← 返回 Skills 市场
cchenwei

Feishu Multi Agent Factory

作者 Cchenwei · GitHub ↗ · v1.0.5 · MIT-0
cross-platform ✓ 安全检测通过
189
总下载
1
收藏
0
当前安装
6
版本数
在 OpenClaw 中安装
/install feishu-multi-agent-factory
功能描述
批量配置 OpenClaw 多 agent + 飞书 channel。当用户说「添加 agent」「新建 agent」「配置飞书」「批量创建 agent」「添加新机器人」时触发。通过对话引导收集信息,一键写入 openclaw.json 并初始化工作区。
使用说明 (SKILL.md)

Agent Factory — 多 agent 批量配置向导

通过对话快速创建新 agent,自动完成所有配置:

  • 写入 openclaw.json(agents / feishu accounts / bindings)
  • 创建 workspace 目录 + IDENTITY.md 等基础文件
  • 创建 agentDir + defaults.json + auth-profiles.json

核心脚本

~/.openclaw/workspace/skills/feishu-multi-agent-factory/scripts/setup_agents.py

对话示例

用户说:

帮我创建2个 agent:
1. id: painter,图片处理专家🌞,飞书 AppId: \x3CPAINTER_APP_ID>,Secret: \x3CPAINTER_SECRET>
2. id: take,行业资讯抓手👋,飞书 AppId: \x3CTAKE_APP_ID>,Secret: \x3CTAKE_SECRET>

你的回应(先 dry-run 确认):

好的,我来预览一下即将创建的配置:

▶ 🌞 图片处理专家 (id: painter)
  • workspace → ~/.openclaw/workspace-painter/
  • agentDir  → ~/.openclaw/agents/painter/agent/
  • 飞书 account 'painter' 已配置

▶ 👋 行业资讯抓手 (id: take)
  • workspace → ~/.openclaw/workspace-take/
  • agentDir  → ~/.openclaw/agents/take/agent/
  • 飞书 account 'take' 已配置

确认执行吗?

用户确认后执行,回应:

✅ 创建: 2 个
新 agents: painter, take

请运行 `openclaw configure` 为各 agent 配置 API key。
已重启 gateway,两个飞书机器人上线后即可收发消息。

安全说明

  • App Secret 不回显:收到 Secret 后不在对话中重复展示,确认预览时只显示 AppId
  • Secret 脱敏输出:脚本日志只显示 Secret 前 4 位(如 R1T8****
  • 凭据仅写入本地:所有凭据只写入 ~/.openclaw/openclaw.json,不上传、不外传
  • id 严格校验:只允许小写字母、数字、连字符,防止路径穿越攻击

对话流程

第一步:收集信息

用户说想添加新 agent 时,按以下格式引导(可批量,一次多个):

我需要以下信息来创建 agent:

1. **Agent ID**(英文小写,如 `coder`、`hr-bot`)
2. **中文名称**(如「代码专家」)
3. **Emoji**(如 💻)
4. **飞书 App ID**(格式:cli_xxxxxxxx)
5. **飞书 App Secret**
6. **简介**(可选)

可以一次提供多个,直接列出来就行。

第二步:确认预览(dry-run)

收到信息后,先 dry-run 预览,确认无误再执行:

python3 ~/.openclaw/workspace/skills/feishu-multi-agent-factory/scripts/setup_agents.py \
  --dry-run \
  --config '\x3CJSON>'

第三步:执行创建

用户确认后,正式执行并重启 gateway:

python3 ~/.openclaw/workspace/skills/feishu-multi-agent-factory/scripts/setup_agents.py \
  --config '\x3CJSON>' \
  --restart

