Rendex
Get API key
Recipe

Extract Clean Content from a URL and Summarize It with AI in n8n

Raw HTML is noisy — ads, nav, cookie walls, and JavaScript get in the way of feeding a page to an LLM. Rendex Extract returns clean, reader-mode Markdown from the fully-rendered page, so your n8n AI step gets exactly the content, and nothing else.

Last updated 2026-07-21

Get Free API KeyDownload n8n templateSetup Guide

The Problem

Building a research assistant, a content digest, or a RAG ingest means getting clean text out of live pages. Plain HTTP fetches miss JavaScript-rendered content and drag in boilerplate; DIY readability parsing is brittle and breaks on paywalls and consent walls.

The Solution

Rendex Extract renders the page in a real browser, strips the chrome, and returns clean Markdown (or structured JSON) under data.content — with the title, byline, and excerpt. Pass data.content straight to an LLM Chain or an AI Agent for summarization, or into a vector store for retrieval. It reads JS-heavy pages and can hide cookie banners before extracting.

How the Workflow Runs

Target URL

A webhook, RSS feed, or Sheets row supplies the page URL

Rendex Extract

Return clean Markdown — no ads, nav, or boilerplate

Summarize (AI)

An LLM turns the content into a summary or embeds it for RAG

Input → Rendered Output

Left: a cluttered web article with ads and navigation. Right: the same article as clean, reader-mode Markdown with just the title and body — ready for an LLM.

Rendered by Rendex

What You Need

  • A Rendex API key — free tier, 100 renders/month, no card
  • n8n with the Rendex community node installed
  • A chat-model credential (OpenAI, Anthropic, or any n8n-supported model)
  • A source of URLs — webhook, RSS, or a Sheets column

What This Recipe Uses

Clean Markdown, not raw HTML

Reader-mode extraction strips ads, nav, and boilerplate, returning just the article text — ideal input for any LLM.

Reads JS-heavy pages

Extract renders the page in a real browser, so content that only appears after JavaScript runs is captured — unlike a plain HTTP fetch.

Ingestion-ready for AI

data.content drops straight into an LLM Chain, an AI Agent, or a vector store — no cleanup step between extraction and the model.

Build It

rendex-extract-node.json
{
  "parameters": {
    "resource": "document",
    "operation": "extract",
    "url": "={{ $json.url }}",
    "extractFormat": "markdown",
    "additionalFields": { "blockCookieBanners": true }
  },
  "name": "Rendex — Extract",
  "type": "n8n-nodes-rendex.rendex",
  "typeVersion": 1
}
extract.sh
curl -X POST https://api.rendex.dev/v1/extract \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://example.com/article", "extractFormat": "markdown" }'
# -> { "data": { "title": "...", "content": "# ...clean markdown...", "byline": "..." } }

100 free API calls/month — no credit card required. Get your API key and start building.

This is the front door to any AI-over-the-web workflow in n8n. Whether you're summarizing competitor announcements, building a knowledge base from documentation, or ingesting articles for retrieval, the hard part is getting clean, LLM-ready text — and that's exactly what Extract returns. Because it renders the page first, it handles single-page apps and content that only appears after load, and the blockCookieBanners option clears consent overlays that would otherwise pollute the text. Feed data.content to a Basic LLM Chain for a quick summary, or swap in an AI Agent plus a vector store node for a full RAG ingest. Extraction is a free-tier capability, so prototyping a pipeline costs nothing beyond a credit per page.

Frequently Asked Questions

Related Resources

Documentation