Overview
Workspace MCP connects AI assistants to all twelve major Google Workspace services through a single MCP server exposing 120+ tools. It supports both local stdio operation for legacy clients and remote streamable HTTP deployment conforming to the latest MCP spec, and ships with a full-featured CLI plus a Code Mode for tools like Claude Code and Codex.
Key Features
- 120+ tools covering Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Tasks, Contacts, and Chat with fine-grained editing capabilities
- Multi-user OAuth 2.1 authentication, enabling centralized, secure hosting for an entire organization
- Three progressive tool tiers and a read-only mode for granular capability control
- Full CLI and Code Mode for use inside developer tools such as Claude Code and Codex
Use Cases
- Letting AI assistants manage email, calendars, and documents end-to-end in natural language
- Centrally hosting a shared Workspace MCP instance behind a reverse proxy for a team or organization
- Automating document editing, spreadsheet operations, and Drive file workflows from coding agents
Technical Details
- Python 3.10+ server published to PyPI as
workspace-mcp, with dependencies pinned inuv.lock - Stateless container deployment mode with zero disk writes, plus optional credential store backends (GCS/CMEK)
- Sensitive-path blocking via
validate_file_path()that guards.env*files and home-directory credential stores like~/.ssh/and~/.aws/even with broadenedALLOWED_FILE_DIRS - No telemetry or external network calls beyond Google APIs; supports external auth server and gateway passthrough auth