JSON 格式

{
  "agents": [
    {
      "id": "coder",
      "name": "代码专家",
      "emoji": "💻",
      "description": "负责写代码和 code review",
      "feishu_app_id": "\x3CYOUR_APP_ID>",
      "feishu_app_secret": "\x3CYOUR_APP_SECRET>"
    },
    {
      "id": "hr-bot",
      "name": "HR 助手",
      "emoji": "👔",
      "feishu_app_id": "\x3CYOUR_APP_ID>",
      "feishu_app_secret": "\x3CYOUR_APP_SECRET>"
    }
  ]
}

其他命令

查看当前所有 agents

python3 ~/.openclaw/workspace/skills/feishu-multi-agent-factory/scripts/setup_agents.py --list

删除某个 agent(仅移除配置,不删目录)

python3 ~/.openclaw/workspace/skills/feishu-multi-agent-factory/scripts/setup_agents.py --remove \x3Cid>

自动完成的事项清单

每个新 agent 执行完成后会自动:

步骤 内容
✅ workspace 目录 ~/.openclaw/workspace-{id}/
✅ IDENTITY.md 包含名称、emoji、描述
✅ SOUL / AGENTS / TOOLS.md 基础工作区文件
✅ agentDir ~/.openclaw/agents/{id}/agent/
✅ defaults.json 继承全局模型配置
✅ auth-profiles.json 从 main agent 复制
✅ agents.list 写入 openclaw.json
✅ feishu accounts 写入 openclaw.json
✅ bindings agent ↔ feishu account
✅ agentToAgent.allow 加入协作白名单

飞书 App 说明

每个 agent 对应一个独立的飞书机器人应用(需在飞书开放平台预先创建)。

需要的权限(在飞书开放平台开通):

  • im:message — 收发消息
  • im:message.group_at_msg — 群消息 @ 机器人

连接模式默认为 websocket(长连接),无需公网 IP。

常见问题

Q: 飞书凭据在哪里找? 飞书开放平台 → 我的应用 → 选择应用 → 凭证与基础信息 → App ID / App Secret

Q: 创建后 agent 没反应? 确认已运行 openclaw gateway restart,并在飞书开放平台确认机器人已上线。

Q: 想给某个 agent 单独设置模型? 创建完成后编辑 ~/.openclaw/agents/{id}/agent/defaults.json,修改 model.primary

