doloop decorations paints an expert's reading order onto the file you already have open. It dims the connective tissue, lights the code that carries the decisions, and marks where to start, computed from the whole repo. Deterministic, local, zero tokens.
All deterministic, all local, zero tokens. It rides VS Code's and Cursor's own decoration surface, so there is nothing new to learn and nothing leaves your machine.
The order follows how developers are documented to read unfamiliar code: focus on an entry point, expand to what it touches, connect the pieces, integrate the whole (Sillito et al.). That is the loop an expert runs in their head. doloop runs the same loop over your whole repo and paints the result, the doors to focus on, the muscle to read, the connective tissue to skip. And it recurses, the same gaze at every zoom, repo to region to file to unit, because comprehension works the same way at every scale. It is a model of how code gets read, computed deterministically, not a heuristic we invented.
What it computes is structure: which files everything hangs off, which units hold the branching and the state, which lines are pass-through. That runs once ahead of time; the paint itself is plain editor decorations, so nothing lags as you type.
No. The three things that look like it each do something else.
Colors by language rules, the same in every repo. Nothing is ranked.
Explains in prose. A different answer each run, tokens each run, a slice of a large repo.
Reads this repo whole, ranks it, and paints the result in your open file. The same paint every run, zero tokens.
It runs unmodified in VS Code and Cursor (same extension API). Bring nothing but the editor you already use.
Routes you to a real doloop page, asks when your question is ambiguous, or tells you when there is no answer. No model runs on the answer path, so it cannot invent one.