使用说明
brightdata mcp服务器代理ai web scraper模板
BrightData MCP服务器代理AI网络爬虫模板
An intelligent, plug-and-play web scraper that combines Bright Data's Mobile Carrier Proxy (MCP)
with Anthropic’s Claude for agentic, natural language-based scraping. A lightweight
Tkinter GUI lets you describe what to scrape in plain English — no coding required.
I can envison tool also reading step from QC/ALM and feed the agent the steps
重要图书馆
| 库 | 描述 |
|---|---|
mcp |
与任务控制平台的接口，用于管理参数、会话和遥测。 |
langchain_mcp_adapters |
与LangChain兼容的MCP工具和代理。 |
langgraph |
启用基于图形的反应式代理流。 |
langchain_anthropic |
集成Claude语言模型。 |
dotenv |
从 .env 文件。 |
asyncio |
启用代理操作的异步执行。 |
os |
用于访问环境变量。 |
✨ 主要特点 🌐 Bright Data MCP代理集成
Bypass geo-restrictions, CAPTCHAs, and anti-bot defenses using mobile carrier IPs.
🧠 Claude的代理AI
Claude interprets your natural language instructions and dynamically generates scraping behavior.
🧾 自然语言接口
Just describe the task in plain English — the scraper figures out the rest.
💻 Tkinter图形用户界面
A desktop interface to enter instructions, monitor scraping progress, and view logs.
🔁 Retry & Adapt
Uses reasoning to handle failed pages, change in structure, or blocking events.
🧱 Modular & Scalable
Clearly separated logic (GUI, agent, scraper, MCP) for easier maintenance and extension.
📷 演示示例
"Go to zillow.com, search Denver, and scrape the price, address, and number of beds for the first 5 listings."
✅ Type this into the GUI.
✅ Claude interprets it.
✅ Scraper executes it using Bright Data’s mobile proxy.
✅ Results saved — no code touched.
🛠 需求
Node.js (v16 or higher)
Python 3.x
Bright Data account with MCP access
Claude Desktop App (Anthropic)
Internet connection
⚙️ 设置 🔹 1.克劳德桌面设置
Download and install from Anthropic
Launch the app
Go to Settings → Developer Settings
🔹 2. Create anthropic accound and Bright Data account with the correct tools
- Anthropic API Key (will require credit card) https://console.anthropic.com/settings/keys - This key will be used in the python project as a environment variable
- Bright data create and account and add Browser API tool and a Unlocked API tool see images below. You can name it what you would like but on step three you will need to use the name WEB_UNLOCKER_ZONE = <Web unlocker name>
🔹 3. Bright Data MCP Server
Option A – Use config.json
json
Copy
Edit
{
"mcpServers": {
"Bright Data": {
"command": "npx",
"args": ["@brightdata/mcp"],
"env": {
"API_TOKEN": "<BrightData API TOKEN>",
"WEB_UNLOCKER_ZONE": "<BrightData Web Unlocker Zone>",
"BROWSER_AUTH": "<BrightData Browser Auth>"
}
}
}
}
有关mpc设置的更多明亮数据，请访问：https://github.com/brightdata/brightdata-mcp
需要的工具
Python设置
⚙️ 设置
🔹 环境变量
你 必须创建 .env Python根目录中的文件
API_TOKEN=your_brightdata_api_token
WEB_UNLOCKER_ZONE=your_web_unlocker_zone
BROWSER_AUTH=your_browser_auth
ANTHROPIC_API_KEY=anthropic_key
