Token导航 LogoToken导航

Agent Skill · 开发

feishu-chat(飞书聊天)

feishu-chat 用于辅助前端页面、组件、样式和交互逻辑开发,适合在 OpenClaw 中需要维护前端项目、生成组件或检查界面实现时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

feishu-chat 是开发类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为需要联网;许可证为 MIT-0,第三方扫描已通过,按安装量在开发类的 21,746 个 Skill 里排第 1,400。

总安装 30,647周安装 1,303GitHub Stars 公开资料未说明下载 10,737核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.1.1

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:feishu-chat(飞书聊天)
来源仓库:https://github.com/heimo-he/feishu-chat
安装命令:
openclaw skills install feishu-chat
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install feishu-chat

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

提供飞书群聊消息发送指南,包括原始与卡片模式。支持 @提及 格式和群组成员管理功能。feishu-chat 属于开发类 Skill,可作为该场景下的辅助能力补充。帮助 Agent 正确构造邮件和消息内容。需根据实际项目调整消息结构和权限设置。适用宿主包括 OpenClaw,接入前应确认版本、权限和运行环境要求。

安装前请核实维护状态和 API 版本兼容性。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install feishu-chat;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

feishu-chat 怎么安装?

在终端执行 openclaw skills install feishu-chat。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

feishu-chat 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

feishu-chat 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

feishu-chat 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

feishu-chat 用的人多吗?

公开数据是累计安装 30,647 次,近一周安装 1,303 次,下载 10,737 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

feishu-chat 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
feishu-chat

飞书聊天 · 本页

30,647—需要联网MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
feishu-chat
description
Feishu (Lark) group chat messaging guide for OpenClaw. Includes Raw/Card message modes, @ mention formatting, and group member management. Use when sending messages in Feishu groups, mentioning users/bots, or formatting messages with Markdown.

Feishu Group Chat Guide

Quick Start: Configure renderMode

⚠️ IMPORTANT: Set renderMode explicitly for stable message formatting

# Recommended configuration
channels:
  feishu:
    renderMode: "card"  # Always use card format

Why? The default "auto" mode causes unpredictable format switching (raw vs card), creating poor user experience.

Configuration:

openclaw config set channels.feishu.renderMode "card"
openclaw gateway restart

Verify:

openclaw config get channels.feishu.renderMode

Prerequisites

  1. In a Feishu group - OpenClaw connected to Feishu
  2. Configure renderMode - Set to "card" for consistent formatting
  3. Know member IDs:

- Human: open_id (ou_xxx format) - Bot: App ID (cli_xxx format)


0. Message Format Stability

The Problem: Raw vs Card Mixing

Without proper renderMode configuration, messages can inconsistently appear as:

  • Raw text: Plain markdown source, no formatting
  • Card format: Rendered markdown with syntax highlighting, tables, links

The Solution: Explicit renderMode

ModeBehaviorUse Case
autoDetects content, uses card for code/tables❌ Avoid - unpredictable
rawAlways plain textSimple text-only responses
cardAlways interactive card✅ Recommended - consistent

Recommended: renderMode: "card" for all production use.


1. Message Sending

⚠️ CRITICAL: Only Use ONE Method

DO NOT use both direct reply and message tool for the same content! This will send two messages.

Choose ONE:

  • Plain text → message tool only
  • Markdown → direct reply only

Two Methods

MethodToolRender ModeUse Case
message toolmessageRaw (plain text)Plain text, @ mentions
Direct replySession replyCard (with Markdown)Markdown formatted messages

Raw Mode (Plain Text)

When to use: Plain text messages, @ mentions

How to send:

message({
  action: "send",
  channel: "feishu",
  target: "oc_xxx",
  message: "Plain text content"
})

@ Mention format:

message({
  action: "send",
  channel: "feishu",
  target: "oc_xxx",
  message: "<at user_id=\"ou_xxx\">nickname</at> Hello!"
})

Raw Mode Limitations:

  • ❌ No Markdown rendering (bold, italic, code blocks won't render)
  • ❌ Tables won't render
  • ❌ Links won't render
  • ✅ @ mentions work correctly

Card Mode (Markdown)

When to use: Formatted messages with code blocks, tables, bold/italic text

How to send: Direct reply in session with Markdown elements.

