Played through an entire level from start to finish with a few failures, and I think I have sort of an understanding of how the game plays. The game looks nice with that pixel art style, and 2d character on 3d worlds is one of my favourite style that I would like to try myself. Game runs smoothly and it works from start to finish with a gameplay loop.
Okay now the elephant in the room. This is clearly, and I meant clearly, AI performing most of the design choices and heavy lifting (which is I tend not to comment on these as people will get unhappy/the comments itself is irrelevant). One of the biggest tell is how unbounded the scope will become once you've got infinite resource in game developement without restrictions like technical difficulty, personal ability and man-hour. Which what tends to happened is that the game will have an overabundant of features and mechanics that will stray away from the core feature, making the game felt unfocused. Just because you can don't mean you should. For people developing in the stone age way (yes I mean myself) feature are usually carefully selected and prioritiesd as you simply have not enough time to do so. But that's the beauty of game development, and able to overcome those is what praises for.
Sorry for the rant, lets get back to the game itself and see what's problematic and what can be done to improve:
- There are a lot of graphic noise in the game, like a lot of particle effects, attack area, UI etc to detract people from the core gameplay. All effects should only focus on the player character and strictly necessary UI that matter in the current scene. For example. no the baking tray do not need to be shown during battle. Only show attack area for the attacks that truely matter without any telegraph. These are reserved for thing like instant lasers and falling rocks not slow projectile and bombs.
- Gameplay systems are too many and is not introduced. Like I don't even know how an order works in the game and it already showing me the baking tray with 6 options in just 20 seconds. Each element must be taught to the player clearly and allow time for the player to digest. Keys? Luck? Cooldown? Special? Through them all at once to the player is not a good feeling. And what the heck B4O2 even mean.
- Going back to attack the player character attack felt strangly sluggish a simple attack animation that doesn't feels good to play. It doesn't not have the attacking feel and it's really strange seeing every other stuff looks good but not the player character, which should be the centerpiece of the game. It's funny cause it is also the case with some other mainly AI games, I guess it's what Claude decided to do.
- Okay you are going for character selections, is the characters necessary or differ enough to warrent a second playthrough? Even if you need them, don't show them straight away that would just confuses the player before even starting the game. This goes the same for pantary and the other thing I can't remember.
- The lucky number theming felt weak in the game, and it feel like the game can function perfectly without it, which in my book is not a good thing in jam. We are supposed to be inspired by the theme and create something around it, not retroactively slapping a theme in a game, which to be fair also happened to a lot of games in non-AI jam. But seeing you are allowed to use AI, there are less excuse to do so.
Anyway, forgive my preachy tone and rant, and as I said you would probably not like it a lot. My suggestion for if you want to make a game better, try making something by yourself first without any AI/assistant, learn what is important for a game and the feeling of prioritise/giving up features and idea, and then using that knowledge with AI, and that would yield a much better product. As an old saying in software developement: Idea is cheap, execution is king. With AI, idea is much more cheaper, and it's sadly not just do it, but you need to do it well.