Gemini CLI Google Docs MCP: Connect and Check Access
Gemini CLI can connect to Google Docs through an MCP server or extension that uses the Google Drive API. Here is how to evaluate connection options, set up the right access scope, and keep shared project decisions and follow-up work outside your document drafts.
Updated 5 min read
Gemini CLI can connect to Google Docs through an MCP server that wraps the Google Drive API, letting it read, search, and in some configurations write to documents in your Drive. There is no single official Google Docs MCP extension; instead, several community and third-party servers support this use case with varying feature sets and access scopes. fenbs can hold the shared project decisions and follow-up work that belong outside the document itself, keeping drafts clean and context available to any connected assistant.
Gemini CLI is a supported MCP client and connects to fenbs over MCP at Integrations, so the same project context your Google Docs work draws on is available to Gemini CLI at the start of each session.
How a Google Docs MCP connection works
A Google Docs MCP server uses the Google Drive API (and sometimes the Google Docs API for structured document operations) to read and modify documents. You authorize the server with your Google account through OAuth. The server then exposes tools that Gemini CLI can call: read a document by ID, search Drive for documents matching a query, append text to a document, and similar operations.
The Google Drive API is well-documented and stable, which makes it a reliable base for MCP servers. The key variables are which OAuth scopes the server requests and whether it stores any document content after returning it to the client.
Choosing the right OAuth scope
The Google Drive API offers several OAuth scopes with different levels of access. Two are most relevant for document work:
https://www.googleapis.com/auth/drive.readonly: read all files in Drive. Suitable for document review and research tasks.https://www.googleapis.com/auth/drive.file: read and write only files created by the app or files the user explicitly opens with it. More restrictive and appropriate when you want to limit the server's reach.https://www.googleapis.com/auth/drive: full Drive access, including metadata and all files. Request this only if the task genuinely requires it.
Start with drive.readonly for a read-only setup or drive.file for a limited read-write setup. Grant full Drive access only when you have reviewed the server's code or documentation and confirmed that broader access is necessary.
Evaluating a Google Docs MCP server
Several open-source MCP servers exist for Google Drive and Docs. When evaluating one for use with Gemini CLI, check: does it use the official Google Drive API or scrape the Docs web interface? What OAuth scopes does it request, and can you configure narrower scopes? Does it store document content on its own servers or pass it directly to the calling client? When was it last updated?
A server that stores retrieved documents on a third-party server is riskier than one that passes content directly to your Gemini CLI session. Review the server's documentation and, if it is open source, its source code before authorizing Drive access.
For general Gemini CLI setup and how to install and configure MCP tools, see how to install Gemini CLI. For a comparison of MCP server versus CLI extension patterns, see MCP vs CLI.
Using Google Docs for drafts, fenbs for decisions
A common mistake when using AI to work on documents is embedding project decisions inside the document itself. A document draft starts to accumulate notes like "still deciding on the scope of section 3" or "awaiting client approval on this paragraph." Those notes belong in a project tool, not in the draft.
Keep documents clean. Put the content in Google Docs. Put the decisions, the outstanding questions, and the approval state in fenbs. That separation means the document is shareable without exposing internal process notes, and the decisions are available to any assistant without reading the whole draft.
Keep decisions and follow-up outside the document
Add decisions about the document to fenbs as Decisions and rules: the approved outline, the agreed tone, the sections that need another review. Save outstanding tasks as fenbs tasks. When Gemini CLI reads the project context at the start of a session, it sees those decisions and can work on the document accordingly, without you re-explaining the editorial choices.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
Organizational Google accounts and admin restrictions
If your Google Drive is managed by an organization through Google Workspace, the administrator may restrict which OAuth apps can connect to Drive. A third-party MCP server may be blocked until it is approved in the Google Workspace Admin console under Apps, OAuth Apps, or the equivalent in your current admin interface.
If authorization fails or the server cannot access Drive, contact your Google Workspace administrator. You may need to submit the app for approval before the connection can proceed.
How to do this with fenbs
- Create a free account at fenbs. Add a project for the document work.
- Write the decisions that govern the document (scope, audience, structure, approval state) as Decisions and rules in the project.
- Connect Gemini CLI to fenbs over MCP from Integrations. Configure your chosen Google Docs MCP server in Gemini CLI's tool config.
- At the start of each session, ask Gemini CLI to read the fenbs project context first, then pull the Google Doc. The decisions and outstanding tasks are in fenbs; the current draft is in Drive.
- Before ending the session, save Where we left off in fenbs with the current state of the document and the next editorial step.
fenbs does not connect to Google Docs. It holds the project decisions and follow-up tasks that belong alongside the document. For other Google and MCP integrations with Gemini CLI, see Google Sheets MCP.