ScrapeGraph MCP Server

A production-ready Model Context Protocol (MCP) server that provides seamless integration with the ScrapeGraph AI API. This server enables language models to leverage advanced AI-powered web scraping capabilities with enterprise-grade reliability.

Table of Contents

• Key Features
• Quick Start
• Available Tools
• Setup Instructions
• Remote Server Usage
• Local Usage
• Google ADK Integration
• Example Use Cases
• Error Handling
• Common Issues
• Development
• Contributing
• Documentation
• Technology Stack
• License

API v2

This MCP server targets ScrapeGraph API v2 (https://v2-api.scrapegraphai.com/api), aligned 1:1 with scrapegraph-py PR #84. Auth uses the SGAI-APIKEY header. Environment variables mirror the Python SDK:

• SGAI_API_URL — override the base URL (default https://v2-api.scrapegraphai.com/api)
• SGAI_TIMEOUT — request timeout in seconds (default 120)
• SGAI_API_KEY — API key (can also be passed via MCP scrapegraphApiKey or X-API-Key header)

Legacy aliases (still honored): SCRAPEGRAPH_API_BASE_URL for SGAI_API_URL, SGAI_TIMEOUT_S for SGAI_TIMEOUT.

Key Features

• Scrape & extract: scrape (POST /scrape, multi-format), extract (POST /extract, URL + prompt)
• Search: search (POST /search; num_results clamped 3–20)
• Crawl: Async multi-page crawl with crawl_start / crawl_get_status / crawl_stop / crawl_resume
• Schema: schema (POST /schema) — generate or augment a JSON Schema from a prompt
• Monitors: Scheduled jobs via monitor_create, monitor_list, monitor_get, pause/resume/delete, monitor_activity (paginated tick history)
• Account: credits, history
• Easy integration: Claude Desktop, Cursor, Smithery, HTTP transport
• Developer docs: .agent/ folder

Migration: v2 → v3

v3 renames every MCP tool that diverged from the v2 API docs. Hard rename, no aliases.

v2 (old)	v3 (new)
smartscraper	extract
searchscraper	search
smartcrawler_initiate	crawl_start
smartcrawler_fetch_results	crawl_get_status
sgai_history	history
generate_schema	schema
markdownify	removed — use scrape with output_format="markdown"

Quick Start

1. Get Your API Key

Sign up and get your API key from the ScrapeGraph Dashboard

2. Install with Smithery (Recommended)

npx -y @smithery/cli install @ScrapeGraphAI/scrapegraph-mcp --client claude

3. Start Using

Ask Claude or Cursor:

• "Convert https://scrapegraphai.com to markdown"
• "Extract all product prices from this e-commerce page"
• "Research the latest AI developments and summarize findings"

That's it! The server is now available to your AI assistant.

Available Tools

Tool	Role
scrape	POST /scrape (output_format: markdown, html, screenshot, branding, links, images, summary)
extract	POST /extract (requires website_url + user_prompt; optional output_schema)
search	POST /search (num_results 1–20; supports country_search, time_range, output_schema)
crawl_start	POST /crawl — extraction_mode markdown / html / links / images / summary / branding / screenshot
crawl_get_status	GET /crawl/:id (poll until status: completed)
crawl_stop, crawl_resume	POST /crawl/:id/stop | resume
schema	POST /schema (generate or augment a JSON Schema from a prompt)
credits	GET /credits
history	GET /history (paginated, service filter)
monitor_create, monitor_list, monitor_get, monitor_pause, monitor_resume, monitor_delete	/monitor API
monitor_activity	GET /monitor/:id/activity (paginated tick history: id, createdAt, status, changed, elapsedMs, diffs)

Removed: sitemap, agentic_scrapper, async-status polling, and (in v3) markdownify — use scrape with output_format="markdown".

Setup Instructions

To utilize this server, you'll need a ScrapeGraph API key. Follow these steps to obtain one:

1. Navigate to the ScrapeGraph Dashboard
2. Create an account and generate your API key

Automated Installation via Smithery

For automated installation of the ScrapeGraph API Integration Server using Smithery:

npx -y @smithery/cli install @ScrapeGraphAI/scrapegraph-mcp --client claude

Claude Desktop Configuration

Update your Claude Desktop configuration file with the following settings (located on the top rigth of the Cursor page):

(remember to add your API key inside the config)

{
    "mcpServers": {
        "@ScrapeGraphAI-scrapegraph-mcp": {
            "command": "npx",
            "args": [
                "-y",
                "@smithery/cli@latest",
                "run",
                "@ScrapeGraphAI/scrapegraph-mcp",
                "--config",
                "\"{\\\"scrapegraphApiKey\\\":\\\"YOUR-SGAI-API-KEY\\\"}\""
            ]
        }
    }
}

The configuration file is located at:

• Windows: %APPDATA%/Claude/claude_desktop_config.json
• macOS: ~/Library/Application\ Support/Claude/claude_desktop_config.json

Cursor Integration

Add the ScrapeGraphAI MCP server on the settings:

Remote Server Usage

Connect to our hosted MCP server - no local installation required!

