首页龙虾技能列表 › Feishu Multi-Bot

Feishu Multi-Bot

v1.0.1

Configure multiple Feishu/Lark bot identities within a single OpenClaw instance — each Agent gets its own Feishu App, name, avatar, and message routing. Use...

1· 96·0 当前·0 累计
by @simonlin1212 (simonlin)·MIT-0
下载技能包
License
MIT-0
最后更新
2026/4/3
安全扫描
VirusTotal
无害
查看报告
OpenClaw
安全
high confidence
The skill and its files are coherent with the stated purpose (generating OpenClaw Feishu multi-bot configuration); nothing in the package appears malicious, but the helper prints app secrets to stdout which is a sensitive-but-intentional behavior you should handle carefully.
评估建议
This package appears to do what it says: generate JSON blocks and documentation for wiring multiple Feishu apps to OpenClaw agents. Before using it: (1) Avoid passing secrets on shared shells—the script prints appSecret values to stdout and those arguments may be stored in shell history; prefer running it in a private terminal or modify the script to read secrets from stdin or a protected file. (2) Review and validate the generated JSON before merging into ~/.openclaw/openclaw.json (the docs alr...
详细分析 ▾
用途与能力
Name/description match the included materials: docs explain a three-block configuration model for OpenClaw + Feishu and the script generates the channels/bindings/agents JSON blocks. Required artifacts (openclaw.json edits, Feishu app IDs/secrets, gateway restarts) are legitimate for this purpose.
指令范围
SKILL.md instructions stay within the claimed scope (creating apps, generating config blocks, validating JSON, restarting gateway, debugging routing). One noteworthy behavior: the provided setup script accepts appSecret values on the command line and prints them into JSON on stdout—this is necessary to produce config but it exposes secrets to terminal logs/history. Troubleshooting steps also suggest pkill/openclaw logs/grep which are expected for admin tasks.
安装机制
No install spec; this is instruction-only plus a small shell helper. No downloads, no archived extracts, and the script does not write files or fetch remote resources. Low install risk.
凭证需求
The skill does not request environment variables or other external credentials in metadata. Practically, it requires you to supply Feishu AppID/AppSecret pairs (via script args) and to edit openclaw.json—which is proportional. The main caution: secrets are provided on the command line and echoed; consider the sensitivity of exposing appSecret values in shell history/logs.
持久化与权限
always:false and default model invocation are fine. The skill does not request persistent system-wide privileges, does not modify other skills, and does not create background services. It only emits configuration blocks and workspace mkdir commands (not executed by the script itself).
安全有层次,运行前请审查代码。

License

MIT-0

可自由使用、修改和再分发,无需署名。

运行时依赖

无特殊依赖

版本

latestv1.0.12026/4/2

Update display name to 'Feishu Multi-Bot'

● 无害

安装命令 点击复制

官方npx clawhub@latest install openclaw-feishu-multi-bot
镜像加速npx clawhub@latest install openclaw-feishu-multi-bot --registry https://cn.clawhub-mirror.com

技能文档

One OpenClaw instance, multiple Feishu bot identities. Each Agent appears as an independent bot in Feishu — own name, own avatar, own group memberships. Users see separate assistants; behind the scenes, a single Gateway dispatches everything.

Architecture in 30 Seconds

OpenClaw Gateway (single instance)
  ├── Agent: orchestrator ←→ Feishu App 1 (总调度 bot)
  ├── Agent: content-writer ←→ Feishu App 2 (写作 bot)
  ├── Agent: code-expert ←→ Feishu App 3 (开发 bot)
  └── Agent: analyst ←→ Feishu App 4 (分析 bot)

Three configuration blocks must align:

BlockLocation in openclaw.jsonPurpose
channels.feishu.accountsFeishu credentials per botMaps accountId → appId/appSecret
bindingsMessage routing rulesMaps accountId → agentId
agents.listAgent definitionsMaps agentId → workspace/model/tools
The accountId is the key that ties them together. Get it wrong in any one place and routing breaks silently.

Quick Start

1. Create Feishu Apps

One enterprise app per Agent at open.feishu.cn/app. Enable "Bot" capability. Record each app's AppID and AppSecret.

2. Run the setup helper

scripts/setup-feishu-bots.sh orchestrator:cli_xxx:secret1 writer:cli_yyy:secret2 coder:cli_zzz:secret3

Generates the three JSON blocks ready to paste into openclaw.json.

3. Register agents and restart

Add the generated config to openclaw.json, then:

openclaw doctor && openclaw gateway restart

4. Test each bot

Send a message to each Feishu bot independently. Verify each responds with the correct Agent identity.

Reference Files

FileRead when...
references/architecture.mdUnderstanding the three-block config model, accountId mechanism, and channel layer design
references/build-guide.mdSetting up from scratch — Feishu developer console through gateway restart, step by step
references/routing-deep-dive.mdDebugging routing issues — accountId consistency checks, binding rules, group-based isolation
references/troubleshooting.mdFixing specific problems — gateway won't start, bot not responding, wrong agent, spawn conflicts

Critical Lessons (Save Hours)

1. accountId must match in 3 places: channels.feishu.accounts.{key}, bindings[].match.accountId, and the agent field inside the account config. One typo = silent routing failure.

2. Binding type must be "route": Using "delivery" or any other value causes gateway startup failure with no helpful error message.

3. Feishu apps must be published: Draft-state apps cannot receive messages. This is the #1 "bot not responding" cause.

4. allowAgents must be complete: If your orchestrator spawns sub-agents, every spawnable agent ID must be in allowAgents. New agent added but not listed = spawn permission error.

5. agentToAgent must stay OFF: Enabling agentToAgent.enabled: true breaks all sub-agent spawning (known bug #5813). Keep it false.

See references/troubleshooting.md for the full diagnostic flowchart.

Scripts

ScriptUsage
scripts/setup-feishu-bots.sh./setup-feishu-bots.sh ... — Generates channels, bindings, and agents.list JSON blocks
数据来源:ClawHub ↗ · 中文优化:龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制

免费技能或插件可能存在安全风险,如需更匹配、更安全的方案,建议联系付费定制

了解定制服务