Post

Agreed on the multi-repo part. The piece I kept missing was the why: it remembers what the code is, less often the decision that made it that way, and that's what the next session needs when it wants to undo it.

Good to hear it holds up in a cluttered multi-repo setup, that's the hard case. Which part does the heavy lifting for you: the files you write, or what it writes down on its own?