TaobaoScraper MCP Server

An MCP (Model Context Protocol) server for scraping product data from Taobao/Tmall and JD.com (Jingdong). Provides 8 tools that can be used directly in Claude Desktop or Claude Code.

Platform: Windows (requires Chrome browser) Language: Python 3.10+

Features

• Multi-platform scraping - Taobao, Tmall, JD.com in one tool

• 8 MCP tools - Scraping, task management, notifications, system control

• Hot search words - Discover trending keywords and market insights

• Excel export - Auto-saves results as .xlsx files

• Notification push - WeChat (ServerChan / PushPlus) and Email alerts

• Dual transport - stdio (local) and SSE (remote) protocols

• Trilingual GUI - Chinese / English / Korean interface (optional)

Related MCP server: Taobao Sourcing Assistant

Architecture

Claude Desktop / Claude Code
        |
   MCP Server (stdio or SSE)
        |
   FastAPI Backend (:8000)
        |
   Selenium + Chrome (:9222)

Quick Start

1. Prerequisites

• Python 3.10+

• Google Chrome browser

• Windows OS

2. Install

git clone https://github.com/jhongjun1981/taobao-scraper-mcp.git
cd taobao-scraper-mcp

# Install MCP server dependencies
pip install -r requirements_mcp.txt

# Install API backend dependencies
pip install -r requirements_api.txt

3. Configure

cp .env.example .env
# Edit .env to set your SCRAPER_API_KEY

4. Start Chrome with debug port

# Close all Chrome instances first, then:
START_GUI.bat
# Or manually:
chrome.exe --remote-debugging-port=9222 --user-data-dir="%LOCALAPPDATA%\Google\Chrome\Debug Profile"

5. Login to Taobao (first time only)

Open Chrome and login to your Taobao account. Cookies will persist in the debug profile.

6. Start API Backend

python run_api.py

7. Configure MCP Server

For Claude Code - Add to .claude/settings.json:

{
  "mcpServers": {
    "taobao-scraper": {
      "command": "python",
      "args": ["run_mcp.py"],
      "cwd": "/path/to/taobao-scraper-mcp"
    }
  }
}

For Claude Desktop - Add to claude_desktop_config.json:

{
  "mcpServers": {
    "taobao-scraper": {
      "command": "python",
      "args": ["run_mcp.py"],
      "cwd": "C:\\path\\to\\taobao-scraper-mcp"
    }
  }
}

SSE mode (remote):

python run_mcp.py --transport sse --port 8001

{
  "mcpServers": {
    "taobao-scraper": {
      "type": "sse",
      "url": "http://your-server:8001/sse"
    }
  }
}

Tools Reference

Scraping Tools

scrape_products

Scrape product data from e-commerce platforms.

Example: "Scrape running shoes from both Taobao and JD"

scrape_hotwords

Scrape trending/related search keywords for market analysis.

Task Management Tools

list_tasks

List recent scraping tasks with status (pending/running/completed/failed).

get_task

Get detailed status of a specific task. Set include_result=true to get full results.

cancel_task

Cancel a running or pending task.

Export Tools

list_files

List all exported Excel data files with filename, size, and modification time.

Notification Tools

send_notification

Send push notifications via WeChat or Email.

System Tools

system_status

Check system health or restart Chrome.

Environment Variables

Data Output

Scraped data is automatically saved as Excel files containing:

• Product title, image URL, price

• Monthly sales volume, review count

• Shop name, shop type, shop region

• Product URL, scrape timestamp

Important Notes

• Taobao requires login - You must login to your own Taobao account in Chrome first

• JD works without login - JD scraping does not require authentication

• One task at a time - Chrome can only handle one scraping task concurrently

• Windows only - The Chrome automation layer uses Windows-specific APIs

• Respect rate limits - Excessive scraping may trigger anti-bot protection

License

MIT License - see LICENSE file.