Hi guys,
LLMs generate a lot of code so now the bottleneck is understanding that code. For this, I created a tool using this principles: increase observability + make everything as explicit as possible + top down approach.
https://github.com/arnoldpredator/pyreplay
If you open the "FEATURES.md" file and scroll down you will see a screen capture for each of the 120+ features. You also have the README with the philosphy and the TUTORIAL.
Basically you can obtain a map to see the connections, who is doing the heavy work, and then record a run and replay it like a video, using a simple html file which you can share.
Any feedback?
Top comments (0)