Hello guys, Do your LLMs generate a lot of code you don't understand? It is not easy to study full codebases that include many folders and files. I was wondering, how could I learn how they work as soon as possible? I think the answer is based on these ideas: increase observability + more explicit information + top down approach. I built pyreplay with those ideas. I hope someone finds it useful. For now it only includes Python. It is in its early stages. Fastest way to understand the project is open the "FEATURES.md" file, you scroll down and will see screen captures for each of the +120 features. Thanks for your time!