Google Drive MCP Server

An MCP server for managing files and folders in Google Drive, integrating seamlessly with AI assistants using the Model Context Protocol.

Features

• List Files/Folders: Search and list files with advanced query support.
• Get Metadata: Retrieve detailed information about specific files or folders.
• Manage Folders: Create new folders and organize content.
• Organize Content: Move, copy, and delete (or trash) files.
• Trash Management: List trashed items and restore by file ID or name.
• Export: Export Google Docs, Sheets, and Slides to formats like PDF or DOCX.
• Download / Upload: Download files or whole folders into a local workspace, edit them, and upload them back (update in place or create new).

Installation

Prerequisites

• Python 3.9 or higher.
• A Google Cloud Project with the Drive API enabled.
• OAuth2 credentials (Client ID and Client Secret).
• OAuth scope: ["https://www.googleapis.com/auth/drive"].

Local Setup

1. Install dependencies:

   pip install -e .
   

2. Run the server:

   python main.py --access-token <TOKEN> --refresh-token <REFRESH_TOKEN> --client-id <CLIENT_ID> --client-secret <CLIENT_SECRET>
   

Configuration

Add the following to your MCP host configuration (e.g., mcp-config.json):

{
  "mcpServers": {
    "barie-google-drive-mcp": {
      "command": "uvx",
      "args": [
        "barie-google-drive-mcp",
        "--access-token",
        "YOUR_ACCESS_TOKEN",
        "--refresh-token",
        "YOUR_REFRESH_TOKEN",
        "--client-id",
        "YOUR_CLIENT_ID",
        "--client-secret",
        "YOUR_CLIENT_SECRET"
      ]
    }
  }
}

Tools

• list_files: Search and list files.
• get_file_metadata: Get file details.
• create_folder: Create a new folder.
• delete_file: Delete or trash a file.
• move_file: Move a file to a new folder.
• copy_file: Copy a file.
• export_file: Export Google Workspace files and return base64 content with metadata.
• list_trashed_files: List files/folders currently in trash.
• restore_file: Restore trashed files/folders by fileId or name.
• download_file: Download a file or folder into the workspace and return the local path(s).
• upload_file: Upload a workspace file or folder; with fileId it updates the existing Drive file in place.

Download / Upload

Files are transferred between Drive and a local workspace directory; file content is never returned in the tool result, only the local path.

Typical agent flow:

1. download_file {"fileId": "abc123"} → {"localPath": "/home/workspace/budget.xlsx", "modifiedTime": "..."}
2. The agent edits /home/workspace/budget.xlsx.
3. upload_file {"path": "/home/workspace/budget.xlsx", "fileId": "abc123", "expected_modified_time": "..."}

Behavior:

• Google Docs/Sheets/Slides/Drawings are exported on download (docx/xlsx/pptx/png by default, or export_format). An edited docx/xlsx/pptx can update the original Google file in place.
• upload_file with fileId keeps the file ID, link and sharing; the previous content stays in Drive version history. Without fileId a new file is created in parentId (or My Drive).
• Folders: a folder ID downloads the whole tree. Uploading a directory with a folder fileId updates changed same-name files (unchanged ones are skipped), creates new ones, and never deletes anything.
• expected_modified_time makes the update fail if the Drive file changed since it was downloaded.

Restore Tool Behavior

restore_file supports two modes:

1. Restore by ID:

{ "fileId": "FILE_ID" }

2. Restore by name:

{
  "name": "report",
  "exactMatch": false,
  "restoreAllMatches": false
}

Notes:

• Provide either fileId or name.
• When restoring by name and multiple files match, only the first is restored unless restoreAllMatches is true.

License

MIT