Trace - Local repository intelligence engine

Hey Product Hunt community! 👋I’m Tarang, the creator of Trace.As developers, we’ve all been there: you drop into a massive, unfamiliar codebase with hundreds of files, and onboarding takes days. While cloud AI tools can help, copying and pasting internal or proprietary codebases into third-party cloud APIs is a massive security compliance risk for most companies and indie hackers.I built Trace to fix this. It brings deep codebase intelligence directly to your local machine with absolute privacy. Zero data ever leaves your device.How Trace helps you master codebases:📊 Auto-Constructed Knowledge Graphs: It maps out your source files into a strict relational graph with 15 node types and 17 edge types to chart how modules connect.🔍 Downstream Impact Analysis: Want to change an export? Trace traces the graph to show you exactly which files will be affected before you commit.🧠 100% Offline AI Chat: Powered by Ollama (defaulting to qwen2.5-coder), you can query your repository's routing, architecture, and bugs on a flight without Wi-Fi.💻 Hardware-Aware Optimizer: Trace automatically scans your Mac's RAM and GPU to recommend the most efficient model size your machine can handle smoothly.It’s open-source, runs natively on macOS (Apple Silicon), and sets up via a few quick terminal commands using pnpm and Ollama.I would love to hear your thoughts, feature requests, and feedback! Are there specific codebases or languages you want to see mapped next?Check out our repository here: github.comThank you for the support! ✨