All projects · AI Agents & RAG
AI Website UX Auditor
Fetches a target URL over HTTP, strips the HTML down to clean text, and sends that (capped at 6000 chars) to GPT-4o-mini acting as a…
Fetches a target URL over HTTP, strips the HTML down to clean text, and sends that (capped at 6000 chars) to GPT-4o-mini acting as a conversion-focused UX auditor. The model returns a strict-JSON audit (overall priority, UX/trust/conversion issues, and prioritized recommended fixes), which is parsed, formatted, and delivered to Telegram.
- Live web fetch feeding an AI audit: Set URL to Fetch Page to HTML-cleaner to model to parser to Telegram
- HTTP Request uses neverError + onError continueRegularOutput so a bad fetch does not break the run
- Regex HTML-to-text cleaning (drops script/style/comments/tags, decodes entities) with a 6000-char cap to control token cost
- UX-audit JSON schema separating UX, trust, and conversion issues plus prioritized fixes
Built with: n8n, HTTP Request node, OpenAI gpt-4o-mini, @n8n/n8n-nodes-langchain.openAi, JavaScript Code nodes, Telegram Bot API