Office Word MCP Server

The Office-Word-MCP-Server is a Model Context Protocol (MCP) server that allows AI assistants to interact with Microsoft Word documents. It exposes document operations as tools and resources, enabling tasks such as:

• Document Management: Creating, reading, merging, and converting documents (e.g., to PDF).
• Content Creation: Adding headings, paragraphs, tables, images, lists, and footnotes with rich formatting.
• Rich Text & Table Formatting: Controlling fonts, colors, cell shading, borders, alignment, and cell merging.
• Advanced Manipulation: Deleting paragraphs, inserting content at specific indices, and applying custom styles.
• Document Protection: Adding password protection, restricted editing, and digital signatures.
• Comment Extraction: Retrieving comments and their metadata (author, date, text).

To deploy the Office Word MCP Server on Render, you must configure specific environment variables to ensure the server uses Server-Sent Events (SSE) instead of the default STDIO transport.

Required Variables:

• MCP_TRANSPORT: Must be set to sse.

Optional Variables:

• MCP_HOST: Set to 0.0.0.0 to bind to all network interfaces (defaults to 0.0.0.0).
• FASTMCP_LOG_LEVEL: Sets the logging level (defaults to INFO).

Follow these steps to deploy on Render:

1. Log in to the Render dashboard.
2. Select your Office-Word-MCP-Server service.
3. Navigate to the Environment section in the sidebar.
4. Add the environment variable MCP_TRANSPORT with the value sse.
5. Save changes. Render will automatically redeploy the service.

Once deployed, the server will be accessible via SSE at your service URL followed by /sse (e.g., https://office-word-mcp-server-bzlp.onrender.com/sse).

To automatically install the Office Word Document Server for Claude Desktop, use the Smithery CLI. This is the recommended method for quick setup with Claude.

npx -y @smithery/cli install @GongRzhe/Office-Word-MCP-Server --client claude

If you prefer a manual installation, ensure you have Python 3.8+ and pip installed. Clone the repository and install the required dependencies from requirements.txt.

# Clone the repository
git clone https://github.com/GongRzhe/Office-Word-MCP-Server.git
cd Office-Word-MCP-Server

# Install dependencies
pip install -r requirements.txt

To enable detailed logging for debugging the MCP server, set the MCP_DEBUG environment variable to 1.

export MCP_DEBUG=1  # Linux/macOS
set MCP_DEBUG=1     # Windows

The repository provides a setup_mcp.py script that automates the installation process, including prerequisite checks, virtual environment setup, dependency installation, and MCP configuration generation.

python setup_mcp.py

You can run the server without a local installation by using the uvx package manager. This is useful for a lightweight, managed execution environment.

{
  "mcpServers": {
    "word-document-server": {
      "command": "uvx",
      "args": ["--from", "office-word-mcp-server", "word_mcp_server"]
    }
  }
}

After installing the server locally, add it to your Claude Desktop configuration file. You must provide the absolute path to the word_mcp_server.py file.

{
  "mcpServers": {
    "word-document-server": {
      "command": "python",
      "args": ["/path/to/word_mcp_server.py"]
    }
  }
}

Common pitfalls when working with tables:

• Cell index errors: Ensure row and column indices are within table bounds (0-based indexing).
• Color format problems: Use hex colors without '#' prefix (e.g., FF0000 for red) or standard color names.
• Padding unit confusion: Specify points or percent explicitly when setting cell padding.
• Column width conflicts: auto_fit_table_columns may override manual column width settings.
• Text formatting persistence: Apply format_table_cell_text after setting cell content for best results.

If you encounter issues with your Render deployment, check the following common causes:

Retrieve comments from a document, either all comments, comments by a specific author, or comments associated with a specific paragraph index.

get_all_comments(filename)
get_comments_by_author(filename, author)
get_comments_for_paragraph(filename, paragraph_index)