Microsoft Office tools and utilities as a Model Context Protocol (MCP) server.

MCP Server for Interaction with Office Apps

This is a Model Context Protocol (MCP) server project.

Warning: Be VERY careful with open Word or PowerPoint apps. Your content may be unexpectedly modified or deleted.

Setup and Installation

Simply run:

make

To create the virtual environment and install dependencies.

Building the Standalone Executable (Windows Only)

To build the standalone executable for this project, you must:

  1. Ensure you are on a Windows system.
  2. Install the development dependencies (including PyInstaller):
    make
    
  3. Run the build command to generate the executable:
    make package
    

This will create a mcp-server-office.exe file inside the dist/ folder.

Running the Standalone Executable

Once built, the executable can be run by simply double-clicking it, or from the command prompt:

./dist/mcp-server-office.exe

The server will start in SSE mode and run on port 25566. To expose the server publicly, use the provided batch file (run_with_devtunnel.bat) to set up a Dev Tunnel and start the server.

Running the Server

Use the VSCode launch configuration, or run manually:

Defaults to stdio transport:
uv run -m mcp_server.start
For SSE transport:
uv run -m mcp_server.start --transport sse --port 25566

To use this MCP server with a hosted Semantic Workbench assistant, go to libraries:mcp-tunnel and run the following command and copy its output into your assistant configuration:

uv run mcp-tunnel --servers "office:25566"
If you need a public-facing server, use the --use-ngrok-tunnel option:
uv run -m mcp_server.start --use-ngrok-tunnel

or for .exe:

mcp-server-office.exe --use-ngrok-tunnel

The SSE URL is:

http://127.0.0.1:25566/sse

Client Configuration

To use this MCP server in your setup, consider the following configuration:

Stdio

{
  "mcpServers": {
    "mcp-server-word": {
      "command": "uv",
      "args": ["run", "-m", "mcp_server.start"]
    }
  }
}

SSE

{
  "mcpServers": {
    "mcp-server-word": {
      "command": "http://127.0.0.1:25566/sse",
      "args": []
    }
  }
}

MCP Server · Populars

MCP Server · New

    googleads

    Google Ads MCP Server

    MCP Server

    Community googleads
    StefanMaron

    AL MCP Server

    MCP Server

    Community StefanMaron
    vintasoftware

    Django AI Boost

    A MCP server for Django applications, inspired by Laravel Boost.

    Community vintasoftware
    idleprocesscc

    Co-Reading MCP

    A local co-reading MCP server for chunked books, reading progress, search, and margin annotations.

    Community idleprocesscc
    JasonJarvan

    Export-Zhihu-Collections

    将知乎收藏夹(公开和私密)批量导出为 Markdown 格式文件. 支持配置收藏夹列表并批量下载, 支持基于URL自动去重, 支持下载专栏和图片, 支持自定义保存路径, 支持调试和错误日志. **同时提供 MCP Server**,可被 AI Agent (如 Claude Code) 直接调用,为大模型提供保存知乎收藏夹的能力。

    Community JasonJarvan