← 返回 Skills 市场
mars82311111

Feishu Agent Team

作者 mars82311111 · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
87
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install feishu-agent-team
功能描述
在 OpenClaw 中构建多 Agent 团队协作系统。Coordinator(调度中心)接收 Feishu 群聊 @mention,自动路由任务到专业 Agent,支持自定义 Agent 角色和数量。适用于:(1) 需要 AI 团队处理多类型任务 (2) 希望用单一入口调度专家 Agent (3) 构建自动化...
使用说明 (SKILL.md)

Feishu Agent Team - 多 Agent 团队协作系统

概念

将多个 AI Agent 组织成"团队":

  • 1 个 Coordinator - 调度中心,接收所有消息并分发
  • N 个 Specialist - 专家 Agent,各司其职
                    ┌─→ Specialist-A (商分)
                    │
User @Coordinator ──┼─→ Specialist-B (市场)
    │               │
    │               └─→ Specialist-C (开发)
    │
    └──→ [State] ←──┘

核心功能

  • 智能路由:关键词匹配 → 自动分发到对应专家
  • Feishu 集成:群 @mention 触发调度,支持多群配置
  • 跨 Agent 通信:MQ 消息队列异步协作
  • 状态持久化:LangGraph Checkpoint 支持断点恢复
  • 可扩展:添加任意数量的专家 Agent

安装

# 进入 OpenClaw workspace
cd ~/.openclaw/workspace

# 克隆项目
git clone \x3Crepo-url> feishu-agent-team

# 安装依赖
pip install langgraph langchain-openai feishu-oapi

# 初始化
python team.py init

快速配置

1. 定义团队角色

编辑 config/team.yaml:

coordinator:
  name: "Coordinator"        # 调度中心名称
  mention_name: "coordinator" # 群里的 @ 名称

specialists:
  - name: "Analyst"          # 专家1
    specialty: "商业分析"
    keywords: ["分析", "市场", "投资", "趋势", "商分"]
    
  - name: "Developer"        # 专家2
    specialty: "技术开发"
    keywords: ["代码", "开发", "bug", "技术", "架构"]
    
  - name: "Marketer"          # 专家3
    specialty: "市场推广"
    keywords: ["推广", "增长", "内容", "运营"]

2. 配置 OpenClaw 绑定

{
  "bindings": [
    {
      "agentId": "coordinator",
      "match": { "channel": "feishu", "accountId": "coordinator" }
    },
    {
      "agentId": "analyst",
      "match": { "channel": "feishu", "accountId": "analyst" }
    },
    {
      "agentId": "developer",
      "match": { "channel": "feishu", "accountId": "developer" }
    },
    {
      "agentId": "marketer",
      "match": { "channel": "feishu", "accountId": "marketer" }
    }
  ]
}

3. 配置 Feishu App

{
  "channels": {
    "feishu": {
      "accounts": {
        "coordinator": {
          "appId": "cli_xxx",
          "appSecret": "xxx",
          "groupPolicy": "open",
          "groups": {
            "YOUR_GROUP_ID": {
              "enabled": true,
              "requireMention": true,
              "groupSessionScope": "group_sender"
            }
          }
        },
        "analyst": { "appId": "cli_yyy", "appSecret": "yyy" },
        "developer": { "appId": "cli_zzz", "appSecret": "zzz" },
        "marketer": { "appId": "cli_aaa", "appSecret": "aaa" }
      }
    }
  }
}

4. 启动 Agent 监听

# 每个 Specialist Agent 运行一个监听进程
python listeners/specialist_listener.py --agent analyst &
python listeners/specialist_listener.py --agent developer &
python listeners/specialist_listener.py --agent marketer &

使用方式

群里 @ Coordinator 发任务

@Coordinator 分析一下当前AI市场的投资趋势

Coordinator 自动识别 → 分发给 Analyst → Analyst 在群里回复。

支持多类型任务

@Coordinator 开发一个用户登录模块
@Coordinator 推广我们的新产品
@Coordinator 分析竞品情况

CLI 方式

# 测试任务路由
python team.py route "我想推广新产品"  # → Marketer

# 运行任务
python team.py run "分析竞争对手"

# 查看团队状态
python team.py info

Python API

from src.api import TeamGraphAPI

api = TeamGraphAPI()

