Glad to hear you liked the game!
I really appreciate the compliment on my "artwork" since, as you have identified, I am very much not a trained sprite artist. (I've also been kicking myself for forgetting to include even a single visible explosion, it's practically genre sacrilege!) Any expansion to this project would definitely require outside help on that front, especially since the genre is so greatly defined by flashy visual effects.
As for the medal scoring mechanics, a brief on-screen explanation/demonstration would probably go a long way (but ultimately would have required a day or more beyond the jam timeframe to implement and refine). That, in conjunction with a gameplay UI widget that tracked the value of the next medal, and a visible callout when a medal is dropped, should probably be enough to communicate the mechanics to most players.
The hidden characters are indeed very hidden, and making them highly unlikely to be uncovered without intentional experimentation was by design, though I do think the last two (Ame and Calli) would benefit greatly from having a conspicuous (and literal) signpost like the other three (but again, time constraints). The fact that I could only manage one stage within the timeframe exacerbated the issue since there were as many secrets as there were types of trigger conditions, which meant the discovery of one secret doesn't really help the player figure out how to trigger the rest. In a more complete game the conspicuous signpost would highlight roughly where in the stage the secret was hidden, and the smol tracker UI widget would indicate the holomem whose reference hints at the trigger condition. Between that and knowledge of the types of trigger conditions (the five used in the first stage are the entire spread), players would (hopefully) have enough information to narrow down which of those trigger conditions should be used in that part of the stage to reveal each smol (though perhaps not always enough to immediately identify the exact condition on the first try).