Card Mode Triggers (any of these):

  • Code blocks `` ``
  • Tables | table |
  • Bold **bold**
  • Italic *italic*
  • Strikethrough ~~text~~
  • Links [link](url)
  • Headings # heading
  • Lists - item or 1. item

@ Mention format:

<at id=ou_xxx></at> This is Card mode @ mention

**Markdown content** renders correctly

Supported in Card Mode:

StyleSyntaxStatus
Bold**bold**✅ Supported
Italic*italic*✅ Supported
Strikethrough~~text~~✅ Supported
Color<font color='red'>text</font>✅ Supported
Links[link](https://xxx)✅ Supported
Headings# heading✅ Supported
Lists- item or 1. item✅ Supported
Code blocks`` code ``✅ Supported
Tables`table`✅ Supported

Not Supported in Card Mode:

StyleSyntaxStatus
Blockquote> quote❌ Not supported
Inline code` code `❌ Unstable
Horizontal rule---❌ Not supported
Complex nestingMultiple levels❌ Unstable

Best Practice for Card Mode:

  • Use **bold** for emphasis
  • Use *italic* for light emphasis
  • Use numbered lists 1. or bullet lists -
  • Avoid ---, >, and inline code
  • Keep formatting simple

Avoid Auto Mode

Problem: Auto mode can incorrectly choose between Raw and Card.

Best Practice:

  • Plain text → Use message tool (Raw mode)
  • Markdown content → Direct reply (Card mode)

Don't:

  • Use message tool for Markdown (won't render)
  • Use complex structures in Card mode (may trigger Raw incorrectly)

2. Group Member Management

⚠️ Important: Bots Only See @ Mentioned Messages

Bots can only see messages that @ mention them! If a group member sends a message without @ mentioning the bot, the bot won't receive it.

This means:

  • To identify a human member's open_id, they must @ mention the bot and send a message
  • The bot cannot see past messages where it wasn't mentioned

How to Get Member IDs

Human open_id (must @ mention bot):

  1. Ask the human to @ mention the bot and send any message
  2. When the message arrives, the system shows:

[Feishu oc_xxx:ou_xxx timestamp] nickname: message content

  1. The ou_xxx is the sender's open_id

Bot App ID (requires user to provide):

  1. Go to Feishu Developer Console: https://open.feishu.cn/app
  2. Click on the bot you want to view
  3. In "Application Credentials" (应用凭证), copy the App ID (cli_xxx format)
  4. @ mention the bot and send the App ID

Maintain Member List

Store member info in memory file:

## Group Members (oc_xxx)
### Humans
- nickname1: `ou_xxx`
- nickname2: `ou_xxx`

### AI Bots
- bot1: `cli_xxx` (must @ to receive messages)

3. @ Mention Tips

@ Format Difference (Critical!)

Mode@ FormatExample
Raw (message tool)<at user_id="ID">nickname</at><at user_id="ou_xxx">kk</at>
Card (direct reply)<at id=ID></at><at id=ou_xxx></at>

@ Different Target Types

@ TargetID TypeRaw FormatCard Format
Humanopen_id (ou_xxx)<at user_id="ou_xxx">nickname</at><at id=ou_xxx></at>
BotApp ID (cli_xxx)<at user_id="cli_xxx">botname</at><at id=cli_xxx></at>
Everyone"all"<at user_id="all">everyone</at><at id=all></at>

@ Bot vs @ Human

Important difference:

  • Bots: MUST be @ mentioned to receive messages! Without @, bots won't get notified
  • Humans: Can see all group messages, @ not required

