theaiautomators/insights-lm-local-package — reverse-engineered prompt

I want a fully private local version of InsightsLM running on my machine, basically a NotebookLM style research assistant that works offline. It should let me upload PDFs, text files, and audio, chat with my documents, give answers with citations back to the source, transcribe audio locally, and generate spoken audio summaries or podcast style recaps from my materials.

Please set it up using the repo as intended, with Docker for the local services, Supabase running locally, n8n for the backend workflows, and local AI through Ollama, Whisper, and Coqui TTS. Use the included env copy files, compose copy files, migration SQL, and Supabase functions, and wire everything together so the app actually works end to end. If this repo depends on the base local AI packaged project, bring that in and connect it properly.

I do not want any cloud APIs unless the docs clearly require it. Keep everything local and private. Look up current docs or the walkthrough video if you need to.