Sujan30/docs-mcp-server

📝 Docs MCP Server

Welcome to the Docs MCP Server!
This tool gives your AI the ability to create and edit Google Docs via MCP, turning natural language into powerful document workflows.

So you can do things like this:

⚙️ Core Functionalities

With this server, your AI (like Claude or GPT) can:

✅ Create a new Google Doc with just a title
✅ Create a new Google Doc with both title and content
✅ Add more content to an existing document
✅ Create google calendar events & list out the next x events with just plain text

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/your-username/docs-mcp-server.git
cd docs-mcp-server

2. Install MCP CLI

uv add "mcp[cli]"

3. Create & Activate a Virtual Environment

python3.11 -m venv venv
source venv/bin/activate

4. Install Python Dependencies

pip install -r requirements.txt

5. Set Up Google Docs API

• Go to Google Cloud Console.
• Enable the Google Docs API.
• Create OAuth credentials and choose "Desktop Application" when asked for the app type.
• Download the credentials.json file.
• Place the file in your project folder.
• In authentication.py, set the CREDENTIALS_PATH to the path of your credentials.json.

6. Run the Server

python server.py

• This will open a Google OAuth window in your browser.
• Once authenticated, a token.pickle file will be generated.
• In authentication.py, set the TOKEN_FILE to the path of your token.pickle.

🎉 Your Docs MCP server is now running!

🧠 Connecting to Claude (Anthropic)

1. Ensure Claude Desktop is Installed

2. Run the MCP server install command:

mcp install server.py

3. Open Claude Desktop App Settings
   (Cmd + , or click on the Cluade button in the top bar, and navigate to settings)

4. Navigate to the Developer tab
   You should see google-docs-mcp listed.

⚠️ If you see errors, don’t worry — continue below.

5. Click "Edit Config" at the bottom left. This opens your claude_desktop_config.json.

7. Locate your google-docs-mcp entry. Update the args to include necessary dependencies:

🔧 Before:

"args": [
  "run",
  "--with",
  "mcp[cli]",
  "mcp",
  "run",
  "$YOUR PATH"
]

✅ After:

"google-docs-mcp": {
  "command": "/usr/local/bin",  // or wherever your Python binaries live
  "args": [
    "run",
    "--with",
    "mcp[cli]",
    "--with",
    "google-api-python-client",
    "--with",
    "google-auth",
    "--with",
    "google-auth-oauthlib",
    "--with",
    "google-auth-httplib2",
    "--with",
    "python-dotenv",
    "mcp",
    "run",
    "/Users/sujannandikolsunilkumar/docs mcp/docs-mcp/server.py" #Your path will be different, don't need to change it. 
  ]
}

7. Save the file and restart Claude Desktop.

✅ Final Test

After restarting Claude:

• You should be prompted to complete Google OAuth.

• Once authorized, click on "Search & Tools" inside Claude.

• You should see your Google Docs MCP Server listed and ready.

✨ You Can Now Ask Claude To:

• Create a Google Doc and give it a title
• Create a Google Doc with a title and content
• Add more text to an existing Google Doc

🆘 Need Help?

Feel free to reach out:
📧 nandikolsujan@gmail.com

🧠 About MCP

This project uses Model Context Protocol (MCP) to enable LLMs to interact with tools directly through self-describing interfaces—no hardcoding needed. Claude reads and understands the tool capabilities and executes accordingly.