🧠 Second Brain & Graph-RAG Setup - 📺 Second Brain Video Overview: Watch on YouTube - 📖 Second Brain Installation Guide: SecondBrain_install_guide.pdf - 📑 Architecture Report: Second Brain Graph-RAG Architecture Report.pdf - 📝 Notes App: Download Obsidian 🤖 AI Agents & Tools - ⚙️ AI Agents & Usage: Overview & Workflows - 💻 Local/Host AI Setup: grok_cli_install.pdf (Host AI Grok Installation) 🛠️ Infrastructure & Data Sync - 🔄 Data Synchronization: Rclone (Sync data sources) - 🌐 Free App Hosting: Cloudflare Worker Skool Post

A while back I posted about turning my Claude chat history into a living, self-organizing knowledge base instead of letting good conversations die in scrollback. Now it has a home: https://second-sight.sudedev.workers.dev/ What's changed since that post: It updates itself. Every day my pinned Claude chats get scanned, and anything new or changed gets categorized, cross-linked, and pushed to the site automatically, no manual exporting. TR/EN, one toggle. No more searching the same concept twice in two languages and finding neither. It reminds me before I forget. Spaced-repetition surfaces old notes right when they're about to fade, instead of me re-learning the same thing a third time. Still rough in places, but real now instead of a plan I keep describing. If you're doing something similar with your own chat history, I'd love to compare notes, especially on the "make it update itself" part.

For a while now, something has been bothering me: I'd have these genuinely great conversations with Claude, working through a concept, untangling a confusing architecture, finally getting something to click💡 and then... it would just disappear into my chat history. Gone🤦‍♀️ Never to be seen again unless I happened to remember the right search terms. It really hit me while I was deep in prep for the Claude Certified Architect (CCA-F) exam under the supervision of Rifat Erdem Şahin. I kept noticing I'd ask Claude almost the same question about the same architecture pattern, weeks apart, because I'd simply forgotten I'd already worked through it once. That was the moment I thought: this knowledge shouldn't be disappearing, it should be compounding. So I started building something to fix that. The short version: I'm turning my Claude chat history into an actual second brain in Obsidian, organized around the PARA method (Projects/Areas/Resources/Archive) , one that grows on its own. Concepts I discuss get pulled out, organized, and linked to what I already know. Instead of a pile of disconnected chat logs, I end up with a living, breathing knowledge graph that reflects how I actually think and learn. One fun side effect: since I switch between Turkish and English depending on what I'm working on, I used to end up searching for the same concept twice, in two languages, and often couldn't find either. Building this system forced me to actually organize that mess. What I care most about isn't just "storage", it's making the system smart enough to surface what I've forgotten, flag the topics I keep getting confused about, and turn raw conversations into something I actually review and retain, spaced-repetition style. There's something oddly satisfying about the system quietly reminding you of something you struggled with a month ago, and realizing you actually know it now. Not a dead archive. An active one. I'm still refining things behind the scenes, but I'm getting close to a proper, polished version of this, something I'm hoping to actually put out there soon 👀