Very interesting little experiment! I'd agree that it was successful. Using the workflow you did, how well do you think it could be scaled for a bigger project? Really curious to hear your thoughts! I could see devs that are more art forward using agents to do lots of coding so they can focus on the art more, but I have personally felt like for bigger scope projects AI has mainly been helpful for a task that I generally know what I'm doing, otherwise things fill up with weird errors I can't decipher. But this was very cool to see! Overall it was quite a fun game, which is really what matters :)