Game review

Mykonos Island Voxels

7.5 / 10

Developer
boona13 (GitHub)
Publisher
boona13 (open source)
Released
1 January 2026
Platforms
Web browser (desktop and mobile)
Genre
Sandbox / Builder
Monetisation
Free, open source, no ads and no in-game purchases
Players
Single-player only

The short version

Mykonos Island Voxels has no goal, no score, and nothing to lose — it hands you a fixed catalogue of whitewashed walls, bougainvillea, cypress trees, and Cycladic domes and lets you arrange them on an isometric grid until the scene looks right. That is the entire game, and it works because the two things it needed to get right, it did: the art is genuinely lovely, and the touch controls were built for a finger from the start rather than bolted on afterward. Tap-to-place, drag-to-brush, a real long-press to erase, and pinch-to-zoom were all confirmed working through actual touch emulation, not just present in the code. What holds it back is the price of admission — a cold load that took over 30 seconds in testing while it fetches its asset pack one image at a time — and the fact that a sandbox with no goals runs out of things to do for anyone who isn't specifically looking for a digital diorama kit.

Biggest strengths

  • Every core action has a genuine touch implementation, confirmed by testing rather than reading the code alone — a tap placed an asset, a 600ms long-press through real CDP touch events erased it, and a drag gesture brush-placed across several cells in one motion
  • Every tool (place, fill, erase, pan, grid, save, reset) and every category tab is a real HTML button, so nothing on the shelf depends on the keyboard-only shortcuts also listed in the code — 1-5 for categories, E for erase, G for grid, S to save
  • The isometric art is genuinely attractive and cohesive: whitewashed cubic walls, weathered blue shutters and domes, bougainvillea rendered as a mass of small pink blocks cascading over a pergola, all sharing one restrained cobalt-and-terracotta palette

Biggest weaknesses

  • The asset pack loads sequentially, one `await` per image, and a cold load took over 30 seconds in testing on both the phone and desktop profiles before the scene became interactive — a genuinely slow start for what is a lightweight building toy once it's actually running
  • The repository ships roughly 80MB of image assets, which is a heavy one-time download for a browser toy, particularly on mobile data
  • There is no goal, challenge, or progression of any kind — once the novelty of arranging the fixed asset catalogue wears off, there is no mechanical reason to come back, unlike sandbox builders that add new pieces or scenarios over time

Who it's for: Anyone who wants a small, pretty, low-stakes building toy for a phone — the Townscaper crowd more than the Minecraft crowd

Our recommendation: Worth the download once the load finishes; don't expect a game with an ending

7.5 / 10

Open Mykonos Island Voxels and, once it finishes loading, you're handed a small whitewashed house, a chapel, a windmill, and a strip of blue water, arranged as a starter village on an isometric grid. There is a palette of terrain tiles, plants, props, water pieces and buildings along the bottom of the screen, a stack of tools on the side, and nothing else — no score, no timer, no win state. You either enjoy rearranging a small pretty diorama or you don't, and the whole review is really about whether the two things that determine that experience — the art and the controls — hold up. They do.

Built for a thumb, not adapted for one

InputManager.js is the part of this project worth reading closely, because it does something most of the candidates on this shelf don't bother with: it treats touch as its own first-class input model rather than a thin wrapper around mouse events. A single tap places the selected asset. A drag brushes it across every cell the finger crosses, tracked by cell key so it doesn't re-fire on the same tile twice. A 420ms stationary hold is scheduled the moment a finger touches down and fires as an erase if nothing moves before the timer completes — cancelled cleanly the instant the finger drifts past an 8-pixel threshold, so panning never accidentally erases a tile. Two fingers promote the same touch state to a pinch-zoom-and-pan gesture, anchored at the midpoint between them.

None of that is inference from a comment. Running the build through playtest.js on a simulated 390×844 touch phone, with real touch events fired through Chromium's own input pipeline, a tap on an unoccupied grass tile placed a cypress tree; a 600ms long-press on that same tile through a raw CDP touch-and-hold removed it again; and a drag across two rows of grass brush-painted trees along the path in one motion, confirmed in the resulting screenshots. Switching to the Pan tool and dragging moved the camera correctly. This is the rare case on this shelf where the touch layer isn't a workaround for something designed around a mouse — the mouse path in the same file is a parallel implementation, not the primary one.

Just as importantly, none of the game's actions live only behind a keyboard shortcut. Toolbar.js and AssetPalette.js build every tool, category tab, and asset swatch as a real <button> element wired to a click listener, which fires from a tap exactly as it would from a mouse — Place, Fill, Erase, Pan, Grid, Save, and Reset all sit in an always-visible dock, and the category tabs and asset grid are laid out the same way. The one exception is flipping an asset before placement, bound to H and V with no on-screen equivalent — a real but minor gap, since a touch player can still place, remove, and replace an asset in the orientation they want; they just can't rotate it in place first.

