Google MCP Remote
A Cloudflare Workers-based MCP server implementation that provides Google API tools (Gmail, Calendar, Drive, etc.) for the Model Context Protocol, designed to integrate seamlessly with AI clients like Claude or Cursor.
接入你的智能体
從發布者安裝。不是 PlugMCP 端點。
遠端 URL 還是本機指令? 兩種安裝方式的差異
來自發布者
A Cloudflare Workers-based MCP server implementation that provides Google API tools (Gmail, Calendar, Drive, etc.) for the Model Context Protocol, designed to integrate seamlessly with AI clients like Claude or Cursor.
Collection of Google-native tools (e.g., Gmail, Calendar) for the MCP
⚠️ SECURITY WARNING: Do not use someone else's deployed instance of this server as it requires access to your Google account and personal data. Always deploy your own instance to maintain control over your data and API access.
Google Cloud Project:
Create a project in the Google Cloud Console
Set up OAuth 2.0 credentials (Client ID and Client Secret)
Enable the APIs you need (Gmail, Calendar, Drive, Tasks, YouTube, People/Contacts)
Add authorized JavaScript origins and redirect URIs for your Cloudflare Worker
Redirect URI should be in the format:
Deployed URL + /callback
https://your-project.your-username.workers.dev/callback
For local testing:
http://localhost:8788/callback
Cloudflare Account:
bun install -g wranglerwrangler loginClone the repository:
git clone https://github.com/vakharwalad23/google-mcp-remote.git
cd google-mcp-remote
Install dependencies:
bun install
Configure secrets: Set your Google OAuth credentials as secrets in Cloudflare:
wrangler secret put GOOGLE_OAUTH_CLIENT_ID
wrangler secret put GOOGLE_OAUTH_CLIENT_SECRET
wrangler secret put COOKIE_ENCRYPTION_KEY
(For the COOKIE_ENCRYPTION_KEY, generate a random string to secure cookies)
Create a KV namespace:
wrangler kv:namespace create OAUTH_KV
Then update your wrangler.jsonc with the ID from the output
Deploy to Cloudflare Workers:
bun run deploy
Note your deployment URL:
After deployment, Wrangler will provide a URL like: https://your-project.your-username.workers.dev
Once deployed, configure your AI client (Claude, Cursor, etc.) to use your MCP server.
Edit your claude_desktop_config.json or for cursor mcp.json:
{
"mcpServers": {
"google-mcp-remote": {
"command": "npx",
"args": [
"mcp-remote",
"https://your-project.your-username.workers.dev/sse"
]
}
}
}
You can now ask Claude to perform tasks using your Google account:
Send an email to [email protected] with the subject "Meeting Notes" and body "Here are the notes from today."
List my upcoming calendar events for the next 3 days.
Create a calendar event titled "Team Sync" tomorrow at 10 AM for 1 hour.
Search YouTube for recent videos about machine learning.
You can test the MCP server online using Cloudflare AI Playground:
/sse path:
https://your-project.your-username.workers.dev/sse
For local development and testing without deploying to Cloudflare:
Create a .dev.vars file in your project root with the necessary environment variables:
GOOGLE_OAUTH_CLIENT_ID="your-client-id"
GOOGLE_OAUTH_CLIENT_SECRET="your-client-secret"
COOKIE_ENCRYPTION_KEY="your-random-encryption-key"
Use Bun:
# Install dependencies
bun install
# Run local development server
bun run dev
Test with MCP Inspector:
bunx @modelcontextprotocol/inspector@latest
This launches a local interface to test your MCP server functionality
The first time you use the server with an AI client, you'll need to authorize access to your Google account:
To run the server locally:
bun install
bun run dev
This will start a local development server, typically at http://localhost:8788
Thank you for using Google MCP Remote! If you have any questions or suggestions, feel free to open an issue or contribute to the project.
Collection of Google-native tools (e.g., Gmail, Calendar) for the MCP
工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
這筆記錄未列出必要的環境變數。
Google MCP Remote 透過 stdio 在本機執行。
上次檢查時間:2026-10-28。 Collection of Google-native tools (e.g., Gmail, Calendar) for the MCP