⚠️ Feishu Limitation: Bot-to-Bot Messages

  • A bot CAN send messages and @ mention another bot
  • But bots can only receive messages from human accounts
  • Bots cannot receive messages from other bots
  • Therefore, @ mentioning another bot will NOT notify that bot

Practical implication:

  • If you need to communicate with another bot, ask a human to send the message instead
  • Bot-to-bot communication is not possible in Feishu groups

Example - @ Bot:

// Raw mode
message({
  action: "send",
  channel: "feishu",
  target: "oc_xxx",
  message: "<at user_id=\"cli_xxx\">botname</at> Please reply!"
})
// Card mode
<at id=cli_xxx></at> Please reply!

**Markdown content**

@ Everyone

Note: Requires group permission (Group Settings > Group Management > Who can @everyone)

// Raw mode
message({
  action: "send",
  channel: "feishu",
  target: "oc_xxx",
  message: "<at user_id=\"all\">everyone</at> Important announcement!"
})
// Card mode
<at id=all></at> Important announcement!

Common @ Mistakes

Raw mode errors:

  • @nickname - Plain text, no notification
  • @{nickname} - Plain text
  • <at id="ou_xxx"></at> - id attribute doesn't work in Raw mode

Card mode errors:

  • <at user_id="ou_xxx">nickname</at> - user_id attribute doesn't work in Card mode

4. Quick Reference

Message Decision Tree

Need to send message
    │
    ├─ Plain text only?
    │   └─ YES → message tool (Raw mode)
    │
    └─ Need Markdown?
        └─ YES → Direct reply (Card mode)

@ Mention Decision Tree

Need to @ mention
    │
    ├─ Using message tool?
    │   └─ YES → <at user_id="ID">nickname</at>
    │
    └─ Direct reply (Card mode)?
        └─ YES → <at id=ID></at>

ID Format Reference

TypeFormatExample
Human open_idou_xxxou_example123abc
Bot App IDcli_xxxcli_example456def
Group chat_idoc_xxxoc_example789ghi

5. Important Notes

  1. Card and Raw mode use different @ formats - Wrong format = @ fails
  2. Bots MUST be @ mentioned to receive messages - Humans don't need @
  3. Bot-to-bot communication is NOT possible - Bots can only receive messages from humans, not other bots
  4. message tool only sends Raw mode - No Markdown rendering
  5. @everyone requires group permission
  6. Avoid unsupported styles in Card mode (blockquote, inline code, horizontal rule)

6. Configuration Reference

renderMode Settings

Location: OpenClaw config file or environment

Quick Setup:

# Set to card mode (recommended)
openclaw config set channels.feishu.renderMode "card"
openclaw gateway restart

Environment-specific configurations:

Development:

channels:
  feishu:
    renderMode: "raw"  # Easier debugging
    dmPolicy: "open"   # Easier testing

Production:

channels:
  feishu:
    renderMode: "card"      # Consistent formatting
    dmPolicy: "pairing"     # Security
    requireMention: true    # Avoid spam

Troubleshooting

Messages appear as raw markdown:

  1. Check: openclaw config get channels.feishu.renderMode
  2. Set: openclaw config set channels.feishu.renderMode "card"
  3. Restart: openclaw gateway restart

Format switches between raw and card:

  • Cause: Using renderMode: "auto"
  • Fix: Change to "card" or "raw" explicitly

Proactive messages not using card:

  • Expected: Proactive messages (via message tool) always use plain text
  • Workaround: Have user send a message first, then reply

Source

  • Tested on: OpenClaw Feishu integration
  • Test date: 2026-02-27 ~ 2026-02-28
  • Source files:

- /home/admin/.openclaw/extensions/feishu/src/bot.ts - /home/admin/.openclaw/extensions/feishu/src/reply-dispatcher.ts - /home/admin/.openclaw/extensions/feishu/src/mention.ts - /home/admin/.openclaw/extensions/feishu/src/send.ts

平台分布

OpenClaw

89.28%
按下载量换算9,586

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

未展示

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills