> Commit code, it trains. Checkout, it reconstructs from memory.
I see a "Commit" button, but I don't see any checkout, I guess that is what happens automatically? Although that panel says "Reconstruction", not "Checkout"... You might want to do a pass after the AI created the initial version, and make it "humanly consistent" or similar, as right now it seems a bit incoherent.
> Run the JS code before and after a commit
How am I suppose to "run the JS code", manually writing it in my local JS console? Or is this just "Click the green button in the UI" but expressed in the way only a LLM would write it?
I see a "Commit" button, but I don't see any checkout, I guess that is what happens automatically? Although that panel says "Reconstruction", not "Checkout"... You might want to do a pass after the AI created the initial version, and make it "humanly consistent" or similar, as right now it seems a bit incoherent.
> Run the JS code before and after a commit
How am I suppose to "run the JS code", manually writing it in my local JS console? Or is this just "Click the green button in the UI" but expressed in the way only a LLM would write it?