Claude Desktop Configuration (Remote)

Add this to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "scrapegraph-mcp": {
      "command": "npx",
      "args": [
        "mcp-remote@0.1.25",
        "https://mcp.scrapegraphai.com/mcp",
        "--header",
        "X-API-Key:YOUR_API_KEY"
      ]
    }
  }
}

Cursor Configuration (Remote)

Cursor supports native HTTP MCP connections. Add to your Cursor MCP settings (~/.cursor/mcp.json):

{
  "mcpServers": {
    "scrapegraph-mcp": {
      "url": "https://mcp.scrapegraphai.com/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Benefits of Remote Server

• No local setup - Just configure and start using
• Always up-to-date - Automatically receives latest updates
• Cross-platform - Works on any OS with Node.js

Local Usage

To run the MCP server locally for development or testing, follow these steps:

Prerequisites

• Python 3.13 or higher
• pip or uv package manager
• ScrapeGraph API key

Installation

1. Clone the repository (if you haven't already):

git clone https://github.com/ScrapeGraphAI/scrapegraph-mcp
cd scrapegraph-mcp

2. Install the package:

# Using pip
pip install -e .

# Or using uv (faster)
uv pip install -e .

3. Set your API key:

# macOS/Linux
export SGAI_API_KEY=your-api-key-here

# Windows (PowerShell)
$env:SGAI_API_KEY="your-api-key-here"

# Windows (CMD)
set SGAI_API_KEY=your-api-key-here

Running the Server Locally

You can run the server directly:

# Using the installed command
scrapegraph-mcp

# Or using Python module
python -m scrapegraph_mcp.server

The server will start and communicate via stdio (standard input/output), which is the standard MCP transport method.

Testing with MCP Inspector

Test your local server using the MCP Inspector tool:

npx @modelcontextprotocol/inspector python -m scrapegraph_mcp.server

This provides a web interface to test all available tools interactively.

Configuring Claude Desktop for Local Server

To use your locally running server with Claude Desktop, update your configuration file:

macOS/Linux (~/Library/Application Support/Claude/claude_desktop_config.json):

{
    "mcpServers": {
        "scrapegraph-mcp-local": {
            "command": "python",
            "args": [
                "-m",
                "scrapegraph_mcp.server"
            ],
            "env": {
                "SGAI_API_KEY": "your-api-key-here"
            }
        }
    }
}

Windows (%APPDATA%\Claude\claude_desktop_config.json):

{
    "mcpServers": {
        "scrapegraph-mcp-local": {
            "command": "python",
            "args": [
                "-m",
                "scrapegraph_mcp.server"
            ],
            "env": {
                "SGAI_API_KEY": "your-api-key-here"
            }
        }
    }
}

Note: Make sure Python is in your PATH. You can verify by running python --version in your terminal.

Configuring Cursor for Local Server

In Cursor's MCP settings, add a new server with:

• Command: python
• Args: ["-m", "scrapegraph_mcp.server"]
• Environment Variables: {"SGAI_API_KEY": "your-api-key-here"}

Troubleshooting Local Setup

Server not starting:

• Verify Python is installed: python --version
• Check that the package is installed: pip list | grep scrapegraph-mcp
• Ensure API key is set: echo $SGAI_API_KEY (macOS/Linux) or echo %SGAI_API_KEY% (Windows)

Tools not appearing:

• Check Claude Desktop logs:
  • macOS: ~/Library/Logs/Claude/
  • Windows: %APPDATA%\Claude\Logs\
• Verify the server starts without errors when run directly
• Check that the configuration JSON is valid

Import errors:

• Reinstall the package: pip install -e . --force-reinstall
• Verify dependencies: pip install -r requirements.txt (if available)

Google ADK Integration

The ScrapeGraph MCP server can be integrated with Google ADK (Agent Development Kit) to create AI agents with web scraping capabilities.

