Unverified50% confidenceFactExact time
System、user、assistant三个角色分别承担不同职责，共同构成多轮对话的上下文结构
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
程序员转AI工程师：你已完成80%，只差这六项核心能力
Related Claims
Unverified在现代LLM对话架构中消息被划分为System、User和Assistant三个角色层级，其中System Prompt处于最高权重层级67% similarUnverifiedIn LLM-based multi-agent systems, there are three main methods of inter-Agent communication: shared memory, message passing, and blackboard systems.63% similarUnverifiedThe system is designed with two core roles: student user and administrator.62% similarUnverifiedReasons for the sharp increase in token consumption in multi-agent architecture include: each sub-agent call must carry the full system prompt and context, inter-agent communication generates a large number of intermediate message tokens, and the coordinator agent must process the outputs of all sub-agents.61% similarUnverifiedCodex plugin sharing adopts a three-level sharing model: individual, designated members, and the entire workspace.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/515080API
curl https://kongchang.com/api/v1/knowledge/claims/515080MCP
get_claim(id=515080)