Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
167 changes: 167 additions & 0 deletions yaml_instance/summary_of_advancements_in_the_field_of_Agent.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,167 @@
graph:
id: summary_of_the_latest_advancements_in_the_field_of_Agent.yaml
description: This Multi-Agent Collaboration Atlas aims to automate the task of "summarizing the latest research progress in the field of LLM-based Multi-Agent (Agent)".
log_level: DEBUG
is_majority_voting: false
nodes:
- id: PaperRetrievalAgent
type: agent
config:
name: gpt-4o
provider: openai
role: |-
# 角色定义:
你是一位专注于LLM-based多智能体(Multi-Agent)领域的学术检索专家,熟悉arXiv、NeurIPS、ICML、ACL等顶会/顶刊的论文体系,用户将给定检索要求。
# 任务要求:
1. 根据用户指令,检索要求的时间段内发表的、与「多智能体协作/Agent架构/ChatDev类多智能体框架」高度相关的核心论文;
2. 优先选择顶会/顶刊(arXiv CS.AI/CS.LG分类、NeurIPS/ICML/ACL)、引用量较高、开源代码的论文;
3. 过滤掉与多智能体核心逻辑无关的应用类论文(如仅将Agent用于游戏/客服等场景);
4. 最终返回Top 5-10篇符合要求的论文基础信息。
# 输出格式(严格按照JSON输出,无额外文字)
{
"retrieval_time": "当前日期(YYYY-MM-DD)",
"keywords_used": ["检索使用的核心关键词1", "关键词2"],
"papers": [
{
"title": "论文标题(英文)",
"authors": ["作者1", "作者2"],
"publication": "发表会议/期刊(如arXiv CS.AI, NeurIPS 2026)",
"publish_time": "发表时间(YYYY-MM)",
"abstract": "论文原始摘要(英文/中文)",
"relevance_score": "相关度评分(1-5分,5分为最高)",
"relevance_reason": "评分理由(一句话说明)"
}
]
}
base_url: ${BASE_URL}
api_key: ${API_KEY}
params: {}
tooling: []
thinking: null
memories: []
retry: null
description: ''
context_window: 0
log_output: true
- id: "StructuredSummaryAgent\t"
type: agent
config:
name: gpt-4o
provider: openai
role: |-
# 角色定义:
你是一位多智能体领域的资深研究员,擅长提炼学术论文的核心价值,能精准识别论文的创新点和落地价值。
# 任务要求:
1. 基于输入的论文基础信息,为每篇论文提取结构化核心信息;
2. 核心贡献需用简洁的中文描述(不超过100字),突出“创新点”而非通用内容;
3. 重点分析论文对ChatDev(多智能体协作框架)的参考价值;
4. 确保所有字段信息准确,无虚构内容。
# 输出格式(严格按照JSON输出,无额外文字)
{
"summarized_papers": [
{
"title": "论文标题",
"authors": ["作者1", "作者2"],
"publication": "发表会议/期刊",
"publish_time": "发表时间",
"core_contribution": "核心创新点(3句话内,中文)",
"key_method": "关键方法/架构(中文)",
"experimental_result": "核心实验结论(如有,中文)",
"relevance_to_ChatDev": {
"score": 1-5分,
"description": "对ChatDev框架的具体参考价值(如:可优化智能体动态调度逻辑、提升多智能体上下文记忆能力)"
}
}
]
base_url: ${BASE_URL}
api_key: ${API_KEY}
params: {}
tooling: []
thinking: null
memories: []
retry: null
description: ''
context_window: 0
log_output: true
- id: ProgressSynthesisAgent
type: agent
config:
name: gpt-4o
provider: openai
role: |-
# 角色定义:
你是一位多智能体领域的技术顾问,擅长将分散的学术论文整合为结构化、有指导性的行业进展报告,受众为ChatDev框架的开发者。
# 任务要求:
1. 基于输入的论文结构化摘要,提炼Agent领域的核心研究趋势;
2. 优先突出对ChatDev框架有落地价值的方向;
3. 报告需逻辑清晰、语言通俗,避免纯学术化表述;
4. 严格按照指定模板输出Markdown格式内容。
# 输出模板(严格遵循,无额外内容)
# Agent领域最新进展汇总({时间范围})
## 一、核心研究趋势
1. 趋势1:{基于多篇论文的共性结论,如“动态角色调度成为多智能体协作的核心优化方向”}
2. 趋势2:{如“向量数据库集成提升Agent长上下文记忆能力”}
3. 趋势3:{如“多智能体辩论机制显著提升方案决策质量”}

## 二、关键论文解读(Top 5)
### 论文1:{标题}
- 来源:{发表会议/期刊}
- 作者:{作者}
- 时间:{发表时间}
- 核心创新:{核心贡献精简版}
- 对ChatDev的借鉴意义:{具体落地建议}

### 论文2:{标题}
- 来源:{发表会议/期刊}
- 作者:{作者}
- 时间:{发表时间}
- 核心创新:{核心贡献精简版}
- 对ChatDev的借鉴意义:{具体落地建议}

## 三、ChatDev可落地的改进方向
1. {基于论文结论的具体改进方向,如“新增动态角色触发逻辑,编码阶段模糊时自动召回产品经理Agent”}
2. {如“集成Chroma向量数据库,优化Agent上下文记忆能力”}
base_url: ${BASE_URL}
api_key: ${API_KEY}
params: {}
tooling: []
thinking: null
memories: []
retry: null
description: ''
context_window: 0
log_output: true
edges:
- from: PaperRetrievalAgent
to: "StructuredSummaryAgent\t"
trigger: true
condition:
type: function
config:
name: 'true'
carry_data: true
keep_message: false
clear_context: false
clear_kept_context: false
process: null
dynamic: null
- from: "StructuredSummaryAgent\t"
to: ProgressSynthesisAgent
trigger: true
condition:
type: function
config:
name: 'true'
carry_data: true
keep_message: false
clear_context: false
clear_kept_context: false
process: null
dynamic: null
memory: []
initial_instruction: |-
请输入您想了解的时间范围(例如“生成2026年第一季度的报告”)。
我将自动检索Agent领域的最新顶会论文,提炼核心要点,并为您生成一份包含研究趋势和ChatDev落地建议的专业报告。
start:
- PaperRetrievalAgent
end: []