Prerequisites

• Python 3.13 or higher
• Google ADK installed
• ScrapeGraph API key

Installation

1. Install Google ADK (if not already installed):

pip install google-adk

2. Set your API key:

export SGAI_API_KEY=your-api-key-here

Basic Integration Example

Create an agent file (e.g., agent.py) with the following configuration:

import os
from google.adk.agents import LlmAgent
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from mcp import StdioServerParameters

# Path to the scrapegraph-mcp server directory
SCRAPEGRAPH_MCP_PATH = "/path/to/scrapegraph-mcp"

# Path to the server.py file
SERVER_SCRIPT_PATH = os.path.join(
    SCRAPEGRAPH_MCP_PATH, 
    "src", 
    "scrapegraph_mcp", 
    "server.py"
)

root_agent = LlmAgent(
    model='gemini-2.0-flash',
    name='scrapegraph_assistant_agent',
    instruction='Help the user with web scraping and data extraction using ScrapeGraph AI. '
                'You can convert webpages to markdown, extract structured data using AI, '
                'perform web searches, crawl multiple pages, and automate complex scraping workflows.',
    tools=[
        MCPToolset(
            connection_params=StdioConnectionParams(
                server_params=StdioServerParameters(
                    command='python3',
                    args=[
                        SERVER_SCRIPT_PATH,
                    ],
                    env={
                        'SGAI_API_KEY': os.getenv('SGAI_API_KEY'),
                    },
                ),
                timeout=300.0,)
            ),
            # Optional: Filter which tools from the MCP server are exposed
            # tool_filter=['scrape', 'extract', 'search']
        )
    ],
)

Configuration Options

Timeout Settings:

• Default timeout is 5 seconds, which may be too short for web scraping operations
• Recommended: Set `timeout=300.0
• Adjust based on your use case (crawling operations may need even longer timeouts)

Tool Filtering:

• By default, all registered MCP tools are exposed to the agent (see Available Tools)
• Use tool_filter to limit which tools are available:

  tool_filter=['scrape', 'extract', 'search']
  

API Key Configuration:

• Set via environment variable: export SGAI_API_KEY=your-key
• Or pass directly in env dict: 'SGAI_API_KEY': 'your-key-here'
• Environment variable approach is recommended for security

Usage Example

Once configured, your agent can use natural language to interact with web scraping tools:

# The agent can now handle queries like:
# - "Convert https://example.com to markdown"
# - "Extract all product prices from this e-commerce page"
# - "Search for recent AI research papers and summarize them"
# - "Crawl this documentation site and extract all API endpoints"

For more information about Google ADK, visit the official documentation.

Example Use Cases

The server enables sophisticated queries across various scraping scenarios:

Single Page Scraping

• Markdownify: "Convert the ScrapeGraph documentation page to markdown"
• Extract: "Extract all product names, prices, and ratings from this e-commerce page"
• Extract with scrolling: "Scrape this infinite scroll page with 5 scrolls and extract all items"
• Basic Scrape: "Fetch the HTML content of this JavaScript-heavy page with full rendering"

Search and Research

• Search: "Research and summarize recent developments in AI-powered web scraping"
• Search: "Search for the top 5 articles about machine learning frameworks and extract key insights"
• Search: "Find recent news about GPT-4 and provide a structured summary"
• Search: v2 does not apply time_range; phrase queries to bias recency in natural language instead

Website analysis

• Use crawl_start plus crawl_get_status to map and capture multi-page content; there is no separate sitemap tool on v2.

Multi-page crawling

• Crawl: "Crawl the blog in markdown mode and poll until complete"
• For structured fields per page, run extract on individual URLs (or monitor_create on a schedule)

Monitors and account

• Monitor: "Run this extract prompt on https://example.com every day at 9am" (monitor_create with interval)
• Credits / history: credits, history
• Agentic Scraper: "Execute a complex workflow: login, navigate to reports, download data, and extract summary statistics"

Error Handling

The server implements robust error handling with detailed, actionable error messages for:

• API authentication issues
• Malformed URL structures
• Network connectivity failures
• Rate limiting and quota management

Common Issues

Windows-Specific Connection

When running on Windows systems, you may need to use the following command to connect to the MCP server:

C:\Windows\System32\cmd.exe /c npx -y @smithery/cli@latest run @ScrapeGraphAI/scrapegraph-mcp --config "{\"scrapegraphApiKey\":\"YOUR-SGAI-API-KEY\"}"

This ensures proper execution in the Windows environment.

Other Common Issues

"ScrapeGraph client not initialized"

• Cause: Missing API key
• Solution: Set SGAI_API_KEY environment variable or provide via --config

"Error 401: Unauthorized"

• Cause: Invalid API key
• Solution: Verify your API key at the ScrapeGraph Dashboard

"Error 402: Payment Required"

• Cause: Insufficient credits
• Solution: Add credits to your ScrapeGraph account

Crawl not returning results

• Cause: Still processing (asynchronous operation)
• Solution: Keep polling crawl_get_status() until status is "completed"

Tools not appearing in Claude Desktop

• Cause: Server not starting or configuration error
• Solution: Check Claude logs at ~/Library/Logs/Claude/ (macOS) or %APPDATA%\Claude\Logs\ (Windows)

For detailed troubleshooting, see the .agent documentation.

