Hello Everyone. New to A.I. and having some challenges building an AI second-brain so Claude can search, reuse, and synthesize information in real time across many topics. One major focus is trading. Claude is directly connected to TradingView (the financial charting platform I use), so the goal is to combine live and historical chart data with stored trading knowledge, strategies, examples, screenshots, clips, and indicator behavior so it can act more like a trading partner. One special use case is an older closed-source indicator that I have lifetime access to. The creator passed away years ago, it is no longer maintained, and it was built long before modern AI tools. Since I do not have the source code, the key challenge is teaching Claude what it is seeing on the screen — what the waves, lines, colors, and timeframes represent, and how they behave in relation to price. I attached a PDF showing the build I’m currently leaning toward, but could really use some feedback as I'm quite new to all this.... I had one second-brain built and functioning, (Supabase, n8n, Slack, Telegram, Claude...) but just completely dismantled it cause it wasn't working perfectly & that all comes down to me needing to learn / know more. I want to build one correctly start to finish. The main feedback I’m hopin' for is whether this is the right foundation, and whether the capture strategy for screenshots, short clips, annotated video, and direct TradingView data is strong enough to make the learning side actually work. Would really appreciate feedback as I've done my very best to get this far. Much to learn. - whether the overall architecture makes sense - whether Obsidian + Supabase is the right foundation - whether Cognee should be included from the start - whether the capture stack works for teaching how the indicator system functions. - Other items I should be considering... Thanks A Ton For Any Feedback