AI agents write Ruby but can't navigate it: a 5-model, 13-codebase benchmark

**Project Summary:** The AI agents write Ruby but can't navigate it: a 5-model, 13-codebase benchmark is an open-source project that provides a unique benchmark