I’m actually experimenting with a graph-based IDE (github.com/v-noc/IDE) around this problem. Instead of treating the codebase mainly as files and folders, it represents things like functions, classes, and files as connected nodes. Then documentation, logs, tasks, etc. can be attached to the parts of the code they relate to. The idea is to make it easier to understand a codebase without constantly jumping between files and trying to reconstruct the context yourself. It could also make onboarding easier, or let you start from behavior and execution paths and drill down into the code when you actually need it. I’m also experimenting with a walkthrough generator for AI-generated work. Instead of scrolling through a long AI chat to understand what happened, it uses graph traversal to follow the relevant parts of the code and generate a step-by-step walkthrough. It’s still very much an experiment, but this is the problem I’m trying to explore.