brightdata/brightdata-mcp

What it solves

AI agents often fail when trying to fetch web pages that block bots, require CAPTCHA solving, or have geo-restrictions. Bright Data MCP solves this by giving agents a single interface to search the web, scrape any URL as Markdown or HTML, pull structured JSON from 1,200+ pre-built scrapers for sites like Amazon, LinkedIn, TikTok, npm, and more, and control a remote browser — with bot detection, CAPTCHAs, rate limits, and proxy rotation handled automatically.

How it works

The server exposes 32 tools over the Model Context Protocol, so any MCP-compatible client (Claude, Cursor, VS Code, etc.) can call them. You can use a hosted remote server by adding a URL with your API token, or run a local instance via npx @brightdata/mcp. Every request goes through Bright Data's unblocking infrastructure, meaning pages that typically block ordinary HTTP clients return normally. Tools are grouped (search/scraping, browser automation, GEO/LLM insights, advanced scraping) so you load only what you need and keep context small. For platform-specific data, the flow is search_scrapers -> get_scraper_details -> run_scraper to get clean JSON from pre-built scrapers. Browser automation provides a remote session with navigation, clicking, typing, form filling, screenshots, and more. There's also a 5,000-request/month free tier.

Who it's for

Developers building AI agents, coding assistants, and chat applications that need reliable, live web data. It's aimed at anyone who wants to give LLMs web search, page reading, structured data extraction, or browser control without writing scraping infrastructure themselves. Use cases include real-time research, e-commerce price comparison, market and competitor analysis, social media monitoring, and Generative Engine Optimization (tracking how ChatGPT/Perplexity rank your brand).

Highlights

• 32 tools covering web search (Google, Bing, Yandex), page scraping (Markdown/HTML), structured data extraction, and remote browser automation
• 1,200+ pre-built scrapers for major platforms (Amazon, LinkedIn, TikTok, YouTube, Reddit, npm, PyPI, etc.) returning clean JSON
• Built-in unblocking: bot detection bypass, CAPTCHA solving, proxy rotation, geo-targeting on every request
• Batch operations: up to 10 searches or scrapes per call
• Hosted remote server (no install) or local npx server
• Tool groups to load only what you need, reducing context
• Free tier: 5,000 requests per month, no credit card required