ScraperAPI MCP server

The ScraperAPI MCP server enables LLM clients to retrieve and process web scraping requests using the ScraperAPI services.

This is the self-hosted (local) server. A hosted (remote) version is also available.

Table of Contents

• Features
• Architecture
• Installation
• Available tools
• Configuration
  • Settings
  • Client Setup
• Development

Features

• Full implementation of the Model Context Protocol specification
• Seamless integration with ScraperAPI for web scraping
• Simple setup with Python or Docker

Architecture

          ┌───────────────┐     ┌───────────────────────┐     ┌───────────────┐
          │  LLM Client   │────▶│  Scraper MCP Server   │────▶│    AI Model   │
          └───────────────┘     └───────────────────────┘     └───────────────┘
                                            │
                                            ▼
                                  ┌──────────────────┐
                                  │  ScraperAPI API  │
                                  └──────────────────┘

Installation

The ScraperAPI MCP Server is designed to run as a local server on your machine, your LLM client will launch it automatically when configured.

Prerequisites

• Python 3.11+
• Docker (optional)

Using Python

Install the package:

pip install scraperapi-mcp-server

Add this to your client configuration file:

{
  "mcpServers": {
    "ScraperAPI": {
      "command": "python",
      "args": ["-m", "scraperapi_mcp_server"],
      "env": {
        "API_KEY": "<YOUR_SCRAPERAPI_API_KEY>"
      }
    }
  }
}

Using Docker

Add this to your client configuration file:

{
  "mcpServers": {
    "ScraperAPI": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "-e",
        "API_KEY=${API_KEY}",
        "--rm",
        "scraperapi-mcp-server"]
    }
  }
}

[!TIP]

If your command is not working (for example, you see a package not found error when trying to start the server), double-check the path you are using. To find the correct path, activate your virtual environment first, then run:

bash

Copy

which <YOUR_COMMAND>

Available tools

API

SDEs

Structured Data Endpoints (SDEs) return pre-parsed JSON (or CSV) instead of raw HTML. Every SDE tool also accepts output_format (json by default, or csv), tld, and country_code, shown in each tool's table.

Google

Amazon