Understand Everything
AI-powered toolchain that transforms any code repository into easy-to-understand interactive documentation through Git history analysis and hierarchical summarization.
Understand Everything is a toolchain for deeply understanding code repositories. It analyzes Git history, uses AI to interpret code, generates hierarchical documentation, and creates an interactive website that helps you easily understand any complex codebase.
Key Features
- Visual Analysis: Generate repository structure heatmaps showing file modification frequency
- AI Interpretation: Use Gemini 3 Pro Preview to generate easy-to-understand code explanations
- Hierarchical Docs: Recursively generate README files for each directory using bottom-up summarization
- Interactive Website: Read the Docs style static website with collapsible file tree navigation and syntax highlighting
Technical Highlights
The toolchain follows a clear three-step pipeline: (1) s1_explain_files.py uses async concurrent processing to generate AI interpretations for each file, (2) s2_generate_readme.py recursively builds hierarchical summaries starting from the deepest folders, and (3) s3_website.py generates a responsive static website with Prism.js code highlighting. Successfully analyzed large open-source projects including verl (1100+ files) and Megatron-LM (1330+ files).
Visit the GitHub repository for installation instructions and live demos.