Feishu Multi Agent Messaging

飞书群聊多Agent消息通讯。解决多Bot协作时的消息流转问题，支持群聊@通知、私聊消息、用户ID映射。

Security Scan

Documentation-only skill that gives instructions for configuring multiple Feishu bots and user-ID mapping; the files and instructions are coherent with the stated purpose but you must protect the bot credentials and mapping data described in the docs.

Purpose & Capability

The skill's name/description (multi‑agent Feishu messaging, user ID mapping, accountId routing) matches the content: documentation and examples describe configuring multiple Feishu accounts, maintaining open_id mappings, and calling the platform message tool with accountId. There are no unrelated credentials, binaries, or install steps requested.

Instruction Scope

SKILL.md and the docs only instruct configuration and use of the message tool (specifying accountId, using <at> tags, maintaining mapping files, restart/check gateway logs). They do reference storing mappings in SESSION-STATE.md or openclaw.json and running system diagnostics (journalctl, log files). These are reasonable for this purpose but are operationally broad (touching config and logs) — the agent instructions do not ask to read arbitrary user files or exfiltrate data, but they do imply storing/reading sensitive configuration and mapping files.

Install Mechanism

Instruction-only skill with no install spec and no code files. This minimizes on‑disk execution risk; nothing is downloaded or written by the skill package itself.

Credentials

The skill does not declare environment variables, but examples show appId/appSecret values placed directly in openclaw.json and recommend storing mapping files. That is proportionate to the feature (bots need credentials) but is sensitive: appSecrets and user open_id mappings are secrets/PII and should be stored securely (not committed to VCS).

Persistence & Privilege

Flags show normal defaults (always: false, agent-invocable true, model invocation allowed). The skill does not request persistent system privileges or modify other skills' configs. Autonomy is platform-default and not combined with other concerning requests.

Assessment

This skill is documentation-only and appears coherent for its stated goal of coordinating multiple Feishu bots and mapping per-bot user IDs. Before installing/using it: (1) Do NOT commit appId/appSecret or mapping files to public repos — keep them in a secrets store or use environment-secret mechanisms; (2) Treat the user-ID mapping as sensitive (it contains user identifiers/PII) and restrict access to it; (3) Verify your OpenClaw/message tool actually supports the described accountId routing and behavior in your environment; (4) Consider using Feishu union_id (if you can obtain the permission) to avoid manual mapping when possible; (5) Review gateway logs and restart steps (journalctl, ~/.openclaw logs) only if you have appropriate system access — the docs reference system-level commands but the skill itself does not perform them automatically; (6) If you need stronger guarantees, request the author to provide an implementation that uses a secure secrets mechanism (instead of plain openclaw.json examples). Overall the content is consistent with its purpose and shows no signs of hidden exfiltration, but protect credentials and mapping files carefully.

飞书群聊多Agent消息通讯

解决飞书平台上多 Agent 协作时的消息流转问题

核心功能

• 用户ID映射 - 解决每个Bot看到的用户open_id不同的问题
• 群聊协作流程 - 6步标准协作流程，确保消息正确流转
• @用户通知 - 群聊中正确@指定用户
• 私聊消息 - 子Agent以自己的身份发送私聊

快速开始

1. 配置多个飞书Bot

在 openclaw.json 中配置多个飞书账号：

{
  "channels": {
    "feishu": {
      "accounts": {
        "default": {
          "appId": "cli_xxx",
          "appSecret": "xxx"
        },
        "coder": {
          "appId": "cli_yyy",
          "appSecret": "yyy"
        },
        "reviewer": {
          "appId": "cli_zzz",
          "appSecret": "zzz"
        }
      }
    }
  }
}

2. 维护用户ID映射表

每个Bot看到的用户open_id不同，需要维护映射表：

Bot	accountId	用户的 open_id
主Bot	default	ou_xxx
码农Bot	coder	ou_yyy
审核Bot	reviewer	ou_zzz

3. 发送消息时指定accountId

# 码农Bot发送私聊
message action=send channel=feishu accountId="coder" target="ou_yyy" message="消息内容"

# 审核Bot发送群聊
message action=send channel=feishu accountId="reviewer" target="oc_xxx" message="消息内容"

核心发现

问题：子Agent无法以自己的身份发消息

根因：

1. 每个飞书Bot看到的用户open_id不同
2. message tool不支持union_id
3. 私聊发送必须指定accountId参数

解决方案：

1. 维护用户ID映射表
2. 发送时使用该Bot看到的用户open_id
3. 用户必须先和Bot建立会话

群聊协作流程

步骤1：用户群聊 @主Bot 下发命令
步骤2：主Bot群聊 @码农 @审核 下发任务
步骤3：码农、审核群聊回复「收到」
步骤4：码农执行完成，群聊通知审核
步骤5：审核决定打回/通过
步骤6：主Bot @用户 汇报完成

核心指标：所有节点对应的人要把状态发到群里

@用户格式

<at user_id="ou_xxx">用户名</at>

注意事项

1. 必须指定accountId - 不指定时会用主Bot的身份发送
2. 必须用正确的open_id - 每个Bot看到的用户open_id不同
3. 用户必须先和Bot建立会话 - 用户需要先主动给Bot发过消息
4. 群聊协作核心指标 - 所有节点对应的人要把状态发到群里

文件结构

feishu-multi-agent-messaging/
├── SKILL.md               # 本文件
├── README.md              # 详细说明
├── LICENSE                # MIT许可证
├── docs/
│   ├── architecture.md    # 架构设计
│   ├── id-mapping.md      # ID映射详解
│   ├── workflow.md        # 协作流程
│   └── troubleshooting.md # 问题排查
└── examples/
    ├── openclaw.json      # 配置示例
    └── message-samples.md # 消息示例