workspace-mcp

What it is and what it does

Workspace MCP is an MCP server that bridges AI assistants and LLM clients to Google Workspace by exposing Calendar, Gmail, Drive, Docs, Sheets, Slides, Forms, Tasks, Contacts, Chat, Custom Search, and Apps Script as callable tools. It implements OAuth 2.0 and OAuth 2.1 (PKCE) authentication, supports multi-user deployments, and offers three progressive tool tiers (core, extended, complete) to manage context window size. The server runs locally over stdio for legacy clients or remotely over streamable HTTP with the latest MCP specification.

The package is designed for both individual use and centralized organizational deployment. It includes a CLI for direct invocation, supports stateless container deployment, and provides fine-grained access control through read-only mode, per-service permissions, and tool-level filtering. All credentials remain in your environment, and the server sends data only to Google APIs on behalf of the authenticated user—no built-in telemetry, analytics, or external dependencies beyond what you explicitly configure.

Use it for:

• Enable Claude or other AI assistants to search Gmail, compose drafts, and manage labels and filters on behalf of a user.
• Allow LLM-powered workflows to create, read, and edit Google Docs, Sheets, and Slides programmatically through natural language.
• Build multi-user AI applications that securely delegate Google Workspace access through OAuth 2.0 without storing user credentials.
• Deploy a centralized MCP server behind a reverse proxy to serve an entire organization with fine-grained tool and permission controls.
• Integrate Google Calendar event management, free/busy queries, and focus time scheduling into AI agent workflows.
• Use Apps Script integration to write, deploy, and debug Google Workspace automation scripts through an AI assistant.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Workspace MCP is an MCP server that exposes Google Workspace services (Calendar, Gmail, Drive, Docs, Sheets, Slides, Forms, Tasks, Contacts, Chat, Custom Search, and Apps Script) as tools for AI assistants and LLM clients through a single standardized interface.

Yes, if you need AI assistants to interact with Google Workspace services. The package is actively maintained, MIT licensed with no restrictions, has low install friction, and offers comprehensive coverage of twelve Google Workspace products with flexible deployment options. The substantial dependency tree (17 runtime packages) and requirement for Google Cloud OAuth setup are expected trade-offs for the breadth of functionality. No known security vulnerabilities as of the query date.

Install

workspace-mcp on PyPI

pip install workspace-mcp

uv add workspace-mcp

poetry add workspace-mcp

Installing workspace-mcp

Before you install

Low install friction with a pure-Python wheel. Active maintenance (last commit 2026-08-13, released 1 day ago) and a substantial dependency tree of 17 runtime packages, all permissively licensed. No known vulnerabilities as of the query date.

License in practice

MIT licensed with no dual licensing, commercial restrictions, or contributor license agreement. You may use, fork, embed, and redistribute commercially without restriction, provided you retain attribution.

Quickstart

# Install
pip install workspace-mcp

# Set credentials
export GOOGLE_OAUTH_CLIENT_ID="..."
export GOOGLE_OAUTH_CLIENT_SECRET="..."

# Launch
uvx workspace-mcp --tool-tier core

Requires Python >=3.10. Requires a Google Cloud project with OAuth 2.0 credentials and enabled APIs for the services you plan to use.

Verify before relying

• Whether the 120+ tools claim and exact tool counts per service (e.g., 15 for Gmail, 16 for Drive) are current and accurate.
• Performance characteristics and latency under typical usage patterns with multiple concurrent users.
• Completeness of Google Workspace API coverage relative to official Google APIs and any known limitations or gaps.

Package facts

Evidence: workspace_mcp-1.24.1-py3-none-any.whl

Keywords: mcp, google, workspace, llm, ai, claude, model, context, protocol, server

Tags

More Artificial Intelligence packages

Further reading