安全使用建议
This skill appears to do what it says: it helps create agent directories and adds Feishu account entries to your local ~/.openclaw/openclaw.json. Before running it, consider: 1) Secrets are stored in openclaw.json in plaintext — ensure that file's filesystem permissions are tight (chmod 600) or use an encrypted secret store if you need stronger protection. 2) The script may restart the OpenClaw gateway when run with --restart; confirm what command it executes (the file's tail was truncated) to ensure no unexpected shell execution. 3) There's a small mismatch: SKILL.md mentions copying auth-profiles but the script currently defers that to `openclaw configure`; expect to run that manually. 4) Review the full script contents on your machine before invoking it with real credentials. If you need higher assurance, ask the author for the complete script or run it in a controlled environment first.
功能分析
Type: OpenClaw Skill Name: feishu-multi-agent-factory Version: 1.0.5 The feishu-multi-agent-factory skill bundle is a legitimate utility designed to automate the batch configuration of OpenClaw agents and their associated Feishu (Lark) channel credentials. The core logic in `scripts/setup_agents.py` performs file system operations (creating workspace directories) and modifies the local `openclaw.json` configuration file, which is consistent with its stated purpose. The script includes proactive security measures such as regex-based validation of agent IDs to prevent path traversal attacks and masking of sensitive application secrets in logs. No evidence of data exfiltration, unauthorized remote execution, or malicious intent was found.
能力评估
Purpose & Capability
Name/description match what the files do: the script creates workspace/agent dirs, updates ~/.openclaw/openclaw.json with agent entries and Feishu accounts, and adds bindings and allow lists. No unrelated credentials, binaries, or external services are requested.
Instruction Scope
SKILL.md and the script keep scope to local configuration and workspace initialization. The instructions require running the included Python script with JSON input; the script writes to ~/.openclaw/openclaw.json, creates files/dirs, and prints masked secrets. It does not appear to collect or transmit data externally. Note: SKILL.md states auth-profiles.json will be 'copied from main', but the script contains a placeholder (get_main_auth_profiles returns None) and instead asks the user to run `openclaw configure` — a minor mismatch.
Install Mechanism
No install spec; the skill is instruction + a Python script bundled with the skill. Nothing is downloaded at install time and no external packages are pulled automatically.
Credentials
No environment variables or external credentials are requested by the skill itself. It prompts the user to provide Feishu AppId/AppSecret (appropriate). However, those secrets are written into ~/.openclaw/openclaw.json in plaintext (the script masks them in printed notes/logs but stores the full appSecret in the config). That is expected for a local config tool but is a secrecy risk if the file is not protected.
Persistence & Privilege
always is false and the skill does not request system-wide privileges beyond writing into the user's ~/.openclaw directory and (optionally) restarting the OpenClaw gateway. Modifying the user's OpenClaw config is consistent with the tool's purpose. The file tail was truncated, so the exact restart implementation couldn't be fully verified.
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-multi-agent-factory
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /feishu-multi-agent-factory 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
v1.0.5
- 增加安全说明板块,强调 App Secret 不回显、日志脱敏、凭据仅写入本地、Agent ID 严格校验等措施 - 优化隐私保护流程:预览和脚本日志不再显示完整 Secret,防止敏感信息泄露 - 其余使用说明与配置流程保持不变,核心命令无更改
v1.0.4
- Replaced sample Feishu AppId and Secret values in documentation with generic placeholders (e.g., `<PAINTER_APP_ID>` instead of concrete IDs). - Clarified documentation samples to avoid exposing sensitive information and promote safer sharing. - No code or behavior changes; documentation update only.
v1.0.3
- 路径变更:脚本路径统一为 feishu-multi-agent-factory(原 agent-factory),相关命令行和文件路径已同步更新。 - 添加了完整的对话示例与标准回应文本,便于实际交互时参考。 - 说明文档结构优化,前置对话示例和简化常见操作的指引。 - 所有命令示例、核心脚本路径、干运行/执行步骤均更新至新版 skill 的命名与目录。 - 其余功能逻辑未变化,主要为文档和用法指引优化。
v1.0.2
- Updated core script scripts/setup_agents.py. - No user-facing documentation or feature changes.
v1.0.1
- 新增 skill,支持批量配置 OpenClaw 多 agent + 飞书 channel - 通过对话引导收集 agent 信息,一键生成配置文件和初始化工作区 - 支持 dry-run 预览、正式创建及一键重启 gateway - 提供 agents 查看、删除等辅助命令 - 自动生成基本目录、配置文件并绑定飞书账号 - 文档详细说明各创建步骤、所需信息及常见问题
v1.0.0
- Initial release of feishu-multi-agent-factory. - Enables batch configuration of OpenClaw agents and Feishu channels via guided conversation. - Automates agent setup: writes openclaw.json, creates workspace directories, and initializes required files. - Supports easy preview (dry-run), confirmation, and execution of agent creation. - Provides commands to list or remove agents and detailed instructions for necessary Feishu settings.
元数据
Slug feishu-multi-agent-factory
版本 1.0.5
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 6
常见问题

Feishu Multi Agent Factory 是什么?

批量配置 OpenClaw 多 agent + 飞书 channel。当用户说「添加 agent」「新建 agent」「配置飞书」「批量创建 agent」「添加新机器人」时触发。通过对话引导收集信息,一键写入 openclaw.json 并初始化工作区。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 189 次。

如何安装 Feishu Multi Agent Factory?

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

Feishu Multi Agent Factory 是免费的吗?

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

Feishu Multi Agent Factory 支持哪些平台?

Feishu Multi Agent Factory 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 Multi Agent Factory?

由 Cchenwei(@cchenwei)开发并维护,当前版本 v1.0.5。

💬 留言讨论