Cooking for Skully

Cooking for Skully opens with a joke that only really lands once you've played it: the Devil is cooking dinner for his skeleton girlfriend, Skully, in a kitchen built into the side of a lava lake, and everyone in his life — the Goblin who actually runs the place, Skully herself on the phone — treats this as completely ordinary. It's a strong hook for a js13k 2022 entry from Gabor Héja, and the writing keeps that comic register through a genuinely long cinematic opening before handing control over. Once it does, the game is a fairly conventional Overcooked-style loop — grab ingredients, cook them in the right order, plate them before a customer's patience runs out — dressed in isometric pixel art that reads clearly even at phone resolution.

The writing is real, and the judges said so unprompted

It's worth taking seriously that Cooking for Skully's competition comment thread is unusually warm for a jam game. Salvatore Previti called it "a little gem" that "made me laugh (in a good way)"; Jerome Lecomte called it "a nice spin on Overcooked" with "bonus points for the cinematic intro"; Taufik Sani, a self-described Overcooked fan, praised how the concept stayed fresh while fitting into 13 kilobytes. That's not manufactured enthusiasm — js13k judges play dozens of entries in a short window and rarely single out writing specifically, and several did here independently.

A frame-rate bug the source confirms, and a judge caught in the wild

The most important technical finding here didn't come from playing the game — it came from reading the shipped bundle. Searching it for the two calls a game needs to measure real elapsed time between frames, performance.now() and Date.now(), turns up neither; the only timing primitive present is requestAnimationFrame itself, called with no delta-time calculation visible anywhere in the update logic. That matters because requestAnimationFrame fires once per display refresh, not once per fixed time slice — on a standard 60Hz screen that's roughly 60 times a second, but on a 120Hz gaming monitor it's roughly 120 times a second. A game that advances its state by a fixed amount per callback, rather than scaling that amount by real elapsed time, will simply run twice as fast on a 120Hz screen. Js13k judge Rob Louie reported exactly this in 2022, in enough technical detail to confirm the mechanism rather than just the symptom: "the game ran at double speed, most likely because I have a 120hz monitor," to the point that he broke out an old 2015 MacBook specifically to get a 60Hz screen and play it properly. This isn't a hypothetical edge case either — a growing share of Android phones, the exact device class this shelf serves, now ship 90Hz or 120Hz displays by default, so the same bug plausibly affects a meaningful slice of mobile players too, not just desktop gamers with high-refresh monitors.

Recipes confused more than one player, not just one

Separately from the speed issue, the competition comments show a pattern rather than a single complaint: Brian Hambley "couldn't get the order to accept" until realising an order had to be selected first; Thiago Romão Barcala noted "the order in the recipes appears wrong. it says to grab a pan first, while the meat is to be grabbed at first"; João Lopes and Jonas Olmstead both separately hit a "forgot the recipe" message they couldn't interpret. Four different people independently getting stuck on the same category of confusion, across the same one-week judging window, is a genuine interface signal rather than a coincidence of four unrelated players missing an obvious cue.

Controls: solid on a keyboard, genuinely uncertain on a touchscreen

Desktop testing was unambiguous and positive. Pressing W repeatedly moved the character cleanly across the kitchen floor toward an NPC, and the game correctly displayed a live "In front of you: [name]" prompt that updated as the player's position changed — real, working proximity detection rather than a static hint. The UI also swaps its displayed button icons between touch glyphs and a keyboard's "Space" label depending on which input was used last, a small adaptive touch that works well in practice.

Mobile is where this review has to be honest about the limits of its own evidence. A real emulated touchscreen drag on the on-screen movement joystick produced only a small, uncertain amount of on-screen movement — nowhere near as clear a result as the same testing method produced on other games' virtual joysticks in this same review cycle. That's not proof the mobile controls are broken; a synthetic touch-drag is an imperfect stand-in for a sustained real finger hold, and this review would rather say so than overclaim a verdict its own tooling can't fully support. What tips the balance is that a 2022 js13k judge reported the identical class of problem independently, in the judge's own words: "Mobile controls are a bit janky but plays well on pc." Testing here didn't contradict that; it just couldn't independently confirm it beyond a suggestive, inconclusive result.

Final verdict

There's a genuinely likeable game under all of this — the writing is better than almost anything else in its jam-game weight class, the art is charming, and taking 1st place in js13k's Decentralized bonus category for a real NEAR-testnet integration is a legitimate, citable achievement. But this shelf exists specifically to test whether a game respects the phone it's played on, and between a source-confirmed frame-rate bug that can double the game's speed on increasingly common high-refresh displays, recipe instructions that tripped up several independent judges rather than one confused player, and a first-party report of janky touch controls this review's own testing could only partially interrogate, the honest score sits just under the pass line. None of these are fatal to the writing or the art, but together they're more than "notable limitations" — they're the specific list of things this section is built to catch.