MCP Servers · taylorwilsdon

google_workspace_mcp

A Python-based MCP server that connects AI assistants and agent platforms to Google Workspace services (Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Chat, Tasks, Contacts, Apps Script, Search). It runs locally or can be deployed centrally using OAuth 2.0/2.1 with no third-party service dependencies.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/taylorwilsdon/google_workspace_mcp

2.8k

GitHub stars

849

Forks

Python

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	taylorwilsdon/google_workspace_mcp
Owner	taylorwilsdon
Primary language	Python
License	MIT — OSI-approved
Stars	2.8k
Forks	849
Open issues	133
Latest release	v1.22.0 (2026-06-22)
Last updated	2026-07-07
Source	https://github.com/taylorwilsdon/google_workspace_mcp

What google_workspace_mcp is

Stateless Python application exposing 12 Google Workspace services via the Model Context Protocol. Supports OAuth 2.0/2.1 (including PKCE), multi-user bearer tokens, remote deployments, and fine-grained API scoping. No external backends or telemetry; all calls flow directly to Google APIs.

Quickstart

Get the google_workspace_mcp source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/taylorwilsdon/google_workspace_mcp.git2cd google_workspace_mcp3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

AI-assisted email and calendar workflows

Natural language control over Gmail inbox management, calendar event scheduling, and meeting coordination through Claude, ChatGPT, or other MCP-compatible assistants without manual UI navigation.

Centralized organizational deployment with SSO

Deploy a single Workspace MCP instance behind your organization's reverse proxy, using OAuth 2.1 bearer tokens to serve multiple users while keeping credentials in-house and respecting granular permission scopes.

Document and data automation in AI workflows

Use Claude Code or similar tools to read, generate, and manipulate Docs, Sheets, Slides, and Forms directly within LLM-driven workflows—combining fine-grained editing with AI reasoning for report generation, template population, and data processing.

Implementation considerations

OAuth client setup is prerequisite: create credentials in a GCP project, configure scopes (recommend read-only or service-specific permissions rather than full access), and manage redirect URIs for your deployment model.
Choose between stateful mode (local credentials storage) or stateless mode (bearer tokens via environment) based on your infrastructure—stateless mode is container-friendly but requires external token management.
Deployment complexity ranges from local CLI (`uvx workspace-mcp`) to centralized OAuth 2.1 multi-user server; reverse proxy and CORS setup required for production multi-user deployments.
Audit the full dependency tree: project pins dependencies in `uv.lock`. No AGPL or copyleft present; all dependencies are MIT, Apache 2.0, or BSD compatible.
Set `ALLOWED_FILE_DIRS` and review `validate_file_path()` logic: default blocks `.env*`, `~/.ssh/`, `~/.aws/` but can be expanded—review boundary conditions if broadening access.

When to avoid it — and what to weigh

Requires real-time bi-directional sync — This is a request/response MCP tool layer, not an event-driven synchronization engine. Changes originate from the AI/user, not from Workspace events triggering the server.
Need no OAuth complexity or offline-first operation — Requires active Google authentication and internet access to Google APIs. Not suitable for fully disconnected or offline-first applications.
Strict requirement to avoid any external API calls — By design, this server calls Google APIs on behalf of authenticated users. If your threat model prohibits any egress to third parties (even Google), this is unsuitable.
Windows-first or non-Python environments — Python 3.10+ required. While cross-platform, primary development and testing assumes Unix/Linux/macOS; Windows support is undocumented in the available data.

License & commercial use

MIT License (permissive). Full commercial use permitted: build products, sell services, deploy internally, fork, embed, redistribute with attribution only. No dual licensing, no CLA, no commercial tier gating.

MIT is a permissive OSI license with no restrictions on commercial use, derivative works, or redistribution. No telemetry, licensing server, or usage reporting—safe for proprietary deployment. However, ensure your use of Google Workspace APIs complies with Google's ToS and licensing; this project does not grant Google API rights.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Strong
Assessment confidence	High

Security considerations

No claims of absolute security made here. Key considerations: credentials (OAuth client secret) must be protected at rest and in transit (use environment variables, secrets managers, not version control). Server sends data only to Google APIs on behalf of authenticated user—no intermediary servers or telemetry. Stateless mode has no disk writes, reducing local credential exposure. `validate_file_path()` blocks sensitive paths by default but can be overridden—review custom `ALLOWED_FILE_DIRS` settings. Review dependency licenses and update cadence. Deployment behind reverse proxy recommended for centralized use. OAuth 2.1 (PKCE) preferred for untrusted clients.

Alternatives to consider

Google's built-in Claude integration or ChatGPT plugin

Simpler setup with no server deployment, but limited feature coverage, less fine-grained control, and no custom scoping. Suitable if pre-built integrations meet your needs and you accept vendor lock-in.

Zapier or Make (formerly Integromat)

Low-code/no-code workflow automation for Workspace. Easier for non-developers; hosted by vendor. Less expressive than AI-driven tools but no coding required; higher operational cost at scale.

Custom Google Apps Script or direct Google API client

Full control, no third-party server. Much higher development effort, poor ergonomics for AI assistants, and no MCP standard. Suitable for internal tooling only, not AI agent integration.

Software development agency

Build on google_workspace_mcp with DEV.co software developers

Start with the interactive quick-start guide at workspacemcp.com/quick-start, or dive into the CLI: uvx workspace-mcp --tool-tier core

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

excel-mcp-server mcp-atlassian atlassian-mcp-server

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI application development API development Enterprise software Talk to DEV.co

google_workspace_mcp FAQ

Can I use this with free Google accounts?

Yes. Free accounts are supported. Some features (Chat, Spaces, advanced security) are limited to paid Google Workspace plans. Scopes are configurable to match your account's permissions.

Does the server store, log, or relay my credentials?

No. OAuth tokens are managed by your application or the built-in OAuth flow. Credentials never leave your infrastructure. In stateless mode, bearer tokens are passed via environment; no persistent storage occurs. No telemetry or usage reporting.

How do I deploy this for my whole team?

Use OAuth 2.1 multi-user mode: run a single server instance, configure your OAuth client, and have each user authenticate via the built-in flow. Deploy behind a reverse proxy for HTTPS. Token refresh and multi-user state are handled automatically or via stateless bearer token management.

What if my MCP client doesn't support all 12 services?

You can cherry-pick services at launch: `uv run main.py --tools gmail drive calendar` starts only those services. Or use the CLI directly: `uvx workspace-mcp` provides full CLI access independent of MCP client support.

Work with a software development agency

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If google_workspace_mcp is part of your mcp servers roadmap, our team can implement, customize, migrate, and maintain it.

Ready to connect your AI to Google Workspace?

Start with the interactive quick-start guide at workspacemcp.com/quick-start, or dive into the CLI: uvx workspace-mcp --tool-tier core

Talk to DEV.co Schedule a Consultation