# 执行任务
result = api.process("市场调研报告", team_config="config/team.yaml")
print(result["specialist"])  # 自动路由到对应专家

自定义指南

添加新的专家 Agent

  1. config/team.yaml 添加:
specialists:
  - name: "Designer"
    specialty: "UI设计"
    keywords: ["设计", "界面", "UI", "UX", "图标", "海报"]
  1. 创建监听器:
# listeners/designer_listener.py
from src.listeners.base import SpecialistListener

class DesignerListener(SpecialistListener):
    name = "designer"
    specialty = "UI设计"

if __name__ == "__main__":
    DesignerListener().start()
  1. 添加 OpenClaw 绑定和 Feishu 账号

  2. 启动监听: python listeners/designer_listener.py &

修改路由逻辑

编辑 src/graph.py 中的路由函数:

def route_to_specialist(state: TeamState) -> str:
    task = state["task"].lower()
    
    # 自定义路由规则
    if any(kw in task for kw in ["分析", "研究", "报告"]):
        return "analyst"
    elif any(kw in task for kw in ["开发", "代码", "bug"]):
        return "developer"
    # ... 更多规则
    
    return "coordinator"  # 默认保留给调度者处理

多群支持

配置多个群组 ID:

{
  "channels": {
    "feishu": {
      "accounts": {
        "coordinator": {
          "groups": {
            "GROUP_ID_1": { "enabled": true, "requireMention": true },
            "GROUP_ID_2": { "enabled": true, "requireMention": true },
            "GROUP_ID_3": { "enabled": true, "requireMention": true }
          }
        }
      }
    }
  }
}

工作原理

1. 用户在 Feishu 群 @Coordinator 发送任务
                          ↓
2. Coordinator 接收消息,提取任务内容
                          ↓
3. 任务路由函数分析关键词,匹配专家
                          ↓
4. 通过 MQ 消息队列分发任务给对应专家
                          ↓
5. Specialist Agent 接收任务,在群里回复

技术栈

  • 状态机: LangGraph
  • 持久化: SQLite Checkpoint
  • 消息队列: 轻量级 MQ 实现
  • 飞书集成: feishu-oapi

项目结构

feishu-agent-team/
├── SKILL.md
├── README.md
├── config/
│   └── team.yaml           # 团队配置(角色、关键词)
├── src/
│   ├── __init__.py
│   ├── api.py              # Python API
│   ├── graph.py            # LangGraph 图定义
│   ├── nodes.py            # 节点逻辑
│   ├── state.py            # 状态定义
│   ├── routing.py          # 路由逻辑
│   ├── feishu_group.py     # Feishu 群操作
│   └── persistence/         # Checkpoint 持久化
├── listeners/
│   ├── base.py              # 监听器基类
│   ├── analyst_listener.py
│   ├── developer_listener.py
│   └── marketer_listener.py
├── team.py                 # CLI 入口
└── tests/