Development

Prerequisites

• Python 3.13 or higher
• pip or uv package manager
• ScrapeGraph API key

Installation from Source

# Clone the repository
git clone https://github.com/ScrapeGraphAI/scrapegraph-mcp
cd scrapegraph-mcp

# Install dependencies
pip install -e ".[dev]"

# Set your API key
export SGAI_API_KEY=your-api-key

# Run the server
scrapegraph-mcp
# or
python -m scrapegraph_mcp.server

Testing with MCP Inspector

Test your server locally using the MCP Inspector tool:

npx @modelcontextprotocol/inspector scrapegraph-mcp

This provides a web interface to test all available tools.

Code Quality

Linting:

ruff check src/

Type Checking:

mypy src/

Format Checking:

ruff format --check src/

Project Structure

scrapegraph-mcp/
├── src/
│   └── scrapegraph_mcp/
│       ├── __init__.py      # Package initialization
│       └── server.py        # Main MCP server (all code in one file)
├── .agent/                  # Developer documentation
│   ├── README.md           # Documentation index
│   └── system/             # System architecture docs
├── assets/                  # Images and badges
├── pyproject.toml          # Project metadata & dependencies
├── smithery.yaml           # Smithery deployment config
└── README.md               # This file

Contributing

We welcome contributions! Here's how you can help:

Adding a New Tool

1. Add method to ScapeGraphClient class in server.py:

def new_tool(self, param: str) -> Dict[str, Any]:
    """Tool description."""
    url = f"{self.BASE_URL}/new-endpoint"
    data = {"param": param}
    response = self.client.post(url, headers=self.headers, json=data)
    if response.status_code != 200:
        raise Exception(f"Error {response.status_code}: {response.text}")
    return response.json()

2. Add MCP tool decorator:

@mcp.tool()
def new_tool(param: str) -> Dict[str, Any]:
    """
    Tool description for AI assistants.

    Args:
        param: Parameter description

    Returns:
        Dictionary containing results
    """
    if scrapegraph_client is None:
        return {"error": "ScrapeGraph client not initialized. Please provide an API key."}

    try:
        return scrapegraph_client.new_tool(param)
    except Exception as e:
        return {"error": str(e)}

3. Test with MCP Inspector:

npx @modelcontextprotocol/inspector scrapegraph-mcp

4. Update documentation:

   • Add tool to this README
   • Update .agent documentation

5. Submit a pull request

Development Workflow

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Run linting and type checking
5. Test with MCP Inspector and Claude Desktop
6. Update documentation
7. Commit your changes (git commit -m 'Add amazing feature')
8. Push to the branch (git push origin feature/amazing-feature)
9. Open a Pull Request

Code Style

• Line length: 100 characters
• Type hints: Required for all functions
• Docstrings: Google-style docstrings
• Error handling: Return error dicts, don't raise exceptions in tools
• Python version: Target 3.13+

For detailed development guidelines, see the .agent documentation.

Documentation

For comprehensive developer documentation, see:

• .agent/README.md - Complete developer documentation index
• .agent/system/project_architecture.md - System architecture and design
• .agent/system/mcp_protocol.md - MCP protocol integration details

Technology Stack

Core Framework

• Python 3.13+ - Modern Python with type hints
• FastMCP - Lightweight MCP server framework
• httpx 0.24.0+ - Modern async HTTP client

Development Tools

• Ruff - Fast Python linter and formatter
• mypy - Static type checker
• Hatchling - Modern build backend

Deployment

• Smithery - Automated MCP server deployment
• Docker - Container support with Alpine Linux
• stdio transport - Standard MCP communication

API Integration

• ScrapeGraph AI API - Enterprise web scraping service
• Base URL: https://v2-api.scrapegraphai.com/api
• Authentication: API key-based

License

This project is distributed under the MIT License. For detailed terms and conditions, please refer to the LICENSE file.

Acknowledgments

Special thanks to tomekkorbak for his implementation of oura-mcp-server, which served as starting point for this repo.

Resources

Official Links

• ScrapeGraph AI Homepage
• ScrapeGraph Dashboard - Get your API key
• ScrapeGraph API Documentation
• GitHub Repository

MCP Resources

• Model Context Protocol - Official MCP specification
• FastMCP Framework - Framework used by this server
• MCP Inspector - Testing tool
• Smithery - MCP server distribution
• mcp-name: io.github.ScrapeGraphAI/scrapegraph-mcp

AI Assistant Integration

• Claude Desktop - Desktop app with MCP support
• Cursor - AI-powered code editor

Support

• GitHub Issues - Report bugs or request features
• Developer Documentation - Comprehensive dev docs

---

Made with ❤️ by ScrapeGraphAI Team