Library / RAG & Knowledge Bases
RAG & Knowledge Bases
Split web pages into heading-based chunks with Ragfeed for embedding
Runs on demand to fetch a web page via Ragfeed and return one item per heading-based chunk, ready for embedding in a RAG pipeline.
What it does
## Split web pages into vector-ready chunks with Ragfeed ### Who it's for Anyone building a RAG chatbot or knowledge base in n8n who wants web pages as clean, ready-to-embed records instead of raw HTML. ### How it works 1. Set page and chunk size holds the URL and the maximum chunk size. 2. Fetch page as chunks calls the Ragfeed API (POST /v2/scrape, format chunks). Ragfeed renders the page, drops navigation, footers and cookie notices, and splits the main content along its headings. 3. One item per chunk turns the list into one n8n item per chunk. Each chunk has a stable id (UUID from the normalized URL and position, so re-fetching the same page overwrites instead of duplicating), text, embedText, tokens, the heading path and flat metadata (URL, title, language, fetch time, hash). ### Requirements A Ragfeed API key (starts with sk-). The free plan has 1,000 credits a month; one page costs 1 credit, chunks cost nothing extra. Sign-up and dashboard are in German: https://ragfeed.de/en ### How to customize Connect One item per chunk to your embeddings and vector store nodes (Qdrant, Pinecone, Supabase, Postgres/pgvector). Use id as the point id and embedText as the text to embed. Change maxTokens (100 to 4,000) for smaller or larger chunks.
The steps
- Run manually
- Set page and chunk size
- Fetch page as chunks
- One item per chunk
Similar automations
RAG & Knowledge BasesLoad help center pages into Pinecone with OpenAI embeddings on manual run18 stepsBuilding Blocks & UtilitiesUpload Large Files to Kommo CRM in Chunks20 stepsDocuments, PDF & OCRPrepare PDFs for RAG with pdfRest OCR and Markdown chunks28 stepsRAG & Knowledge BasesEmbed Files from FTP into a Qdrant Vector Database10 steps