Generate Twitter Content in Personal Style with OpenAI & Supabase RAG

🎯 Self-Learning X Content Engine (Creator RAG Booster)

Learn your voice. Generate posts that sound like you — not AI.

🧩 Overview This n8n workflow builds a personal RAG (Retrieval-Augmented Generation) system for creators.
It learns from your own past posts and generates new tweets, replies, and image prompts in your tone.

⚙️ How it works

Step 1 — Ingest Use the “Add to KB” Form to upload your past posts or notes. Text + metadata (topic, style) are stored in Supabase as vectors.

Step 2 — Generate Use the “Generate Posts” Form to create new post ideas. The Agent fetches the most relevant style snippets (via Supabase VectorStore) Output includes:
📝 post
💬 quote
💭 reply
🎨 image_prompt

🔧 Setup (3–5 min) Connect Supabase (URL + Key)
Make sure the table name is documents
Enable vector extension (pgvector)
Connect OpenAI API Key
Activate both Forms and open the URLs to test.
Optionally replace Forms with Webhooks.

💡 Tip: RLS enabled? Ensure your API key allows insert/select for documents.

🧠 Tech Stack n8n (self-hosted) Supabase (Vector Store) OpenAI (gpt-4.1-mini) HTML-based completion form

🪄 Credits Built by Yusuke | @yskautomation
License: MITView on GitHub