Downloads

Chat trigger + Whisper STT + AI Agent + Simple Memory. Fixed session key — works with intermediate nodes between trigger and memory.

Manual trigger → AI Agent (Groq llama-3.3-70b) generates a tip of the day → HTTP Request posts it to the COS Heidelberg #AI-test channel via webhook. Shows the full loop: LLM output → real-world integration.

Schedule (every 1h) → SearXNG web search → Code (extract top 5) → AI Agent summarises → posts digest to #AI-test. Change the search query and webhook URL to make it your own.

Chat trigger → fetch any URL → AI Agent answers questions grounded in that document. Edit the Document URL node to use your own paper or page.

Router → IF node → two branches: 'summarize' (3-5 sentences) or 'extract' (numbered facts). Shows how n8n routes text to different AI pipelines based on keywords.

Chat trigger → Code (build PubMed URL) → esearch → Extract IDs → efetch plain-text abstracts → AI Research Assistant. Two-step API pattern: search for IDs first, then fetch content.

Chat trigger → Code (extract PMID) → IF node → TRUE: fetch paper with error output enabled → AI summarize / HTTP error handler. FALSE: guide message. Shows IF routing and Continue on error.

Chat trigger → PubMed search tool → Paper details tool → Format paper list → AI Research Synthesizer. Pipeline implements the tool-use / ReAct pattern deterministically in n8n.

5-stage pipeline: Search (esearch) → Fetch (efetch) → Extract (AI structured JSON) → Filter (removes incomplete) → Format (Markdown comparison table). AI extracts method, finding, limitation per paper.