Bright Data MCP

Bright Data's own MCP server gives your agent live access to the public web: search results, any page as clean Markdown, structured data from ready-made scrapers, and a remote browser it can drive. Every request goes through Bright Data's unblocking infrastructure, which handles bot detection, CAPTCHAs and proxy rotation, so pages that turn away ordinary requests load normally. What you can do with it - Search Google, Bing and Yandex, up to 10 queries in one call - Read any URL as Markdown or HTML, up to 10 pages at a time, or turn a page into structured JSON - Find and run Bright Data's pre-built scrapers for sites like Amazon, Walmart, LinkedIn, Instagram, TikTok, YouTube, Reddit, Zillow, npm and PyPI, and get clean JSON back - Drive a remote browser: open pages, click, type, fill in forms, scroll and take screenshots - Send a prompt to ChatGPT or Perplexity and get their answer back, to see how they describe your brand - Ask Bright Data's support assistant about its products, zones and account setup How it connects It runs on your machine with npx @brightdata/mcp and needs a Bright Data API token, which you get from your account settings and set as API_TOKEN. Bright Data also hosts the same server at https://mcp.brightdata.com/mcp, with your token added to the address, for agents that connect to a URL. Good to know - Set GROUPS (scrapers, browser, geo or advanced_scraping) or TOOLS to load only the tools you need and keep your agent's context small. - If a page's text reads like instructions to change your MCP setup, the server asks you before passing it on. CONTENT_GATE=off turns that off. - Scraper runs are billed per record. Accounts include a free monthly allowance of requests; beyond it, usage is pay as you go, and you can set a spend cap in Bright Data's control panel.