常见问题

Q: 需要多少个 Feishu App? A: 最少 2 个(1 个 Coordinator + 1 个 Specialist),最多 N+1 个。

Q: 如何添加更多专家? A: 见"自定义指南 - 添加新的专家 Agent"。

Q: 任务失败怎么办? A: Coordinator 会保留任务状态,可通过 python team.py retry \x3Ctask_id> 重试。

许可证

MIT License - 免费使用

支持

遇到问题提交 Issue

安全使用建议
This skill appears to implement a Feishu-based coordinator + specialist routing system, but there are a few things to check before installing: - Verify provenance: SKILL.md asks you to git clone a <repo-url> but registry lacks a homepage/source. Ask the publisher for the repository URL and verify its origin (GitHub org, commit history, author). - Secrets handling: Examples place Feishu appId/appSecret in config files (plaintext). Prefer storing credentials in a secure secret store or environment variables, and avoid committing config files with secrets to VCS. - Undeclared credentials: The skill implies use of langchain-openai (usually requires an OpenAI API key) but the registry declares no required env vars; clarify which environment variables or secret storage the skill expects and whether it supports reading credentials from environment variables instead of config files. - Incomplete implementations: listeners contain TODOs for MQ and Feishu API calls — the shipped code may not actually connect to Feishu or a message queue. Expect to implement or configure those integrations yourself. - Dependency safety: pip-installing third-party packages (langgraph, feishu-oapi, langchain-openai) is normal but verify package maintainers and pin versions. Consider installing in an isolated virtual environment. If you plan to proceed: run the code in an isolated/test environment, inspect the full source (especially any network calls and logging), and do not place production secrets into config files until you confirm how credentials are used and stored. If the publisher cannot provide a verifiable source or explain the credential handling, treat the skill as higher risk.
功能分析
Type: OpenClaw Skill Name: feishu-agent-team Version: 1.0.0 The skill bundle provides a legitimate framework for a multi-agent coordination system integrated with Feishu (Lark). The code consists of standard Python logic for task routing based on keywords, configuration management via YAML, and placeholder listener classes for handling messages, with no evidence of data exfiltration, malicious execution, or harmful prompt injection.
能力评估
Purpose & Capability
Name/description, routing code (src/routing.py), config/team.yaml and CLI (team.py) all align with a Feishu multi-agent coordinator/specialist design. However, SKILL.md (metadata) requires python3 and pip packages (langgraph, langchain-openai, feishu-oapi) while the registry requirements list no binaries/envs — minor mismatch but explainable. Overall functionality matches stated purpose.
Instruction Scope
Runtime instructions tell the user to clone a repo, pip install third-party packages, and place Feishu appId/appSecret values into config (examples show plaintext appSecret values). The code contains TODO placeholders for MQ reads and Feishu API calls (listeners/BaseListener has get_mq_messages and send_to_group marked TODO), so runtime behavior is incomplete and will require additional integration. Instructions do not ask to read unrelated system files, but they do direct storing of secrets in project config files and assume external services (Feishu, LangGraph/OpenAI) which are not declared in requires.env.
Install Mechanism
There is no formal install spec in the registry (instruction-only), but SKILL.md recommends pip installing packages from PyPI (langgraph, langchain-openai, feishu-oapi). Using pip is typical but carries moderate risk: packages should be verified (authors, versions). No downloads from arbitrary URLs in the install spec, but the README/SKILL.md asks to git clone an unspecified <repo-url> which requires caution if the source is unknown.
Credentials
The skill requires Feishu app credentials (appId/appSecret) in the example config and recommends langchain-openai which implies an OpenAI API key, yet the registry declares no required environment variables or primary credential. That mismatch is concerning: secrets are expected but not declared, and examples suggest storing app secrets in plaintext config files. This is disproportionate to what the registry metadata advertises and increases risk of accidental secret exposure.
Persistence & Privilege
The skill is not always-enabled and does not request system-wide privileges or claim to modify other skills. It uses local files (config/team.yaml, SQLite checkpoint per docs) and runs listener processes; this is consistent with its function and not overly privileged.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install feishu-agent-team
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /feishu-agent-team 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release
元数据
Slug feishu-agent-team
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Feishu Agent Team 是什么?

在 OpenClaw 中构建多 Agent 团队协作系统。Coordinator(调度中心)接收 Feishu 群聊 @mention,自动路由任务到专业 Agent,支持自定义 Agent 角色和数量。适用于:(1) 需要 AI 团队处理多类型任务 (2) 希望用单一入口调度专家 Agent (3) 构建自动化... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 87 次。

如何安装 Feishu Agent Team?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install feishu-agent-team」即可一键安装,无需额外配置。

Feishu Agent Team 是免费的吗?

是的,Feishu Agent Team 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Feishu Agent Team 支持哪些平台?

Feishu Agent Team 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Feishu Agent Team?

由 mars82311111(@mars82311111)开发并维护,当前版本 v1.0.0。

💬 留言讨论