Feishu/Lark OpenAPI MCP

English | 中文

Developer Documentation Retrieval MCP | Official Document

> ⚠️ Beta Version Notice: This tool is currently in Beta stage. Features and APIs may change, so please stay updated with version releases.

This is the Feishu/Lark official OpenAPI MCP (Model Context Protocol) tool designed to help users quickly connect to the Feishu/Lark platform and enable efficient collaboration between AI Agents and Feishu/Lark. The tool encapsulates Feishu/Lark Open Platform API interfaces as MCP tools, allowing AI assistants to directly call these interfaces and implement various automation scenarios such as document processing, conversation management, calendar scheduling, and more.

Features

• Complete Feishu/Lark API Toolkit: Encapsulates almost all Feishu/Lark API interfaces, including message management, group management, document operations, calendar events, Bitable, and other core functional areas.
• Dual Authentication Support:
  • Supports App Access Token authentication
  • Supports User Access Token authentication
• Flexible Communication Protocols:
  • Supports standard input/output stream (stdio) mode, suitable for integration with AI tools like Trae/Cursor/Claude
  • Supports Server-Sent Events (SSE) mode, providing HTTP-based interfaces

• Supports multiple configuration methods, adapting to different usage scenarios

Tool List

A complete list of all supported Feishu/Lark tools can be found in tools.md, where tools are categorized by project and version with descriptions.

Preparation

Creating a Feishu/Lark Application

Before using the lark-mcp tool, you need to create a Feishu/Lark application:

1. Visit the Feishu Open Platform or Lark Open Platform and log in

2. Click "Console" and create a new application

3. Obtain the App ID and App Secret, which will be used for API authentication

4. Add the necessary permissions for your application based on your usage scenario

5. If you need to call APIs as a user, set up OAuth 2.0 redirect URLs and obtain user access tokens

For detailed application creation and configuration guidelines, please refer to the Feishu Open Platform Documentation - Creating an Application or the Lark Open Platform Documentation.

Installing Node.js

Before using the lark-mcp tool, you need to install the Node.js environment.

Installing Node.js on macOS

1. Using Homebrew (Recommended):

brew install node

2. Using the Official Installer:

node -v
     npm -v

Installing Node.js on Windows

1. Using the Official Installer:

node -v
     npm -v

2. Using nvm-windows:

nvm install latest
     nvm use

Installation

Install the lark-mcp tool globally:

npm install -g @larksuiteoapi/lark-mcp

Usage Guide

Using with Trae/Cursor/Claude

To integrate Feishu/Lark functionality in AI tools like Trae，Cursor or Claude, add the following to your configuration file:

{
  "mcpServers": {
    "lark-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@larksuiteoapi/lark-mcp",
        "mcp",
        "-a",
        "",
        "-s",
        ""
      ]
    }
  }
}

To access APIs with user identity, you can add a user access token:

{
  "mcpServers": {
    "lark-mcp": {
     "command": "npx",
      "args": [
        "-y",
        "@larksuiteoapi/lark-mcp",
        "mcp",
        "-a",
        "",
        "-s",
        "",
        "-u",
        ""
      ]
    }
  }
}

Custom API Configuration

By default, the MCP service enables common APIs. To enable other tools or only specific APIs or presets, you can specify them using the `-t` parameter (separated by commas):

lark-mcp mcp -a  -s  -t im.v1.message.create,im.v1.message.list,im.v1.chat.create,preset.calendar.default

Preset Tool Collections in Detail

The following table details each API tool and its inclusion in different preset collections, helping you choose the appropriate preset for your needs:

> Note: In the table, "✓" indicates the tool is included in that preset. Using `-t preset.xxx` will only enable tools marked with "✓" in the corresponding column.

Advanced Configuration

Command Line Parameters

The `lark-mcp mcp` tool provides various command line parameters for flexible MCP service configuration:

Parameter Usage Examples

1. Basic Usage (using application identity):

lark-mcp mcp -a cli_xxxx -s yyyyy

2. Using User Identity:

lark-mcp mcp -a cli_xxxx -s yyyyy -u u-zzzz

> Note: User access tokens can be obtained through the Feishu Open Platform's authorization process or Lark Open Platform's authorization process, or you can use the API debugging console to obtain them. After using a user access token, API calls will be made with that user's identity.

3. Setting Specific Token Mode:

lark-mcp mcp -a cli_xxxx -s yyyyy --token-mode user_access_token

> Note: This option allows you to explicitly specify which token type to use when calling APIs. The `auto` mode (default) will be determined by the LLM when calling the API.

4. Specifying Lark or KA Domains:

# Lark international version
   lark-mcp mcp -a  -s  -d https://open.larksuite.com

   # Custom domain (KA domain)
   lark-mcp mcp -a  -s  -d https://open.your-ka-domain.com

5. Enabling Only Specific API Tools or Other API Tools:

lark-mcp mcp -a cli_xxxx -s yyyyy -t im.v1.chat.create,im.v1.message.create

> Note: The `-t` parameter supports the following preset tool collections:

> - `preset.light` - Lightweight tool set with fewer but commonly used tools, suitable for scenarios that require reduced token usage

> - `preset.default` - Default tool set containing all preset tools

> - `preset.im.default` - Instant messaging related tools, such as group management, message sending, etc.

