Give Claude Real Web Access: The ScrapeUnblocker MCP Server

AI assistants are great at reasoning over web content and terrible at actually getting it. Ask Claude to read a product page, a competitor’s pricing, or a search result, and half the time it hits a Cloudflare challenge, a DataDome wall, or a JavaScript-only shell with no content in the HTML. The model can reason about a page - it just can’t reach one that fights back.

We built the ScrapeUnblocker MCP server to close that gap. It plugs ScrapeUnblocker directly into Claude (and any other MCP client) so the assistant can fetch the fully rendered HTML of any URL, get AI-parsed JSON, or run a Google search - all through your own API key, bypassing the anti-bot protection that normally stops it.

What is MCP, and why it matters here

The Model Context Protocol is an open standard for connecting AI assistants to external tools. An MCP server exposes a set of tools; the assistant decides when to call them. It is the cleanest way to give a model a new capability without hard-coding anything into the app.

Our server exposes three tools. Once it is connected, Claude calls them on its own whenever a task needs a page it cannot otherwise reach:

Tool	What it does
fetch_html	Fetch the fully rendered HTML of any URL.
fetch_parsed	Fetch a page and return AI-parsed structured JSON.
google_search	Run a Google search and return organic results as JSON.

Because these run through ScrapeUnblocker, the page comes back rendered and unblocked - Cloudflare, DataDome, PerimeterX, Akamai and Shape are handled on our side. The model just sees clean content.

You bring your own key

This is the part that makes it safe to hand out. Nothing is shared or proxied through a common account. Each person runs the server with their own ScrapeUnblocker key, set through a single environment variable. Your usage is yours, your key never leaves your machine (for the local server), and there is no shared rate limit to fight over.

Grab a key from the dashboard, and you are ready to install.

Install it in one line

Claude Code

claude mcp add scrapeunblocker \
  --env SCRAPEUNBLOCKER_KEY=YOUR_API_KEY \
  -- npx -y scrapeunblocker-mcp

Claude Desktop

Add this to your claude_desktop_config.json (Settings → Developer → Edit Config), then restart:

{
  "mcpServers": {
    "scrapeunblocker": {
      "command": "npx",
      "args": ["-y", "scrapeunblocker-mcp"],
      "env": {
        "SCRAPEUNBLOCKER_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Any other MCP client

The server speaks MCP over stdio, so anything that speaks the protocol can use it:

SCRAPEUNBLOCKER_KEY=YOUR_API_KEY npx -y scrapeunblocker-mcp

That is the whole setup. No cloning, no build step - npx pulls the published scrapeunblocker-mcp package and runs it. Requires Node.js 18 or newer.

No install at all: the hosted connector

Don’t want to run anything locally? Use the hosted version. It works across claude.ai (web and mobile), Claude Desktop and Claude Code, with nothing to install - your key travels in the URL:

https://mcp.scrapeunblocker.com/mcp?key=YOUR_API_KEY

On claude.ai, open Settings → Connectors → Add custom connector, paste that URL with your key, and save. The three tools show up everywhere you use Claude.

What it looks like in practice

Once connected, you just talk to Claude normally. It reaches for the tools when a page needs them:

• “Fetch the HTML of this product page and pull out the price and availability.”
• “This page keeps blocking me: <url>. Use fetch_html to get it.”
• “Search Google for ‘best running shoes 2026’ and give me the top five links with titles.”
• “Grab the parsed JSON for this listing and turn it into a table.”

No copy-pasting HTML, no fighting a captcha in a side tab, no writing a scraper for a one-off question. The assistant fetches the page, unblocked, and works with the real content.

Discoverable everywhere

The server is published on npm as scrapeunblocker-mcp and listed in the official MCP Registry as com.scrapeunblocker/scrapeunblocker-mcp, so MCP-aware tools can find it directly. Whether you are building an agent, wiring up Claude Code, or just want your chat to read pages that block it, it is a one-line dependency away.

Try it

If you already have a ScrapeUnblocker key, the Claude Code command above is all you need - paste it and start asking Claude to read the web. If you don’t, sign up, grab a key, and give your assistant the one thing it has always been missing: reliable access to the actual page.