Hey Product Hunt! Edwin here, founder of GZOO. I built Cortex because I kept losing context across projects. I run multiple codebases simultaneously, and decisions, patterns, and architecture choices end up scattered across hundreds of files. I'd forget what I decided three months ago and waste time re-solving problems I'd already figured out in another repo. So I built a tool that watches my project directories, extracts knowledge automatically using LLMs (decisions, patterns, components, dependencies), stores everything in a local knowledge graph, and lets me query across all of it in natural language. A few things I'm proud of in the design: Local-first — everything lives in SQLite + LanceDB on your machine. No cloud database, no account required. Hybrid LLM routing — high-volume tasks like entity extraction run on Ollama (local, free), reasoning-heavy tasks like queries go to Claude. You can also go fully local if you want zero data leaving your machine. Privacy controls — mark any directory as "restricted" and it never touches a cloud API. Sensitive files (.env, .key, .pem) are auto-detected and blocked. I've been using it daily on real projects for months before open-sourcing it. It genuinely changed how I work — I query past decisions instead of grep-ing through repos trying to remember what I chose and why. It's MIT licensed and free. Would love to hear what workflows you'd use it for and what features would make it more useful for your setup. GitHub: https://github.com/gzoonet/cortex

GZOO Cortex

A local-first knowledge graph for developers

A local-first knowledge graph for developers