Google Drive

Most of what a team knows is in a file somebody saved to Drive: the proposal template, the signed contract, the notes from last week's call. The Tulina connector lets an agent find those files, read them and put new ones back, under the Google account of the person who authorized it, then use what it read in the CRM, the inbox or a spreadsheet in the same conversation.

Does Google Drive have an MCP server?

Yes. Google documents a remote Drive MCP server at drivemcp.googleapis.com. On the pages we read on 31 October 2026 it is part of the Google Workspace Developer Preview Program, so access needs a membership in that program. It signs in with OAuth 2.0 and exposes eight tools: search_files, list_recent_files, get_file_metadata, read_file_content, download_file_content, create_file, copy_file and get_file_permissions. Google lists Claude as a client, on its Enterprise, Pro, Max or Team plans, through a custom connector.

Does Google support MCP?

For Workspace, Google's documentation describes one remote MCP server per product, covering Gmail, Drive, Docs, Sheets, Slides, Calendar, Chat and Classroom, all marked as a developer preview. Setting one up means a Google Cloud project with the matching APIs enabled, an OAuth consent screen and an OAuth client created for an AI agent.

How do you connect Google Drive to Claude and ChatGPT?

Through Tulina's own MCP server, with no Cloud project of your own:

1. In Tulina, open the Google Drive card and choose Authorize Google Drive. Google asks for the Drive scope only, on the account you pick.
2. Repeat for a second account if you work across two; each call names the account it uses, and Tulina refuses a call rather than guess.
3. Add https://mcp.tulina.ai/mcp to Claude or ChatGPT and sign in with your Tulina account. The MCP page has the same link.

Google's Drive MCP server and Tulina's Drive connector

If Drive is the only system in play and your Workspace admin is in the preview, Google's server is the shorter path. Tulina earns its place when a file is one step in longer work, such as reading a proposal from Drive, checking the deal in the CRM and drafting the reply in Gmail.

What you can ask it

"List the contracts modified this month in the Clients folder." The agent lists the folder with a date query and returns names, modification times and links.

"Read the Acme kickoff notes and write the three follow-ups." The agent exports the Google Doc as markdown, reads it and drafts the follow-ups. A spreadsheet comes back as CSV per tab, and a PDF as its extracted text.

"Share the proposals folder read-only with jane@example.com." The agent grants the reader role through drive_access and says whether Google sent a notification.

What runs on its own: a weekly sweep of a folder for new files can summarise them in Slack, and a sheet the team keeps in Drive can be read and updated through Google Sheets. The rest of the Google suite is in Google Workspace.

What the Drive connector can't do

• It acts as one Google account at a time, with that account's rights. An account that has not authorized Drive is refused, naming the card.
• A Google Doc, Sheet or Slides deck has no binary content to download. Read it with export; download is for uploaded files.
• A spreadsheet read is bounded: 200 rows per tab by default and 5,000 at most, and the result says when it was truncated.
• Upload fetches a file of up to 25 MB, and only some types convert to a Google format.
• delete moves a file to the trash and cannot be undone through the API. Sharing changes live in a separate tool on purpose.
• There is no tool for comments or revision history.