Hmm, I went and read some of your articles, and 'the bad fight' piqued my interest. There is another designer who does short experiments that I think you might find some inspiration in - another way to vary the fixed encounters is to have them arranged nonlinearly with the rewards being expanded character build/mechanical options rather than just power scaling. "Wine and Roses" does this and I think is enjoyable. Sort of a turn based boss-gauntlet 'Metroidvania'. This author, Craze, has also played around with large, swappable parties and attrition in ways that might be inspiring to you, based on your writing.
Perhaps even better - Obelisk: Devilkiller, also by Craze? It has the same kind of non-linear 'metroidvania' style of gameplay, but also has a 'Burst' system that reminds me in some ways of some of your systems. It also has a larger roster of characters you can exchange with swappable equipment so there is some strategy of preparing for specific battles.
Have you tried Visions and Voices (also Craze)? I don't think the mechanical systems can touch my other recommendations, but it is interesting for pitting the player against attrition and a sort of time limit, with lots of optional exploration that furthers the characters/story. There is a tension between taking risks and playing it safe that I liked, reminding me of a good roguelike.
And for a generally different take, but another short experience with what were really memorable and fun battle mechanics for me, In Search of Immortality by Indra and Fomar. There are levels, but they only increase the 'slots' you have to modify your characters.
These might help inspire you if you seek to take some of the "tightness" out of your games ('overdesigned fixed combats' I think you said), either by trying out less linearity in the fixed combats (Obelisk and Wine and Roses) and possible loadouts, or maybe by playing around with more meaningful random encounters (V&V or Immortality)... I hope you also realize that the mechanical 'puzzly' tightness you have been offering has its place, and you are doing it *really* well.
For example, I would say "Slimes" has a similar focus on a gauntlet of battles with a (slowly expanding but basic) toolset. And I enjoyed it, but I was eventually feeling "solved" with that game's strategy, while Ocean OI had me pissing myself laser focused in every battle. Perhaps not for everyone, but a niche that exists and you are equipped to fill well!
Okay, last one for now: Grist of Flies by Razelle. Nice short game with fun mechanics. Interesting ideas - 'characters' are squads with eq slots that represent jobs the individual characters can do. So team members are swappable to different squads, and have different efficacy ratings in the different positions (although the demo never has any reason to put anyone in a position that's not their *best* one as far as I could tell. Action economy that includes moves for building up MP. Buffs and debuffs are really important - ditto status effect, both always work on everything. This is one of my favorites and only forgot it because it's been a while.
visions and voices is the one craze game i have played, back when it was relatively new, although it didn't click with me then. looks like i've got a long overdue crazegame deep dive ahead of me. thanks for all the recs! (i'm a big slimes fan, too)
i'll take what you've said to heart, too, about puzzly tightness. i do appreciate it from time to time - what i think i'm really not interested in is trial and error solutions with a slim margin of error, and for me at least ocean OI dips into perilous guesswork more often than i'd like. i'm a little less interested in explicit reasoning than i am in building up intuition and managing risks, which feels really expressive to me in a way that i like. this is kinda what i arrived at after revisiting the 7th saga after atom OI and playing a bunch of nes DQ last year, both of which involve a lot of unpredictability and variation. cataphract moves a step toward embracing randomness (in kind of a 7th saga way, where things can always go south at the drop of a hat), but i'm hoping to more fully explore a sort of 'soft puzzle-solving' approach in a future title. (it occurs to me now that the piece where i try to dig into this a bit more isn't publicly visible right now, but if you're up for a long, messy read, check out this piece)