mcaln/mega-brain — reverse-engineered prompt
Reverse engineered prompt
Build me a local AI knowledge system for Claude Code that helps me turn expert content into something actually usable. I want to be able to drop in videos, PDFs, transcripts, podcasts, and course materials, then have it organize that into structured playbooks and a searchable knowledge base with clear links back to the original source.
It should extract the big ideas, mental models, heuristics, frameworks, and step by step methods from each source. I also want agent style helpers, like a mind clone for a specific expert, role based assistants for things like sales or marketing, and a conclave mode where multiple agents can reason together and give a structured answer backed by evidence.
Please include a simple setup flow for API keys, keep everything local by default, and support slash commands like /setup, /ingest, /process jarvis, /conclave, /save, /resume, and /jarvis briefing. Session persistence, logs, and an inbox for raw materials should all work. Look up current docs online if you need to.
Want more depth? Deep Reverse