Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@lark-mcp-server show my upcoming calendar events"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
🇨🇳 The first Chinese MCP Server for Lark/Feishu — let any AI client operate Feishu through MCP protocol. 🇨🇳 首个飞书 MCP Server — 让任何 AI 客户端通过标准 MCP 协议操作飞书。
# via pip
pip install lark-mcp-server
# or via uvx (no install needed)
uvx lark-mcp-serverYou need lark-cli installed and authenticated:
# Install lark-cli & login
lark-cli auth loginAdd to your claude_desktop_config.json:
{
"mcpServers": {
"lark": {
"command": "uvx",
"args": ["lark-mcp-server"]
}
}
}Now Claude can send/receive Feishu messages, list chats, and more! 🤖
Related MCP server: lark-cli-mcp
Tool | Description | 说明 |
| Send text/markdown to a chat | 发送消息到聊天 |
| Reply to a message | 回复消息 |
| List group chats | 列出群聊列表 |
| Get chat member list | 获取群成员 |
| View chat history | 查看聊天历史 |
| Send interactive card | 发送交互卡片 |
Tool | Description | 说明 |
| Get user information | 获取用户信息 |
Tool | Description | 说明 |
| View calendar agenda | 查看日程安排 |
# Your AI agent can now:
# - Monitor and reply to messages automatically
# - Send batch notifications to group chats
# - Check meeting schedules and notify attendees# Connect Feishu with other tools via MCP:
# - Read Feishu messages → process with AI → write to database
# - Sync calendar events to your task manager
# - Automate HR onboarding workflows# Feishu + GitHub + Slack = automated workflow
# All connected through standard MCP protocollark-mcp-server/
├── pyproject.toml # Python package config
├── src/lark_mcp_server/
│ ├── server.py # MCP server entry, tool registration
│ └── client.py # LarkClient — lark-cli subprocess wrapper
└── tests/
└── test_client.py # Integration testslark-cli — Lark/Feishu CLI tool
YuanZhiTech — Our Agent Garden organization
MIT © 2026 Agent Garden
Built by AI Agents, for AI Agents. 🤖 由 AI Agent 构建，为 AI Agent 服务。
查看飞书日程安排
Inputs: end, start, calendar_id. Returns: result
获取群聊成员列表
Inputs: chat_id (required). Returns: result
获取飞书用户信息，不传 user_id 查自己
Inputs: user_id. Returns: result
列出飞书群聊列表
Inputs: page_size, page_token. Returns: result
查看聊天历史消息
Inputs: chat_id (required), page_size. Returns: result
回复飞书消息
Inputs: content (required), message_id (required), msg_type, is_thread. Returns: result
发送交互卡片到飞书聊天
Inputs: chat_id (required), card_json (required). Returns: result
发送消息到飞书聊天
Inputs: chat_id (required), content (required), msg_type. Returns: result
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 8 tools
Each tool targets a distinct resource and action: sending, replying, listing chats, getting members, viewing messages, sending cards, user info, and agenda. While send_message and send_card both send to a chat, their descriptions clearly distinguish text from interactive cards, leaving no ambiguity.
All tool names follow the same pattern: lark_verb_noun (e.g., lark_send_message, lark_list_chats). The verbs are consistent (send, reply, list, get) and all use snake_case, making the naming uniform and predictable.
With 8 tools, the server is well-scoped for a collaboration platform covering messaging, chat management, user info, and agenda reading. Each tool earns its place without being overwhelming or sparse.
The server covers core messaging workflows (send, reply, list history) and read operations for chats and users, but lacks write operations for chats (create/update/delete), message editing/deletion, and calendar management beyond reading. These gaps could cause agents to hit dead ends when trying to manage resources.
MCP server connecting AI agents to 100+ apps (Gmail, Slack, Notion, GitHub) via one-click OAuth.
MCP server enabling AI agents to manage Bitrix24 features via standardized protocol
Connect any AI agent to 1,000+ apps and 27,000+ actions through one remote MCP server (OAuth).
Zero-setup MCP gateway securely connecting AI to your tools with authentication and workflows
