Bright Data MCP Server

Bright Data MCP is an official Model Context Protocol server from Bright Data that gives AI agents reliable, unblocked access to the live web. It bundles Bright Data's web unlocker, SERP API, scraping browser, and structured Web Data APIs into a single MCP endpoint so agents can search, fetch, navigate, and extract structured data without hitting CAPTCHAs, rate limits, or IP bans.

In its default Rapid mode the server exposes a small set of free tools (search, markdown scraping, AI-ranked discovery, batch search). Enabling Pro mode unlocks 60+ tools across curated groups including ecommerce, social, browser control, finance, business, research, app stores, travel, and advanced scraping. These include domain-specific Web Data extractors for sites like Amazon, Walmart, LinkedIn, TikTok, YouTube, Crunchbase, ZoomInfo, Yahoo Finance, npm, and PyPI, plus a remote scraping browser that agents can drive interactively with snapshots, clicks, and screenshots.

The server can be self-hosted via the @brightdata/mcp npm package or used as a hosted remote endpoint at https://mcp.brightdata.com/mcp. Bright Data provides 5,000 free requests per month and works with Claude Desktop, Cursor, Windsurf, and other MCP-compatible clients.

Tools

Prerequisites

• A Bright Data account and API token (sign up at brightdata.com, then create a token in the account settings)
• Node.js installed if self-hosting

Option 1: Hosted remote server (no install)

Point your MCP client at the hosted endpoint and include your token as a query parameter:

https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN_HERE

In Claude Desktop: Settings → Connectors → Add custom connector, paste the URL above.

Option 2: Local (npx) install for Claude Desktop, Cursor, etc.

{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "<your-api-token-here>"
      }
    }
  }
}

Enabling Pro mode and tool groups

{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "<your-api-token-here>",
        "PRO_MODE": "true",
        "GROUPS": "ecommerce,browser,advanced_scraping"
      }
    }
  }
}

Environment variables

• API_TOKEN (required): Bright Data API token
• PRO_MODE: set true to expose all 60+ tools
• GROUPS: comma-separated tool group IDs (ecommerce, social, browser, finance, business, research, app_stores, travel, advanced_scraping)
• TOOLS: comma-separated individual tool names to enable
• RATE_LIMIT: custom rate limit
• WEB_UNLOCKER_ZONE: defaults to mcp_unlocker
• BROWSER_ZONE: defaults to mcp_browser
• POLLING_TIMEOUT: timeout in seconds for web_data tools (default 600)
• BASE_TIMEOUT, BASE_MAX_RETRIES: timeout and retry behavior for base tools

• Run competitive ecommerce monitoring by pulling structured Amazon, Walmart, and Google Shopping data for a list of SKUs each day
• Scrape and convert long-form competitor pages or docs into clean markdown so an agent can summarize them without getting blocked
• Build a sales research agent that enriches accounts with Crunchbase and ZoomInfo company profiles plus LinkedIn post activity
• Track GEO (generative engine optimization) by querying ChatGPT, Perplexity, and Grok insight tools to see how your brand surfaces in AI answers
• Drive a remote scraping browser to log into JS-heavy sites, take snapshots, click through flows, and extract data interactively

• "Search Google for 'best CRM for startups 2026' and return the top 10 results as markdown."
• "Scrape this Amazon product URL and tell me the price, rating, and top three negative review themes."
• "Pull the Crunchbase profile and recent LinkedIn posts for Acme Corp and draft an outbound email."
• "Use the scraping browser to open this page, click 'Pricing', and screenshot the resulting plan table."
• "Check how ChatGPT and Perplexity describe 'Bright Data' and summarize differences in positioning."

• Official server maintained by Bright Data with both self-hosted (npx @brightdata/mcp) and hosted remote endpoints
• Built-in unblocking infrastructure means agents rarely hit CAPTCHAs, IP bans, or rate limits on hard targets
• 60+ Pro tools cover search, generic scraping, structured Web Data APIs for major sites, and a remote scraping browser
• Free tier of 5,000 requests/month and granular control via GROUPS and TOOLS env vars

• Beyond the free tier, usage is billed against Bright Data zones, which can get expensive at scale
• Pro mode exposes 60+ tools, which can overwhelm an LLM's tool-selection if not narrowed with GROUPS or TOOLS
• Some web_data_* extractors are asynchronous and can take many seconds (up to the 600s POLLING_TIMEOUT) to return

• Firecrawl MCP for scrape, crawl, and structured extraction
• Apify MCP for running 4,000+ Apify Actors as tools
• Exa MCP for neural web search and content retrieval