Insights and updates from the Supermodel team- Code Graphs, Benchmarking MCPs, building useful primitives for code factories and more.
Context loss during conversation compaction is one of the biggest challenges facing AI coding agents today. Learn how persistent code graphs solve this problem and keep your agents productive across long sessions.
We built Uncompact to solve a real problem we faced daily — agents losing track of codebase structure after auto-compaction. Here's what we learned building a tool that maintains persistent context.
A deep dive into how we built a real-time code analysis API that processes millions of lines across multiple languages. From AST parsing to graph storage, here's the technical story.
