google-cli-mcp
An MCP server that exposes 17 Google Workspace APIs (e.g., Gmail, Drive, Calendar) as auto-generated tools for AI assistants, enabling natural language control of Google services.
README
google-cli-mcp
An MCP (Model Context Protocol) server that exposes Google Workspace APIs as tools for AI assistants like Claude, Cursor, and others. It uses Google's Discovery API to auto-generate tools at runtime — no manual tool definitions required.
Built with Claude Code by Anthropic.
What it does
Connects your AI assistant to 17 Google Workspace services:
| Alias | Service |
|---|---|
drive |
Google Drive |
sheets |
Google Sheets |
gmail |
Gmail |
calendar |
Google Calendar |
docs |
Google Docs |
slides |
Google Slides |
tasks |
Google Tasks |
people |
Google People / Contacts |
chat |
Google Chat |
classroom |
Google Classroom |
forms |
Google Forms |
keep |
Google Keep |
meet |
Google Meet |
admin-directory |
Admin SDK - Directory |
admin-reports |
Admin SDK - Reports |
events |
Workspace Events |
modelarmor |
Model Armor |
Requirements
- Node.js >= 18
- A Google account with access to the services you want to use
- Google credentials (see Authentication below)
gcloud CLI is not required — but it is one of the two supported auth methods.
Installation
git clone https://github.com/ChrisNoronha/google-cli-mcp.git
cd google-cli-mcp
npm install
npm run build
Authentication
The server supports two credential methods. Use whichever fits your setup.
Option 1: Bearer token (no gcloud needed)
Set the GOOGLE_WORKSPACE_CLI_TOKEN environment variable to a valid Google OAuth 2.0 bearer token:
export GOOGLE_WORKSPACE_CLI_TOKEN="ya29.your-token-here"
Note: Bearer tokens expire (typically after 1 hour) and must be rotated manually. This option is best for quick testing or short-lived automation.
Option 2: Application Default Credentials (ADC)
This method uses gcloud to authenticate once and handles token refresh automatically — recommended for long-running use.
# Install gcloud CLI if you haven't already
# https://cloud.google.com/sdk/docs/install
gcloud auth application-default login
After logging in, the server will pick up your credentials automatically.
Configuration
All configuration is done via environment variables.
| Variable | Default | Description |
|---|---|---|
GOOGLE_WORKSPACE_CLI_TOKEN |
— | Bearer token (Option 1 auth) |
GWS_MCP_SERVICES |
all services | Comma-separated list of service aliases to load (e.g. gmail,drive,calendar) |
GWS_MCP_SCOPES |
all scopes | Space-separated OAuth scopes (only used with ADC auth) |
Load only specific services
export GWS_MCP_SERVICES="gmail,drive,calendar"
Leave GWS_MCP_SERVICES unset to load all 17 services.
Usage
Run the server
# Development (no build step)
npm run dev
# Production (after npm run build)
node dist/index.js
Connect to Claude Desktop
Add this to your Claude Desktop MCP config (claude_desktop_config.json):
{
"mcpServers": {
"google-workspace": {
"command": "node",
"args": ["/path/to/google-cli-mcp/dist/index.js"],
"env": {
"GWS_MCP_SERVICES": "gmail,drive,calendar"
}
}
}
}
Connect to Claude Code
claude mcp add google-workspace -- node /path/to/google-cli-mcp/dist/index.js
Development
npm run dev # run with tsx (no build)
npm run build # compile TypeScript to dist/
npm test # run tests
npm run test:watch # watch mode
How it works
- Auth — resolves credentials from env var or ADC
- Discovery — fetches Google API Discovery Documents in parallel (10s timeout per service)
- Tool generation — builds MCP tool definitions from the Discovery Documents automatically
- Server — starts a stdio MCP server and registers all tools
- Dispatch — each tool call is proxied to the corresponding Google REST API with the resolved auth token
License
MIT
Created with Claude Code
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。