FAQ: MCP

Find answers to common questions about integrating, configuring, and using Bright Data’s MCP server.

What is Bright Data MCP?

Bright Data MCP (Model Context Protocol) is a server that enables LLMs, AI agents, and applications to access, discover, and extract web data in real-time. It allows MCP clients like Claude Desktop, Cursor, and Windsurf to search the web, navigate websites, take actions, and retrieve data without getting blocked, making it perfect for web scraping tasks.

The MCP server provides advanced capabilities, including:

• Bypassing geo-restrictions to access content regardless of location
• Navigating websites with bot detection protection using Web Unlocker technology
• Structured data extraction from platforms like Amazon, LinkedIn, Instagram, and many other data feeds
• Remote browser automation for complex web interactions
• Access to a global IP network to avoid blocking or rate limiting

Get started with Bright Data MCP:

1. Explore the MCP GitHub repository to review the source code, contribute, or deploy your own instance.
2. Read the MCP documentation to understand the API, request structure, and supported features.
3. Get your API key
4. Try out the MCP Smithery playground to test and experiment with live requests in a no-code environment.

Where do I get my API keys?

You can get your API token from the user settings page in your Bright Data account.

Make sure you have an account on brightdata.com - new users get free credit for testing, and pay-as-you-go options are available after that.

How do I enable browser control tools?

To enable browser control tools:

1. Visit your Bright Data control panel at brightdata.com/cp/zones
2. Create a new ‘Browser API’ zone
3. Once created, copy the authentication string from the Browser API overview tab
4. The authentication string will be formatted like: brd-customer-[your-customer-ID]-zone-[your-zone-ID]:[your-password]
5. Add this authentication string to your MCP configuration as the BROWSER_AUTH environment variable

What tools are available in the Bright Data MCP?

For the full list of tools, please refer to our GitHub repo: MCP Tools

Bright Data MCP offers a comprehensive set of tools, including:

Search and basic scraping:

• search_engine: Scrape search results from Google, Bing or Yandex
• scrape_as_markdown: Scrape a webpage and get results in Markdown format
• scrape_as_html: Scrape a webpage and get results in HTML format
• session_stats: View tool usage during the current session

Structured data extraction:

• Amazon: web_data_amazon_product, web_data_amazon_product_reviews
• LinkedIn: web_data_linkedin_person_profile, web_data_linkedin_company_profile
• ZoomInfo: web_data_zoominfo_company_profile
• Instagram: web_data_instagram_profiles, web_data_instagram_posts, web_data_instagram_reels, web_data_instagram_comments
• Facebook: web_data_facebook_posts, web_data_facebook_marketplace_listings, web_data_facebook_company_reviews
• Various others: web_data_x_posts, web_data_zillow_properties_listing, web_data_booking_hotel_listings, web_data_youtube_videos

Browser automation:

• scraping_browser_navigate: Navigate to a URL
• scraping_browser_go_back/scraping_browser_go_forward: Navigate browser history
• scraping_browser_click: Click on an element
• scraping_browser_links: Get all links on the current page
• scraping_browser_type: Type text into an element
• scraping_browser_wait_for: Wait for an element to appear
• scraping_browser_screenshot: Take a screenshot
• scraping_browser_get_html/scraping_browser_get_text: Get page content

How do I handle timeout issues when using Bright Data MCP?

Some web data tools can take longer to execute, especially when dealing with complex websites. To ensure your agent can consume the data:

1. Set a high enough timeout in your agent settings (180 seconds is recommended for 99% of requests)
2. For particularly slow sites, you may need to increase this value further
3. Use the specialized web_data_* tools when available, as they’re often faster than general scraping
4. For browser automation sequences, keep operations close together in time

What security considerations should I keep in mind?

Always treat scraped web content as untrusted data. Never use raw scraped content directly in LLM prompts to avoid potential prompt injection risks.

Instead:

• Filter and validate all web data before passing it to the LLM
• Use structured data extraction rather than raw text when possible
• Be cautious with executing JavaScript from scraped content

Why am I getting a 'spawn npx ENOENT' error?

This error occurs when your system cannot find the npx command. To fix it:

1. Find your Node.js path:
   • On macOS: Run which node in Terminal
   • On Windows: Run where node in Command Prompt
2. Update your MCP configuration to use the full path instead of npx:

{
  "mcpServers": {
    "Bright Data": {
      "command": "/usr/local/bin/node", // Replace with your actual Node.js path
      "args": ["node_modules/@brightdata/mcp/index.js"],
      "env": {
        "API_TOKEN": "<your-api-token>"
      }
    }
  }
}

How do I monitor my usage and costs?

You can use the session_stats tool to check your usage during the current session. For comprehensive usage tracking and billing information, log in to your Bright Data account dashboard.

The session_stats tool will show you information about:

• Number of requests made during the current session
• Tools used and their frequency

Is there a playground where I can try Bright Data MCP?

Yes, you can try Bright Data MCP without any setup using the playground on Smithery.

This platform provides an easy way to explore the capabilities of Bright Data MCP without any local setup. Just sign in and start experimenting with web data collection!