Google Docs MCP Server

A Python-based MCP (Message Control Protocol) server that can read Google Docs and add comments to documents using a service account.

Features

• Authenticate with Google Docs API using a service account (no user interaction required)
• Read content from Google Docs
• Add comments to documents with references to specific text
• Programmatically interact with documents owned by your platform

Why Use a Service Account?

A service account is ideal for server applications that need to access Google Docs without user interaction. This approach is perfect for:

• Platforms that own the Google Docs and need to add comments programmatically
• Automated systems that process documents
• Background services that don't have a user interface

Setup

1. Clone this repository

2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up a Google Cloud Project and service account:

   • Go to Google Cloud Console
   • Create a new project
   • Enable the Google Docs API and Google Drive API
   • Create a service account and download the key file
   • Rename the downloaded key file to service-account-key.json and place it in the project root (Use service-account-key.example.json as a reference for the expected format)
   • Share your Google Docs with the service account's email address

4. Run the server:

   python server.py
   

See setup_instructions.md and service_account_guide.md for detailed setup instructions.

Usage

The server exposes several endpoints:

• /docs - List available documents (shared with the service account)
• /docs/<doc_id> - Get document content
• /docs/<doc_id>/comments - Add a comment to a document

Example of adding a comment:

curl -X POST http://localhost:8000/docs/DOCUMENT_ID/comments \
  -H "Content-Type: application/json" \
  -d '{"comment": "This is a test comment", "search_text": "text to comment on"}'

Comment Functionality

Limitations

The Google Docs API has limitations when it comes to adding comments to specific text using service accounts:

1. The Docs API's createComment method doesn't fully support attaching comments to specific text when using service accounts.
2. The Drive API's comments are file-level and don't support direct attachment to specific text.

Current Approach

Our current approach is to:

1. Find the specific text in the document using the Docs API
2. Add a comment using the Drive API that references the specific text in its content
   • Example: "Regarding text 'search text': This is a comment"

This approach ensures that:

• Comments are reliably added to documents
• Comments reference the specific text they're about
• The implementation works consistently with service accounts

Security Notes

Important: This repository contains a service account key example file that should be replaced with your own:

• service-account-key.example.json: Template for your service account key
• Never commit actual credentials to GitHub
• Ensure your .gitignore includes service-account-key.json and other credential files
• For production deployment, consider using environment variables or a secure vault for credentials

Testing

You can test the service account authentication with:

python test_auth.py

You can test adding comments with:

python client_example.py "text to comment on"

License

MIT