KAZ makes the theme you pick change how the game plays

KAZ is a grid-based arcade roguelike played on four buttons. You act on a small board while the screen strobes through colour, chasing a score and a spot on the leaderboard. The developer pitches it as a dopamine rush for score chasers and one-more-run addicts, and the plain version of that pitch is worth stating: the input surface is tiny, the readout is enormous, and everything interesting lives in the gap between them.

The minimalism is not the unusual part — plenty of arcade games are four buttons and a board. The unusual part is that KAZ ships two opposite ways to play the same grid. One set of modes is a pure reflex test: find your flow, push your reaction time, keep up with the flashing. Another drops the clock entirely and gives you a limited number of moves instead, which turns the identical board into something you can sit and stare at. Five modes in all, unlocked across more than 145 quests.

Then there are the themes. More than 25 of them, they are nominally the cosmetic layer, and each one carries its own power-up. Choosing the look that feels like you is also choosing an ability, which means the wardrobe and the loadout are the same menu. Some are community-made, some come from artist collaborations, and all of them are a reason to play differently rather than just a reason to look different.

A run is assembled between rounds out of one of six starting weapons and then a growing pile of items — more than 110 of them — plus boosts, spells and consumables. There is also a curse gauge, and when it fills you have to take a malus: one of more than 35 downsides, chosen as the lesser evil. Some of them hide a power-up. Stack the right mix and you survive into endless rounds that keep getting flashier; stack the wrong one and you are back at round one with a number to beat.

Downwell’s Dual Purpose Design is the closest reference. It argues that nearly every part of Downwell holds down two or three jobs at once, and that the overlap is where its speed, depth, risk and hard choices come from. KAZ runs the same trick through the part of a game that is usually inert. A skin is normally a decision with no cost attached; making it carry a power turns taste into a trade-off, and gives the unlock list a reason to exist beyond collection.

The Two Types of Randomness in Game Design is the one that decides whether the items feel fair. The essay’s split is whether the dice fall before the player decides or after, and KAZ puts nearly all of its luck in front of the decision — the weapon, the item, the boost and even the malus are offered to you and picked between rounds. A downside that secretly contains a power-up is only interesting because you agreed to it. Roll the same curse onto a player mid-round and the identical content reads as the game cheating.

Roguelikes, Persistency, and Progression asks what survives your death, and KAZ answers with a lot: quests, themes and modes all accumulate outside the run while the run itself keeps nothing. That is the standard roguelite bargain, and it carries the standard risk — the unlock list can quietly become the reason to play instead of the score. The leaderboard is the counterweight, and how well those two pull against each other is the thing to watch after the first few hours.

KAZ is made by [Kalinarm](https://linktr.ee/playkaz) and published by HAKURO, out now on Windows, listed under Action and Indie. It is single-player, with Steam achievements, Steam Cloud and partial controller support.

For a game whose entire surface is flashing colour, the accessibility flags are the part of the store page worth reading. It lists colour alternatives, camera comfort, custom volume controls, a keyboard-only option, and a mode playable without timed input. That last one is not a bolt-on — it is the tactical, limited-moves mode, doing two jobs at once in exactly the way the rest of the design does.

If four buttons sounds thin, Scarlet Skips is the one to read next. It spends an entire roguelike on a single button — press, hold and release — and it is up against the same question KAZ is: how much depth a very small input surface can hold before the design starts wanting one more thing to press.