语言模型作为决策大脑，结合ReAct、Chain-of-Thought等推理范式
1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
AI Agent开发入门：从零构建智能应用的完整学习路径
bilibili小全栈7/10/2026
Related Claims
Unverified在大模型时代，Agent架构中LLM充当大脑负责推理决策，工具充当四肢负责与外部世界交互，记忆模块保存对话历史和中间状态81% 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78% similarUnverified现代Agent系统能够在单次会话中维持长达数万token的上下文窗口，并通过思维链推理机制对复杂实验方案进行多步骤逻辑分解75% similarUnverifiedThe technical foundation for the shift to the Agent paradigm includes advances in large language models' reasoning capabilities, long-context processing, and Function Calling74% similarPartially Verified主流Agent框架如LangChain、AutoGen、CrewAI均围绕感知-决策-执行循环构建74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/484914API
curl https://kongchang.com/api/v1/knowledge/claims/484914MCP
get_claim(id=484914)