Hello HN ,
I built a knowledge graph engine for codebases
- Records why each change was made — written back by agents as they work, so it gets better the longer you use it - Dashboard stays in sync with commits and updates from GitHub and GitLab - Flags stale files/code and deranks them internally, so the graph stays current - Token savings in the range of 78 to 99.74%. - Connects to claude code, codex , opencode etc I will be happy to talk about Graph , sych updates ,persistent memory and also how it detects stale files and token savings ?
Visit : https://github.com/Novaya-AI/novgraph https://trynovaya.com/
Comments URL: [https://news.ycombinator.com/item?id=49672213](https://news.ycombinator.com/item?id=49672213)
Points: 2
source & further reading
news.ycombinator.com — original article
Dwarf Star Support for DeepSeek 4.1 Flash on MBPro 128GB
GoatCode – open-source terminal AI agent with provider failover
Ask HN: What is the most overlooked risk in the AI security domain?