Treemaps are the default, and are great for finding outliers with heatmaps, but terrible in terms of explainability. So now I'm heading into architecture diagra

Treemaps are the default, and are great for finding outliers with heatmaps, but terrible in terms of explainability. So now I'm heading into architecture diagram territory, and tree-sitter is helping with extracting the names of classes, functions, variables etc. LLMs can make decent diagrams out of this.