drive-file-browser-mcp
MCP server that lets AI assistants browse, search, and read files from a Google Drive folder, supporting local stdio and remote HTTP deployments with tools like list_files, search_files, get_file_info, and read_file.
README
Drive File Browser MCP
A Model Context Protocol (MCP) server that lets AI assistants — Claude Desktop, Claude Code, Cursor, opencode, or any MCP client — browse, search, and read files from a Google Drive folder.
Runs in two modes:
- Local (stdio) — for desktop MCP clients
- Remote (HTTP, Streamable) — deployed on Vercel: https://drive-file-browser-mcp.vercel.app/api/mcp
Tools Provided
| Tool | Description |
|---|---|
list_files |
List files and subfolders in a Drive folder |
search_files |
Search files by name (case-insensitive substring) |
get_file_info |
Get metadata for a specific file (name, type, size, timestamps) |
read_file |
Read file content — Google Docs export as markdown, Sheets as CSV, Slides as plain text; plain text files returned directly; binary files return a link |
Remote Deployment (Vercel)
The server is live at:
https://drive-file-browser-mcp.vercel.app/api/mcp
It uses the stateless Streamable HTTP transport, so it works with any MCP client that supports remote HTTP servers. The endpoint is protected with a Bearer API key.
Connect a client
Claude Desktop / Claude Code (claude_desktop_config.json or .mcp.json):
{
"mcpServers": {
"drive-browser": {
"type": "http",
"url": "https://drive-file-browser-mcp.vercel.app/api/mcp",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}
Any HTTP client — send JSON-RPC over POST with headers Content-Type: application/json, Accept: application/json, text/event-stream, Authorization: Bearer <YOUR_API_KEY>:
curl -X POST https://drive-file-browser-mcp.vercel.app/api/mcp \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-H "Authorization: Bearer <YOUR_API_KEY>" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"curl","version":"1.0"}}}'
Health check (no auth): GET https://drive-file-browser-mcp.vercel.app/api/health
Deploy your own instance
- Fork/clone this repo, then:
npm install
npm i -g vercel
vercel login
vercel link --yes
# env vars:
echo "<your-api-key>" | vercel env add MCP_API_KEY production
echo "<your-drive-folder-id>" | vercel env add DRIVE_ROOT_FOLDER_ID production
echo "<service-account-json-contents>" | vercel env add GOOGLE_SERVICE_ACCOUNT_JSON production
vercel deploy --prod --yes
- Get a Google service account JSON (see below) and share your Drive folder with its
client_email.
Environment Variables (Vercel)
| Variable | Description |
|---|---|
GOOGLE_SERVICE_ACCOUNT_JSON |
The full service account JSON key (inline string) — required for Drive access |
DRIVE_ROOT_FOLDER_ID |
Default folder ID the server browses |
MCP_API_KEY |
Bearer token required on /api/mcp — leave unset to disable auth (not recommended) |
Manage them in the Vercel dashboard: Project → Settings → Environment Variables.
Local Setup (stdio mode)
1. Create a Google service account
- Go to the Google Cloud Console and create (or select) a project.
- Enable the Google Drive API (APIs & Services → Library → search "Google Drive API" → Enable).
- Go to APIs & Services → Credentials → Create Credentials → Service account.
- Open the service account → Keys tab → Add key → Create new key → JSON. A
.jsonkey file downloads. - Share the Drive folder with the service account's email (found in the JSON file as
client_email, looks likexxx@xxx.iam.gserviceaccount.com) — right-click the folder in Drive → Share → enter that email → Viewer.
2. Install and build
git clone https://github.com/ashrafur02-sys/drive-file-browser-mcp.git
cd drive-file-browser-mcp
npm install
npm run build
3. Configure
Set environment variables (or copy .env.example to .env — note: .env files are NOT auto-loaded; set real env vars in your MCP client config, or point GOOGLE_SERVICE_ACCOUNT_JSON at a file path):
| Variable | Description |
|---|---|
GOOGLE_SERVICE_ACCOUNT_JSON |
Inline service account JSON or a path to the JSON key file |
DRIVE_ROOT_FOLDER_ID |
Default folder ID (or full URL) the server browses |
Usage with MCP Clients
Add to your client's MCP configuration:
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"drive-browser": {
"command": "node",
"args": ["/absolute/path/to/drive-file-browser-mcp/dist/src/index.js"],
"env": {
"GOOGLE_SERVICE_ACCOUNT_JSON": "/absolute/path/to/service-account.json",
"DRIVE_ROOT_FOLDER_ID": "1TG4ssc_z0FgG2snDapoR_SRu3y16Wxm5"
}
}
}
}
Claude Code / opencode:
{
"mcpServers": {
"drive-browser": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/drive-file-browser-mcp/dist/src/index.js"],
"env": {
"GOOGLE_SERVICE_ACCOUNT_JSON": "/absolute/path/to/service-account.json",
"DRIVE_ROOT_FOLDER_ID": "1TG4ssc_z0FgG2snDapoR_SRu3y16Wxm5"
}
}
}
}
Once connected, ask your assistant things like:
- "List the files in the Drive folder"
- "Search for the press meet presentation"
- "Read the corporate communication file"
- "Get info about the file with ID ..."
Security Notes
- The server uses read-only Drive scope (
drive.readonly). - Never commit your service account key. It is gitignored by default.
- Anyone you give the key file to gets read access to everything the service account can see — share the repo, not your key.
License
MIT
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。