Bright Data¶

The Bright Data MCP Server connects your ADK agent to Bright Data's web data platform. This tool gives your agent the ability to perform real-time web searches, scrape webpages, extract structured data, control browsers remotely, and access pre-built data feeds from popular platforms.

Use cases¶

Real-Time Web Search: Perform optimized web searches to get up-to-date information in AI-friendly formats (JSON/Markdown).
Structured Data Extraction: Use AI-powered extraction to convert any webpage into clean, structured JSON data with optional custom prompts.
Browser Automation: Control real browsers remotely for complex interactions, JavaScript rendering, and dynamic content extraction.
Pre-Built Data APIs: Access 60+ structured datasets from popular platforms including Amazon, LinkedIn, Instagram, TikTok, Google Maps, and more.
Advertisement Analysis: Extract and analyze advertisements from webpages using industry-standard ad blocking filter lists.

Prerequisites¶

Sign up for a Bright Data account to obtain an API token.
Refer to the documentation for more information.
The server offers a free tier with 5,000 requests/month, which is useful for prototyping and everyday workflows.

Use with agent¶

Local MCP ServerRemote MCP Server

from google.adk.agents import Agent
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset
from mcp import StdioServerParameters

BRIGHTDATA_API_TOKEN = "YOUR_BRIGHTDATA_API_TOKEN"

root_agent = Agent(
    model="gemini-2.5-pro",
    name="brightdata_agent",
    instruction="Help users access web data using Bright Data",
    tools=[
        MCPToolset(
            connection_params=StdioConnectionParams(
                server_params = StdioServerParameters(
                    command="npx",
                    args=[
                        "@brightdata/mcp",
                    ],
                    env={
                        "API_TOKEN": BRIGHTDATA_API_TOKEN,
                        "PRO_MODE": "true",  # Optional: Enable all 60+ tools
                    }
                ),
                timeout=300,
            ),
        )
    ],
)

from google.adk.agents import Agent
from google.adk.tools.mcp_tool.mcp_session_manager import StreamableHTTPServerParams
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset

BRIGHTDATA_API_TOKEN = "YOUR_BRIGHTDATA_API_TOKEN"

root_agent = Agent(
    model="gemini-2.5-pro",
    name="brightdata_agent",
    instruction="""Help users access web data using Bright Data""",
    tools=[
        MCPToolset(
            connection_params=StreamableHTTPServerParams(
                url=f"https://mcp.brightdata.com/mcp?token={BRIGHTDATA_API_TOKEN}",
            ),
        )
    ],
)

Example usage¶

Once your agent is set up and running, you can interact with it through the command-line interface or web interface. Here are some examples:

Sample agent prompts:

Get me the current price and details of the iPhone 15 Pro on Amazon

Search for "climate change news 2025" on Google and summarize the top 5 results

Scrape the homepage of techcrunch.com and extract all article headlines and links

The agent automatically calls the appropriate Bright Data tools to provide comprehensive answers, making it easy to access real-time web data without manual navigation or worrying about getting blocked.

Available tools¶

The Bright Data MCP server operates in two modes:

Rapid Mode (Free Tier - Default)¶

Tool	Description
`search_engine`	Scrape Google, Bing, or Yandex SERPs as JSON or Markdown.
`scrape_as_markdown`	Convert webpages into clean Markdown with built-in unblocking.
`scrape_as_html`	Return raw HTML from webpages while bypassing blockers.
`extract`	Transform Markdown output into structured JSON with custom prompts.
`session_stats`	View session usage statistics and tool call counts.

Pro Mode (60+ Additional Tools)¶

Enable Pro Mode by setting PRO_MODE=true in environment variables to access:

Batch Operations: - search_engine_batch: Run up to 10 search queries simultaneously. - scrape_batch: Scrape up to 10 URLs simultaneously.

Browser Automation: - scraping_browser.*: Full browser control for complex interactions. - Navigate, click, type, scroll, take screenshots, and more.

Web Data APIs (60+ Structured Datasets):

E-commerce: web_data_amazon_product, web_data_walmart_product, web_data_ebay_product, web_data_etsy_products, web_data_bestbuy_products, web_data_zara_products
Social Media: web_data_linkedin_person_profile, web_data_instagram_profiles, web_data_facebook_posts, web_data_tiktok_profiles, web_data_x_posts, web_data_reddit_posts
Business Intelligence: web_data_linkedin_company_profile, web_data_crunchbase_company, web_data_zoominfo_company_profile
Search & Reviews: web_data_amazon_product_search, web_data_amazon_product_reviews, web_data_google_maps_reviews, web_data_facebook_company_reviews
Maps & Local: web_data_google_maps_reviews, web_data_zillow_properties_listing, web_data_booking_hotel_listings
App Stores: web_data_google_play_store, web_data_apple_app_store
Media & News: web_data_youtube_videos, web_data_youtube_comments, web_data_reuter_news
Developer Tools: web_data_github_repository_file
Finance: web_data_yahoo_finance_business

All Web Data API tools return cached or fresh structured data in JSON format, often more reliable than real-time scraping.

Configuration options¶

The Bright Data MCP server supports several environment variables for customization:

Variable	Description	Default
`API_TOKEN`	Your Bright Data API token (required)	-
`PRO_MODE`	Enable all 60+ advanced tools	`false`
`RATE_LIMIT`	Custom rate limiting (e.g., "100/1h", "50/30m")	No limit
`WEB_UNLOCKER_ZONE`	Custom Web Unlocker zone name	`mcp_unlocker`
`BROWSER_ZONE`	Custom Browser API zone name	`mcp_browser`