io.github.ZenRows/zenrows-mcp

Model Context Protocol server for the ZenRows Universal Scraper API.
Give any MCP-compatible AI assistant the ability to scrape any webpage — including JavaScript-rendered content and anti-bot protected sites.

Quick Start

Claude Code

claude mcp add zenrows -e ZENROWS_API_KEY=YOUR_API_KEY -- npx -y @zenrows/mcp

Or ask your AI assistant naturally once configured:

Scrape https://example.com and summarize the content.

Tool

scrape

Fetches a webpage and returns its content as clean markdown (default), plaintext, raw HTML, PDF, structured JSON, or a screenshot. See the ZenRows API docs for full parameter reference.

| Parameter | Type | Default | Description | | --------------------- | -------------------------------------------- | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | url | string | required | Webpage URL to scrape | | js_render | boolean | false | Enable JS rendering for SPAs and dynamic content | | premium_proxy | boolean | false | Use residential proxies to bypass anti-bot systems | | proxy_country | string | — | ISO 3166-1 alpha-2 country code (e.g. US, GB). Requires premium_proxy | | response_type | markdown | plaintext | pdf | html | markdown | Output format. html returns raw source (ZenRows default when no param is sent). Ignored when autoparse, css_extractor, outputs, or screenshot params are set | | autoparse | boolean | — | Auto-extract structured JSON from the page | | css_extractor | string | — | JSON map of CSS selectors: {"title":"h1","price":".price"} | | outputs | string | — | Comma-separated data types to extract as JSON: emails, headings, links, menus, images, videos, audios. Use * for all | | screenshot | boolean | — | Capture an above-the-fold screenshot. Returns an im

... View full README on GitHub