🌐 opc-web-scrape-mcp

MCP Server wrapping Playwright — web scraping, screenshots, and link extraction for AI agents

Overview

opc-web-scrape-mcp leverages Playwright (69K+ stars) to provide AI agents with headless browser capabilities: scrape page text content, capture full-page screenshots, and extract structured link data.

Tools

| Tool | Description | |------|-------------| | scrape_page | Extract text content from any web page with CSS selector support | | take_screenshot | Capture full-page or viewport screenshots in PNG | | extract_links | Extract all links from a page with optional regex filtering |

Prerequisites

npm install playwright
npx playwright install chromium

Installation

npm install opc-web-scrape-mcp

Usage

As an MCP server

{
  "mcpServers": {
    "web-scrape": {
      "command": "npx",
      "args": ["opc-web-scrape-mcp"]
    }
  }
}

Example: scrape_page

Input:  { "url": "https://example.com", "selector": "main" }
Output: Full text content of the <main> element

Example: take_screenshot

Input:  { "url": "https://example.com", "full_page": true, "width": 1440, "height": 900 }
Output: Screenshot saved: ./screenshots/screenshot_1700000000.png

Example: extract_links

Input:  { "url": "https://github.com/trending", "filter": "github.com/[^/]+/[^/]+$" }
Output: [{ href: "https://github.com/user/repo", text: "repo description" }, ...]

Features

• Headless Chromium — Real browser rendering, JavaScript execution, SPAs
• CSS selector targeting — Extract specific elements or full page content
• Full-page screenshots — Capture scrollable content beyond the viewport
• Link extraction — Structured link data with optional regex filtering
• Network idle wait — Waits for dynamic content to load before extraction

License

MIT