How do I scrape a website in n8n?

Add an HTTP Request node with the page URL, then an HTML node with the Extract HTML Content operation and a CSS selector for each value you want. That covers static pages. For pages that build themselves with JavaScript, use a node that runs a real browser, such as Firecrawl, Apify or MiroMiro.

Can n8n scrape JavaScript websites?

Not with the HTTP Request node alone: it downloads the HTML before any JavaScript runs, so React, Next.js, Framer and Webflow pages often come back nearly empty. Use a node that loads the page in a real browser first.

Can n8n turn a website into code or an Elementor template?

Yes, with the MiroMiro node. Its Extract Code operation returns a section of any page as Tailwind, HTML + CSS, React, Vue, an Elementor template (.json) or a Shopify section (.liquid), styled like the original. One section costs 10 credits.

Is scraping with n8n free?

Self-hosted n8n and the HTTP Request and HTML nodes are free. Browser-based scraping services charge per page or per credit; MiroMiro includes 300 free credits a month with no card, which is about 30 sections as code or 300 image lists.

How do I scrape many pages in one n8n run?

Put the URLs in a list (a Code node, a Google Sheet or a Read From File node), and n8n runs every node after it once per item. Add a Wait node if a site rate-limits you.

n8n Website Scraping: Text, Code, Brand Kits and Images

To scrape a website with n8n, chain an HTTP Request node (download the page) and an HTML node (pull values out with CSS selectors). That handles static pages in two minutes. It breaks on JavaScript-built sites, and it only ever gives you text. If what you want is the page's design (code, an Elementor template, brand colors, the images), use a node that reads the page in a real browser and returns that directly.

Here are the three setups, when each one is right, and the problems you will hit.

Scrape text with HTTP Request + HTML

First, an HTTP Request node:

Method: GET
URL: https://example.com/pricing
Response format: String (put the body in a field called data)

Then an HTML node:

Operation: Extract HTML Content
Source data: JSON, property data
Extraction values, one row per value:
- Key title, CSS selector h1, return value Text
- Key plans, CSS selector .plan-name, return value Text, Return array on
- Key image, CSS selector meta[property="og:image"], return value Attribute, attribute content

Run it and you get one item with title, plans and image.

This is the right tool for prices, headlines, links and anything else that is text in the HTML. It is free and it is fast. I would start here every time the page is server-rendered.

Scrape JavaScript websites with a browser node

Open the page with View Source. If the text you want is not in it, the HTTP Request node will not see it either: the site builds its content in the browser. Most React, Next.js, Framer and Webflow sites behave like this.

You need something that runs a real browser and gives the result back to n8n. Firecrawl and Apify both have n8n nodes; Firecrawl returns the page as Markdown, Apify runs whatever scraper you pick from its store. For text from JavaScript sites, either is a good choice.

Scrape the design: code, Elementor, brand and images

Text scrapers return text. If you are rebuilding a site, moving it to WordPress or Shopify, or collecting a brand, you want the design itself, and HTML-plus-selectors gets you nowhere near it.

That is what the MiroMiro node does. It loads the page in a real browser and returns:

Operation	What you get	Credits
Extract Code	A section (hero, pricing, footer...) as Tailwind, HTML + CSS, React, Vue, an Elementor template or a Shopify section	10
Extract Brand	Logo, brand colors with roles, fonts, and whether each was declared by the site or inferred	10
Extract Design Tokens	Every color, font size, spacing and radius with usage counts, or a W3C tokens file	10
Extract Images	Every image URL: img tags, srcset, CSS backgrounds, favicons, og:image	1
Extract SVGs, Fonts, Videos, Audio, Lottie	The files of that type	1 to 5

To try it:

Install the node. On self-hosted n8n: Settings > Community nodes > Install, package n8n-nodes-miromiro. It is verified by n8n and arrives in n8n Cloud's nodes panel with their next release.
Add a MiroMiro node, create the credential with a free API key.
Pick Extract Code, paste a URL, set Section to Pricing and Output to Elementor Template.

The code field is the complete template. Send it through a Convert to File node and import the .json in WordPress. Every result also lists the page's other sections with a ready selector, so a Split + second MiroMiro node can rebuild the whole page. The ready-made templates do exactly that.

Which one to use

You want	Use	Handles JavaScript sites	Cost
Text from static pages (prices, titles, links)	HTTP Request + HTML	No	Free
Text or Markdown from JavaScript pages	Firecrawl or Apify node	Yes	Per page
Code, Elementor, Shopify, brand kit, images	MiroMiro node	Yes	1 to 10 credits per call

Common problems

The HTML node returns empty values. The content is built by JavaScript. Check View Source; if it is missing there, switch to a browser node.

URL is not defined in an expression. n8n's expression sandbox has no URL constructor. To get the domain of a link, use {{ $json.url.split('/')[2] }} instead.

The site returns 403. Bot protection. Browser nodes get through most of it. When a site refuses even a real browser, no scraper is going to work on it, and MiroMiro says so instead of returning an empty result (and does not charge the call).

Images only appear when you scroll. Lazy loading. On the MiroMiro images operation, add the Render option: the page is scrolled in a browser before the images are collected.

One run, hundreds of URLs. Read the list from a Google Sheet or a Code node; n8n runs each following node once per item. Same page, same settings within 24 hours costs nothing on MiroMiro, so re-running a failed batch only pays for what actually re-extracts.

Or start from a template

Every setup above exists as a free n8n template you import in one click: download every image from a website, rebuild a whole page as code, a page in Elementor, brand kits and more. The MiroMiro API includes 300 free credits a month, no card. You can start for free.

Summary

For text on static pages, HTTP Request + HTML is all you need. For text on JavaScript pages, add a browser node. For anything you want to build with, the design itself, skip the selectors and use a node that returns code, templates and brand data directly.