Glowing Web Network
Glowing Web Network
MCP Server

Give your AI assistant the web with the Scrappey MCP server

The Scrappey MCP server exposes web data collection as Model Context Protocol tools, so MCP-compatible clients like Claude Desktop, Claude Code, Cursor, and Windsurf can fetch live pages, run managed browser sessions, and return clean HTML or Markdown. Add one server entry to your client config and your agent can read JavaScript-heavy and modern websites on demand.

Quick start

No install step needed in most clients — npx fetches and runs the server on demand. Set SCRAPPEY_API_KEY in the server's env block.

Run via npx
npx -y mcp-server-scrappey
  1. 1

    Get your Scrappey API key

    Create an account and copy your API key from the dashboard. You start with 150 free requests, no subscription and no card required.

  2. 2

    Add the server to your MCP client config

    Add a scrappey entry to your client's MCP config JSON (for Claude Desktop this is claude_desktop_config.json). Use npx so the server is fetched and run automatically, and pass your key via the env block.

    json
    {
      "mcpServers": {
        "scrappey": {
          "command": "npx",
          "args": ["-y", "mcp-server-scrappey"],
          "env": {
            "SCRAPPEY_API_KEY": "YOUR_API_KEY"
          }
        }
      }
    }
  3. 3

    Restart your client and call a tool

    Restart the MCP client so it picks up the new server. The Scrappey tools (request, session create/destroy, browser action, screenshot) become available to the model. Just ask it to fetch a page.

    text
    Ask your assistant:
    "Use Scrappey to fetch https://example.com and return the main content as Markdown."

Code examples

Full MCP config with a managed session
{
  "mcpServers": {
    "scrappey": {
      "command": "npx",
      "args": ["-y", "mcp-server-scrappey"],
      "env": {
        "SCRAPPEY_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

// Once connected, the model has access to tools such as:
//   scrappey_request          - fetch a URL (HTML or Markdown)
//   scrappey_create_session   - open a persistent browser session
//   scrappey_browser_action   - click, type, scroll, run JS
//   scrappey_screenshot       - capture a page image
//   scrappey_destroy_session  - release the session
//
// Example prompt to the assistant:
//   "Create a Scrappey session, open the product page,
//    read the price and title, then close the session."
The underlying Scrappey API call the tools wrap
# The MCP server calls the same Scrappey HTTP API you can hit directly.
# Useful for debugging or building your own tool outside MCP.

curl -X POST "https://publisher.scrappey.com/api/v1?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "cmd": "request.get",
    "url": "https://example.com",
    "markdownResponse": true
  }'

# markdownResponse: true returns LLM-ready Markdown instead of raw HTML.
# Residential proxies and web access handling are included automatically.

Why connect Scrappey over MCP

Drop-in for any MCP client

Works with Claude Desktop, Claude Code, Cursor, Windsurf, and any Model Context Protocol client. Add one JSON block and the tools appear.

Clean HTML or LLM-ready Markdown

Return pages as structured Markdown so the model gets readable content instead of raw markup and scripts.

Handles modern, JavaScript-heavy sites

Full-browser rendering and automatic web access handling mean dynamic pages load and resolve before content is returned.

Managed browser sessions

Create, reuse, and destroy persistent sessions so multi-step flows like navigation and form input keep their state.

Browser actions and screenshots

Beyond fetching, the agent can click, type, scroll, run JavaScript, and capture screenshots through dedicated tools.

Pay only for what works

Pay-as-you-go with residential proxies included. Failed requests are free, so your agent's retries do not cost extra.

Popular use cases

Research agents

Let an assistant pull live data from public pages mid-conversation, then summarize or compare it without you copy-pasting URLs.

Coding agents that read docs

Give Claude Code or Cursor the ability to fetch external documentation and reference pages while writing code.

Multi-step page automation

Use managed sessions to log into a public flow, navigate, fill fields, and extract the resulting content in one task.

Monitoring and extraction

Ask the agent to grab a page on demand and return just the fields you care about as clean Markdown.

MCP Server FAQ

Which MCP clients does this work with?

Any Model Context Protocol client. That includes Claude Desktop, Claude Code, Cursor, and Windsurf. You add the same scrappey server entry to each client's MCP config JSON.

How do I install the server?

There is usually nothing to install. Configure the command as npx with args ["-y", "mcp-server-scrappey"], and the client fetches and runs the server automatically on startup.

Where do I put my API key?

In the env block of the server config, as SCRAPPEY_API_KEY. The server reads it at launch and uses it for every tool call, so you never paste the key into prompts.

What does it cost to run?

Scrappey is pay-as-you-go with no subscription. You get 150 free requests, then pay €0.20 per 1,000 direct HTTP requests or €1.00 per 1,000 full-browser requests. Residential proxies are included and failed requests are free.

What tools does the server expose?

Tools for sending requests, creating and destroying managed browser sessions, listing and checking session status, running browser actions like clicking and typing, and capturing screenshots.

Can it return Markdown instead of HTML?

Yes. The request tool can return LLM-ready Markdown, which maps to adding markdownResponse: true on the underlying Scrappey API call. That keeps content compact and readable for the model.
footer-frame

Start building with Scrappey

Try It For Free. No Subscription Required. No Credit Card Required. Instant Set-Up. 150 Free Requests Are Waiting For You!