> - `preset.base.default` - Base related tools, such as table creation, record management, etc.

> - `preset.base.batch` - Base batch operation tools, including batch create and update record functions

> - `preset.doc.default` - Document related tools, such as document content reading, permission management, etc.

> - `preset.task.default` - Task management related tools, such as task creation, member management, etc.

> - `preset.calendar.default` - Calendar event management tools, such as creating calendar events, querying free/busy status, etc.

6. Using SSE Mode with Specific Port and Host:

lark-mcp mcp -a cli_xxxx -s yyyyy -m sse --host 0.0.0.0 -p 3000

7. Setting Tools Language to Chinese:

lark-mcp mcp -a cli_xxxx -s yyyyy -l zh

> Note: Setting the language to Chinese (`-l zh`) may consume more tokens. If you encounter token limit issues when integrating with large language models, consider using the default English setting (`-l en`).

8. Setting Tool Name Format to Camel Case:

lark-mcp mcp -a cli_xxxx -s yyyyy -c camel

> Note: By setting the tool name format, you can change how tool names appear in the MCP. For example, `im.v1.message.create` in different formats:

> - snake format (default): `im_v1_message_create`

> - camel format: `imV1MessageCreate`

> - kebab format: `im-v1-message-create`

> - dot format: `im.v1.message.create`

9. Using Environment Variables Instead of Command Line Parameters:

# Set environment variables
   export APP_ID=cli_xxxx
   export APP_SECRET=yyyyy
   
   # Start the service (no need to specify -a and -s parameters)
   lark-mcp mcp

10. Using Configuration File:

Besides command line parameters, you can also use a JSON format configuration file to set parameters:

lark-mcp mcp --config ./config.json

Configuration file example (config.json):

{
      "appId": "cli_xxxx",
      "appSecret": "xxxx",
      "domain": "https://open.feishu.cn",
      "tools": ["im.v1.message.create","im.v1.chat.create"],
      "toolNameCase": "snake",
      "language": "zh",
      "userAccessToken": "",
      "tokenMode": "auto",
      "mode": "stdio",
      "host": "localhost",
      "port": "3000"
    }

> Note: Command line parameters have higher priority than configuration file. When using both command line parameters and configuration file, command line parameters will override corresponding settings in the configuration file.

11. Transport Modes:

lark-mcp supports two transport modes:

1. stdio mode (Default/Recommended): Suitable for integration with AI tools like Trae/Cursor or Claude, communicating through standard input/output streams.

lark-mcp mcp -a  -s  -m stdio

2. SSE mode: Provides an HTTP interface based on Server-Sent Events, suitable for scenarios where local execution is not possible.

# Default listens only on localhost
      lark-mcp mcp -a  -s  -m sse -p 3000
      
      # Listen on all network interfaces (allowing remote access)
      lark-mcp mcp -a  -s  -m sse --host 0.0.0.0 -p 3000

After startup, the SSE endpoint will be accessible at `http://:/sse`.

FAQ

• Issue: Unable to connect to Feishu/Lark API

Solution: Check your network connection and ensure your APP_ID and APP_SECRET are correct. Verify that you can access the Feishu/Lark Open Platform API; you may need to configure a proxy.

• Issue: Error when using user_access_token

Solution: Check if the token has expired. user_access_token usually has a validity period of 2 hours and needs to be refreshed periodically. You can implement an automatic token refresh mechanism.

• Issue: Unable to call certain APIs after starting the MCP service, with insufficient permissions errors

Solution: Check if your application has obtained the corresponding API permissions. Some APIs require additional high-level permissions, which can be configured in the Developer Console or Lark Developer Console. Ensure that permissions have been approved.

• Issue: Image or file upload/download related API calls fail

Solution: The current version does not support file and image upload/download functionality. These APIs will be supported in future versions.

• Issue: Command line displays garbled characters in Windows environment

Solution: Change the command line encoding to UTF-8 by executing `chcp 65001` in the command prompt. If using PowerShell, you may need to change the terminal font or PowerShell configuration.

• Issue: Permission errors during installation

Solution: On macOS/Linux, use `sudo npm install -g @larksuiteoapi/lark-mcp` for installation, or modify the permissions of the npm global installation path. Windows users can try running the command prompt as administrator.

• Issue: Token limit exceeded after starting the MCP service

Solution: Try using `-t` to reduce the number of enabled APIs, or use a model that supports larger tokens (such as claude3.7).

• Issue: Unable to connect or receive messages in SSE mode

Solution: Check if the port is already in use and try changing to a different port. Ensure that the client is correctly connected to the SSE endpoint and is handling the event stream.

Related Links

• Feishu Open Platform
• Lark International Open Platform
• Feishu Open Platform API Documentation
• Lark Open Platform API Documentation
• Node.js Website
• npm Documentation

Feedback

Issues are welcome to help improve this tool. If you have any questions or suggestions, please raise them in the GitHub repository.