Bright Data MCP Server
Enables AI agents to browse and extract data from any public website via Bright Data's cloud infrastructure, handling proxy rotation, CAPTCHA solving, and JavaScript rendering.
README
Bright Data MCP Server
A full-featured Model Context Protocol (MCP) server that lets an AI agent browse and extract data from any public website through Bright Data's Web Access APIs.
All heavy lifting — proxy rotation, browser fingerprinting, CAPTCHA solving, and
JavaScript rendering — happens on Bright Data's cloud infrastructure, so this
server runs fine on machines that can't (or shouldn't) run a browser locally. The
Browser API tool uses puppeteer-core, which speaks CDP to a remote browser and
never downloads or launches Chromium on your machine.
Tools
| Tool | What it does |
|---|---|
unlocker_scrape |
Fetch any URL as HTML, clean Markdown, or a PNG screenshot. Auto-handles blocking & CAPTCHAs. |
unlocker_scrape_async |
Start a long-running unlock job; returns a response_id. |
unlocker_get_async_result |
Poll/fetch the result of an async unlock job. |
unlocker_success_rate |
Per-domain success-rate stats (last 7 days). |
serp_search |
Structured search results from Google / Bing / Yandex / DuckDuckGo. |
web_scraper_trigger |
Trigger a Crawl / Web Scraper dataset job over one or more URLs. |
web_scraper_get_results |
Check progress and download dataset results by snapshot_id. |
browser_scrape |
Drive a real cloud browser: navigate, click, type, scroll, run JS, auto-solve CAPTCHAs, emulate devices, block ads, return rendered HTML / text / screenshot. |
Prerequisites
- A Bright Data account and an API token (Account settings → API tokens, or https://brightdata.com/cp/setting/users).
- One or more zones created under Web Access APIs in the control panel:
- a Web Unlocker zone (for
unlocker_*andserp_search) - a SERP API zone (for
serp_search) - a Browser API zone (for
browser_scrape) — copy itsUSER:PASSfrom the zone's Overview tab - a Dataset id (
gd_...) from https://brightdata.com/cp/datasets for the Crawl / Web Scraper tools
- a Web Unlocker zone (for
- Node.js 18 or newer.
Bright Data's free tier includes monthly credits; adding a payment method unlocks API access and grants a small verification credit. Usage of all of these APIs is billed by Bright Data per their pricing.
Install & build
npm install
npm run build
Configure
Copy .env.example to .env and fill in your values:
cp .env.example .env
| Variable | Required | Default | Notes |
|---|---|---|---|
BRIGHTDATA_API_KEY |
yes | — | Bearer token for all REST calls. |
BRIGHTDATA_UNLOCKER_ZONE |
no | web_unlocker1 |
Web Unlocker zone name. |
BRIGHTDATA_SERP_ZONE |
no | serp_api1 |
SERP API zone name. |
BRIGHTDATA_BROWSER_AUTH |
for browser_scrape |
— | USER:PASS of the Browser API zone. |
BRIGHTDATA_BROWSER_HOST |
no | brd.superproxy.io:9222 |
CDP host:port. |
BRIGHTDATA_DATASET_ID |
for crawl tools | — | Default dataset id (gd_...). |
BRIGHTDATA_API_BASE_URL |
no | https://api.brightdata.com |
Override base URL. |
The server reads variables from the real process environment. If you keep them in
.env, load it before launching (e.g. node --env-file=.env dist/index.js on
Node 20+, or via your MCP client's env block below).
Run
# Node 20+ can load the .env file directly:
node --env-file=.env dist/index.js
# or rely on environment variables already exported in your shell:
npm start
Use it from an MCP client
Add this to your client's MCP config (Claude Desktop, Cursor, etc.). Pass secrets
through the env block rather than committing them:
{
"mcpServers": {
"brightdata": {
"command": "node",
"args": ["/absolute/path/to/brightdata-mcp/dist/index.js"],
"env": {
"BRIGHTDATA_API_KEY": "your_api_key_here",
"BRIGHTDATA_UNLOCKER_ZONE": "web_unlocker1",
"BRIGHTDATA_SERP_ZONE": "serp_api1",
"BRIGHTDATA_BROWSER_AUTH": "brd-customer-XXXX-zone-YYYY:zzzz",
"BRIGHTDATA_DATASET_ID": "gd_xxxxxxxx"
}
}
}
}
Example tool calls
Fetch a page as Markdown:
{ "tool": "unlocker_scrape", "arguments": { "url": "https://example.com", "data_format": "markdown" } }
Search Google:
{ "tool": "serp_search", "arguments": { "query": "best laptops 2026", "engine": "google", "num": 20 } }
Drive a cloud browser through a CAPTCHA and screenshot it:
{
"tool": "browser_scrape",
"arguments": {
"url": "https://site-with-captcha.example",
"solve_captcha": true,
"screenshot": true,
"actions": [
{ "type": "wait_for_selector", "selector": "#results" },
{ "type": "click", "selector": "button.load-more" },
{ "type": "scroll" }
]
}
}
How it maps to the Bright Data docs
- Unlocker / SERP →
POST https://api.brightdata.com/requestwith{ zone, url, format, method, country, data_format }. Async adds?async=trueand is retrieved viaGET /unblocker/get_result. - Crawl / Web Scraper →
POST /datasets/v3/trigger?dataset_id=..., polled viaGET /datasets/v3/progress/{id}and downloaded viaGET /datasets/v3/snapshot/{id}. - Browser API →
puppeteer.connect({ browserWSEndpoint: "wss://USER:PASS@brd.superproxy.io:9222" })plus Bright Data's custom CDP commands:Captcha.solve,Captcha.setAutoSolve,Unblocker.enableAdBlock,Proxy.useSession,Emulation.setDevice,Browser.getSessionId.
Responsible use
This server accesses public web data through a paid, compliant commercial
service. Respect each target site's Terms of Service and robots.txt, applicable
laws (e.g. data-protection rules), and Bright Data's
Acceptable Use Policy. Don't use it
to access content behind logins/authentication you aren't authorized to use, or to
collect personal data without a lawful basis.
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。