worksuite-mcp

worksuite-mcp

Connects AI assistants to Google Workspace (Gmail, Drive, Docs, Sheets, Slides, Calendar, Tasks, Forms, Chat, Contacts) through your own OAuth app, providing around 120 tools for managing email, files, calendar, and more.

Category
访问服务器

README

Worksuite MCP

Connect Claude (or any MCP client) to your Google Workspace. Worksuite MCP is a polished, batteries-included Model Context Protocol server that gives your AI assistant full access to Gmail, Google Drive, Docs, Sheets, Slides, Calendar, Forms, Tasks, Chat and Contacts, all through your own Google account and your own OAuth app. Your data stays yours.

It wraps the excellent open-source workspace-mcp engine and adds a clean launcher, a self-healing startup, and the setup guide below so you can be talking to your inbox and Drive in a few minutes.


What you can do once it is connected

  • Gmail - search, read, draft, send, and label email.
  • Drive - search files, read content, upload, share, and manage folders.
  • Docs / Sheets / Slides - create and edit documents, read and write cells, build decks.
  • Calendar - read your schedule, create and move events, check free/busy.
  • Tasks, Forms, Chat, Contacts - full read/write across the rest of Workspace.

Around 120 tools in total. You talk to your AI in plain language ("find the invoice from Acme and forward it to finance") and it uses these tools to do it.


Requirements

  • macOS or Linux
  • uv (provides uvx). Install it with:
    curl -LsSf https://astral.sh/uv/install.sh | sh
    
  • A Google account
  • An MCP client. These instructions use Claude Code, but any MCP client works.

Step 1 - Get the code

git clone https://github.com/chadvdwww-hash/worksuite-mcp.git
cd worksuite-mcp
chmod +x bin/worksuite-mcp.sh

Step 2 - Create your Google OAuth app

Worksuite MCP signs in as you, using your own Google Cloud OAuth client. You create it once. It takes about five minutes.

  1. Open the Google Cloud Console and create (or pick) a project.
  2. Enable the APIs you want under APIs & Services → Library. At minimum enable: Gmail API, Google Drive API, Google Docs API, Google Sheets API, Google Slides API, Google Calendar API. (Enable Tasks, Forms, Chat and People APIs too if you want those tools.)
  3. Go to APIs & Services → OAuth consent screen:
    • User type: External.
    • Fill in the app name (e.g. "Worksuite MCP") and your email.
    • Under Test users, add the Google account you will sign in with. This is required while the app is in Testing mode, or Google blocks the login with 403 access_denied.
  4. Go to APIs & Services → Credentials → Create credentials → OAuth client ID:
    • Application type: Desktop app (or Web application).
    • If you pick Web application, add this authorized redirect URI: http://localhost:8000/oauth2callback
    • Download the client secret JSON.

Heads up on token lifetime: while your OAuth app is in Testing mode, Google expires the refresh token after 7 days, so you re-authenticate roughly weekly. Publish the app (OAuth consent screen → Publish app) to make it permanent.

Step 3 - Point Worksuite MCP at your credentials

Tell the server where your downloaded client secret lives:

export GOOGLE_CLIENT_SECRET_PATH="/absolute/path/to/your/client_secret.json"

(You can also set GOOGLE_OAUTH_CLIENT_ID and GOOGLE_OAUTH_CLIENT_SECRET directly instead of a file. Either way, keep these secret. They are gitignored.)

Step 4 - Register it with Claude Code

claude mcp add worksuite-mcp -- /absolute/path/to/worksuite-mcp/bin/worksuite-mcp.sh

Then check it:

claude mcp list
# worksuite-mcp: .../bin/worksuite-mcp.sh - ✔ Connected

Step 5 - Sign in to Google (first run)

The first time your AI calls a Workspace tool, Worksuite MCP opens a Google sign-in page in your browser. Approve the scopes (you may see an "unverified app" warning because it is your personal app; click through to continue). After that your credentials are cached and reused automatically.

That's it. Ask your assistant to "list my recent Drive files" or "summarise my unread email" to confirm it works.


Good to know (saves you an hour)

  • Port 8000 must be free. The OAuth sign-in flow listens on localhost:8000 (it can roam up to 8004). The launcher auto-reaps stale Worksuite/Workspace instances holding those ports before each start, so a crashed session won't block the next one. If sign-in still fails with "port in use", something else is on 8000: lsof -nP -iTCP:8000 -sTCP:LISTEN.
  • File uploads are sandboxed. Tools that import a local file only read from ~/.workspace-mcp/attachments/ by default. Either copy the file there first, or set ALLOWED_FILE_DIRS=/your/dir and restart the server.
  • Avoid dots in new document titles. When creating a Doc from a file, a name like Cal.com Setup gets truncated to Cal. Use Cal-com Setup and rename afterward if needed.
  • Where credentials live. Once you sign in, tokens are stored under ~/.google_workspace_mcp/credentials/. Delete that file to force a fresh login.

How it works

Worksuite MCP is a thin wrapper. It pins the upstream workspace-mcp package, applies its own branding and startup banner, runs a port-reaper so a crashed run never blocks the next, then hands off to the upstream server which does the Google API heavy lifting. Upgrading the engine is a one-line version bump in pyproject.toml.

your AI client  <-- MCP -->  Worksuite MCP launcher  -->  Google Workspace APIs

Credits

Built on workspace-mcp by Taylor Wilsdon and contributors. Licensed under the MIT License (see LICENSE).

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
Kagi MCP Server

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。

官方
精选
Python
graphlit-mcp-server

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。

官方
精选
TypeScript
Exa MCP Server

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。

官方
精选
mcp-server-qdrant

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。

官方
精选
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选