Website MCP Server

Give your AI full access to any website's design system — colors, fonts, CSS, images, code, and structure. No browser needed.

What it does

This MCP server downloads a complete website and makes every file queryable through 6 AI tools. Your AI can extract color palettes, read CSS files, find font stacks, detect frameworks, search for class names, and parse page structure — all from a single URL.

How it works

You: "Analyze the design of stripe.com"

AI → scrape_website({ url: "https://stripe.com" })
   ← Downloaded 87 files (HTML, CSS, JS, images, fonts)

AI → get_overview({ siteId: "a1b2c3" })
   ← Colors: #635bff, #0a2540, #ffffff
   ← Fonts: Inter, system-ui
   ← Frameworks: Next.js, Tailwind CSS

AI → search_content({ siteId: "a1b2c3", query: "font-family", fileType: "css" })
   ← 12 matches across 3 files

AI → get_file({ siteId: "a1b2c3", path: "stripe.com/css/main.css" })
   ← [full CSS content ready for your AI to use]

Tools

Tool	What it does
scrape_website	Download an entire website — HTML, CSS, JS, images, fonts. Returns a site ID for all other tools.
list_files	List all downloaded files. Filter by type: html, css, js, images, fonts, or all.
get_file	Read the full content of any file — CSS, HTML, JS, config files. Binary files return metadata.
get_page	Parse an HTML page: title, headings, links, images, stylesheets, scripts. Defaults to index page.
search_content	Search across all files — like grep for the whole website. Find colors, class names, specific text.
get_overview	Instant design analysis: color palette, font stack, detected frameworks, file counts.

Input Parameters

scrape_website

Parameter	Type	Required	Description
url	string	Yes	The website URL to download
depth	number	No	How many levels deep to crawl (1-5, default: 3)

list_files

Parameter	Type	Required	Description
siteId	string	Yes	Site ID from scrape_website
type	string	No	Filter: html, css, js, images, fonts, all (default: all)

get_file

Parameter	Type	Required	Description
siteId	string	Yes	Site ID from scrape_website
path	string	Yes	File path from list_files output

get_page

Parameter	Type	Required	Description
siteId	string	Yes	Site ID from scrape_website
url	string	No	Specific page URL (default: index page)

search_content

Parameter	Type	Required	Description
siteId	string	Yes	Site ID from scrape_website
query	string	Yes	Search term (text, regex supported)
fileType	string	No	Limit search to file type: css, html, js, etc.

get_overview

Parameter	Type	Required	Description
siteId	string	Yes	Site ID from scrape_website

Use Cases

For Designers

Extract the complete design system from any website. Get the exact color palette, font stack, and spacing — no more guessing from screenshots.

For Agencies

Run competitive analysis across multiple client competitors. Compare design systems, frameworks, and tech stacks side by side.

For AI Coding Agents

Build website clones or reference existing designs. Your AI reads the actual CSS and HTML — pixel-perfect results instead of guessing from screenshots.

For Content Migration

Moving to a new platform? Extract all assets, content structure, and styling from the old website automatically.

How to Connect

Apify Standby (Remote)

Connect any MCP-compatible AI tool to the Standby URL:

https://website-mcp-server.geniuslead.apify.actor/mcp

Claude Desktop (Local)

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "website": {
      "command": "npx",
      "args": ["-y", "website-mcp-server"]
    }
  }
}

Cursor

Add to your MCP settings with the same configuration as Claude Desktop.

Compatibility

Works with any MCP-compatible AI tool:

• Claude (Desktop & API)
• Cursor
• ChatGPT (with MCP plugin)
• Windsurf
• Any MCP client

Pricing

$0.05 per website scraped via Apify pay-per-event pricing.

Technical Details

• Download method: wget with full mirror mode (preserves site structure)
• Supported assets: HTML, CSS, JavaScript, images (PNG, JPG, SVG, WebP, GIF), fonts (WOFF, WOFF2, TTF, OTF, EOT), data files (JSON, XML, CSV)
• Framework detection: Next.js, Nuxt, React, Vue, Angular, Svelte, WordPress, Shopify, Webflow, Tailwind CSS, Bootstrap, and more
• Caching: Downloaded sites are cached for 1 hour to avoid redundant downloads
• Security: Path traversal protection, file size limits, timeout handling

Links

• GeniusLead — Built by GeniusLead
• MCP Protocol — Learn about MCP