Unverified60% confidenceFactTime unknown
多智能体框架指把一个复杂研究任务拆解给多个各司其职的 AI 代理完成，而非依赖单一模型一次性生成答案
2
Sources
60%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Claude Code Deep Research：多智能体协作的引用溯源研究框架
githubfranklin798
Related Entities
Related Claims
Unverified多智能体系统的概念最早源自分布式 AI 领域，指多个自主智能体通过交互、协作或竞争完成单一智能体无法独立完成的任务77% similarVerifiedAI Agents can receive a single instruction, autonomously plan the task workflow, invoke tools, and accomplish goals without requiring step-by-step human intervention.77% similarUnverified一次AI代理会话类似于一条分布式调用链，涉及多文件读写、多工具调用、多轮LLM推理，复杂度超出人脑直接追踪的能力边界76% similarUnverifiedAI辅助编程正从单次交互向多Agent任务编排演进，Agent编排成为提升开发生产力的新瓶颈76% similarVerifiedAI Agent具备工具调用（Tool Use）和规划推理（Planning & Reasoning）两大核心能力，能将复杂目标自动分解为可执行的子任务序列76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/917524API
curl https://kongchang.com/api/v1/knowledge/claims/917524MCP
get_claim(id=917524)