The art carries a game with nothing else going on

The isometric sprites — supplied as pre-rendered images rather than generated live, per the comments in assetLoader.js — commit fully to one aesthetic: sun-bleached white walls, weathered cobalt shutters and pergolas, terracotta roof tiles, and bougainvillea rendered as a dense mass of small magenta blocks spilling over walls and archways. It reads as a coherent, considered art direction rather than a generic voxel-game default palette, and the toolbar icons — hand-drawn on inline canvases rather than pulled from an icon font — carry the same restrained cobalt ink throughout. On both the 390px phone profile and the 1280px desktop profile the layout reflowed correctly: a docked side rail and side palette on desktop, a bottom sheet and pinned tool rail on phone, both laid out sensibly rather than the same fixed-width canvas squeezed into a smaller box.

What the art can't do is give the toy a reason to keep opening once a player has built one scene they like. There's no unlockable content, no scenario, no second island with different pieces — Townscaper, the obvious point of comparison, gets away with the same lack of a goal because its generative building logic keeps producing surprises from simple inputs; here the outcome is closer to "how you'd arrange this specific box of dollhouse furniture," which is pleasant once and doesn't obviously reward returning to.

A slow front door

The practical complaint is the load. assetLoader.js fetches the manifest's images with a sequential await inside a for loop, one request at a time rather than in parallel, and in testing that meant over 30 seconds between navigating to the page and the scene becoming interactive, on both the phone and desktop profiles, with the loading card's "crafting X…" label ticking through individual asset names the entire time. Two separate multi-step interactive test sessions on this build also ended with the browser process itself closing partway through rather than finishing cleanly — with no console error, page error, or failed request logged at any point before it happened, which points more toward a heavier memory footprint than an actual bug, but it's a data point worth having on record for a game shipping roughly 80MB of image assets to a phone.

That footprint also means the GitHub star count is worth treating skeptically rather than as a credibility signal in either direction: 1,052 stars and 231 forks all landed within minutes of the repository's single commit on 14 May 2026, and nothing has moved since. That is not what organic interest in a niche building toy looks like, and it shouldn't be read as evidence the game is either better or worse than what's actually here.

Where that leaves it

The two things this shelf exists to test — can you actually play it with a thumb, and is it good enough to send to a friend — both land in the game's favour. The controls are as thorough and honestly-built as anything reviewed for this shelf so far, and the art is good enough to be worth opening on its own. The load time and the complete absence of a goal are real, not dealbreakers: a 7.5 sits it at the top of "good, for the right audience" — recommended to someone who wants a calm building toy for a spare ten minutes, not to someone looking for a game with an ending.

Pros and cons

Pros

  • Every core action has a genuine touch implementation, confirmed by testing rather than reading the code alone — a tap placed an asset, a 600ms long-press through real CDP touch events erased it, and a drag gesture brush-placed across several cells in one motion
  • Every tool (place, fill, erase, pan, grid, save, reset) and every category tab is a real HTML button, so nothing on the shelf depends on the keyboard-only shortcuts also listed in the code — 1-5 for categories, E for erase, G for grid, S to save
  • The isometric art is genuinely attractive and cohesive: whitewashed cubic walls, weathered blue shutters and domes, bougainvillea rendered as a mass of small pink blocks cascading over a pergola, all sharing one restrained cobalt-and-terracotta palette
  • No third-party analytics anywhere in the source, and zero console errors, page errors, or failed requests across two full playtest passes
  • Fully responsive rather than a fixed embed: the same build reflows into a docked side toolbar and side palette on a 1280px desktop and a bottom sheet on a 390px phone, and both looked correct in testing

Cons

  • The asset pack loads sequentially, one `await` per image, and a cold load took over 30 seconds in testing on both the phone and desktop profiles before the scene became interactive — a genuinely slow start for what is a lightweight building toy once it's actually running
  • The repository ships roughly 80MB of image assets, which is a heavy one-time download for a browser toy, particularly on mobile data
  • There is no goal, challenge, or progression of any kind — once the novelty of arranging the fixed asset catalogue wears off, there is no mechanical reason to come back, unlike sandbox builders that add new pieces or scenarios over time
  • Two long interactive test sessions ended in the browser process being closed mid-script rather than a clean finish; no console or page error accompanied either instance, so this reads as the heavier of the two candidates tested this cycle rather than a confirmed defect, but the game's memory footprint is worth watching on lower-end phones
  • Flipping an asset before placement (H/V) is keyboard-only with no on-screen equivalent, so a touch player can't mirror a fence or staircase before dropping it, only place it and redo
  • The GitHub repository's 1,052 stars and 231 forks arrived within minutes of its single initial commit and the count has not moved since — numbers that do not read as organic reception and shouldn't be taken as a signal of real-world popularity either way

Category scores

Category scores are the reviewer's read on each area. They inform the overall score but do not average into it — the number at the top of the page is a judgement, not a calculation.

  • Gameplay 6.5 / 10
  • Visuals 8.5 / 10
  • Performance 5.5 / 10
  • Originality 6.5 / 10
  • Replayability 5.5 / 10