Bright Data MCP Server

To use the Bright Data MCP server in an MCP client (like Claude Desktop), add it to your mcpServers configuration. You must provide a valid API_TOKEN via environment variables.

Basic Setup

Use this configuration for standard web search and markdown scraping capabilities.

Advanced Configuration

Enable PRO_MODE to access over 60+ specialized tools (Browser automation, E-commerce, Social Media, etc.) and customize performance settings like rate limits and timeouts.

{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "your-token-here",
        "PRO_MODE": "true",              // Enable all 60+ tools
        "RATE_LIMIT": "100/1h",          // Custom rate limiting
        "WEB_UNLOCKER_ZONE": "custom",   // Custom unlocker zone
        "BROWSER_ZONE": "custom_browser", // Custom browser zone
        "POLLING_TIMEOUT": "600"         // Polling timeout in seconds (default: 600)
      }
    }
  }
}

Test cases are defined in JSON files within the tool-groups.json/ directory. Each test case object specifies the parameters for an automated evaluation run.

Fields:

• title: Description of the test.
• query: The natural language prompt sent to the LLM.
• runs: Number of times to execute the test.
• model: The specific LLM model string (e.g., gpt-5.1-2025-11-13).
• provider: The LLM provider (e.g., openai).
• expectedToolCalls: An array of tool names that the LLM is expected to invoke.
• selectedServers: An array of server names defined in your server configuration.
• advancedConfig: Optional object containing:
  • instructions: System instructions for the LLM.
  • temperature: LLM temperature.
  • maxSteps: Maximum number of tool call steps.
  • toolChoice: Strategy for tool selection (required or auto).

Example test case:

{
  "title": "Test E-commerce - Amazon product search",
  "query": "Search for wireless headphones on Amazon and show me the top products with reviews",
  "runs": 1,
  "model": "gpt-5.1-2025-11-13",
  "provider": "openai",
  "expectedToolCalls": ["web_data_amazon_product_search"],
  "selectedServers": ["ecommerce-server"],
  "advancedConfig": {
    "instructions": "You are a shopping assistant helping users find products on Amazon",
    "temperature": 0.1,
    "maxSteps": 5,
    "toolChoice": "required"
  }
}

You can customize the available tools by using environment variables in your MCP configuration:

• GROUPS: Enables curated tool bundles. Use comma-separated IDs (e.g., ecommerce,browser).
• TOOLS: Adds specific tool names on top of selected groups (e.g., extract).

Mode Priority: PRO_MODE=true (all tools) → GROUPS / TOOLS (whitelist) → default rapid mode (base toolkit).

Available Group IDs:

• ecommerce: Retail/marketplace (e.g., web_data_amazon_product)
• social: Social/community (e.g., web_data_linkedin_posts)
• browser: Scraping Browser automation (e.g., scraping_browser_screenshot)
• finance: Financial intelligence (e.g., web_data_yahoo_finance_business)
• business: Company/location intelligence (e.g., web_data_crunchbase_company)
• research: News/developer feeds (e.g., web_data_github_repository_file)
• app_stores: App store data (e.g., web_data_google_play_store)
• travel: Travel info (e.g., web_data_booking_hotel_listings)
• geo: GEO & LLM brand visibility (e.g., web_data_chatgpt_ai_insights)
• code: Package intelligence (e.g., web_data_npm_package)
• advanced_scraping: Batch/AI extraction (e.g., scrape_batch, extract)

The Bright Data MCP server is configured using several environment variables. These variables control authentication, zone selection, tool availability, and performance settings.

Required

• API_TOKEN: Your Bright Data API token.

Optional Configuration

• WEB_UNLOCKER_ZONE: The name of the Web Unblocker zone to use (defaults to mcp_unlocker).
• BROWSER_ZONE: The name of the Browser API zone to use (defaults to mcp_browser).
• PRO_MODE: Set to true to enable all available tools. If not set or false, tool access is restricted by GROUPS or TOOLS.
• GROUPS: A comma-separated list of tool group IDs to enable (e.g., group1,group2).
• TOOLS: A comma-separated list of specific tool names to enable (e.g., search_engine,extract).
• RATE_LIMIT: A string defining the rate limit in the format limit/time (e.g., 100/1h or 50/30m).
• POLLING_TIMEOUT: Timeout in seconds for long-running tasks like discover (defaults to 600).
• BASE_TIMEOUT: Base timeout for API requests in seconds.
• BASE_MAX_RETRIES: Maximum number of retries for failed requests (capped at 3).