Steam reviews: issues & requests

FindingNot general instability — most reviewers say the game runs fine for them. What they attack is a narrow set of world-collision and boss-fight failures that cost an entire roguelite run: falling through the floor, getting stranded outside the playable area, enemies drifting out of the arena invulnerable, and crashes that land mid-boss.

• 76% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring run-ender is world geometry rather than performance: players fall through the floor, clip into terrain, or end up outside the arena with no way back — stated most precisely as “Fell through the floor of the world and fell forever, was forced to forfeit the run.” English and Japanese reviewers describe this independently, and the injury named is the lost run, not the glitch itself.
• Crashes and hard freezes cluster on boss fights rather than spreading across play — the final boss, “Freezes without fail mid phase 2 of the second boss,” and the zone 0 boss all recur across English, Russian and Chinese. The resentment is re-entry cost: one Chinese reviewer redid the zone 0 boss four times, each attempt preceded by shader recompilation.
• The same bounds failure applies to enemies, which is the half easiest to miss: Russian and English reviewers report mobs leaving the arena and becoming invulnerable, boss AI “bugging out” mid-fight, and engineers who “get stuck spamming grenades.” The result is a fight that stalls rather than one that is lost.
• Chinese reviews report a materially different bug set — four of nine mention aiming/scope faults (arms blocking the view when scoped, a general “scope bug”) or weapon-mod math that is exploitable rather than broken, such as sniper accuracy absorbing -200% in mods and charge shot behaving as a gold mod. English reviews almost never raise either.
• A sizeable share explicitly report clean runs — “I didn’t have any performance issues or crashing,” “I have had one crash but other than that it has run well” — and most bug reports come from reviewers who still recommend, framed as early-access roughness: “The game is still buggy and incomplete but the foundations are nearly perfect.” The sharper dissent is about fix speed, not bug count: “Gamebreaking bugs, basic QoL features, and so on are lacking in response.”

The 99 figure overstates real defects: many matches are the destruction mechanic (“crashing” buildings onto enemies), the freeze element, “stuck in zone 1” progression gripes, and the game’s deliberate glitch art style.

FindingThe complaint is not that the game is short in the abstract — it is that the run-to-credits happens on the first or second attempt, before the roguelite loop has shown players anything new, and that this is being sold at $30. What runs out is build and enemy variety, not play hours.

• 75% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring fact underneath the theme is credits on run one: "I beat the entire game, on the first run, straight out of the tutorial" is echoed by "beat all 4 zones my first run", "I got credits to roll on my first run", and third-run completions at around three hours in English, Spanish and Russian alike.
• Reviewers then describe a cliff rather than a grind, and the single most-helpful review in the theme states it most precisely: after the only three zones, the player's remaining experience is nothing but the challenge of higher difficulty — the same point as "Decently interesting and difficult gameplay for that first run and then it just drops off a cliff."
• What players name as thin is build and enemy variety, not zone count: "limited mod variety and not enough enemy types" recurs alongside modules "not worth taking", corruptions whose trade-offs don't pay, and one player beating the game "with the default pistol alone".
• The objection is a ratio, not a verdict on the game — "not enough content to justify its price currently" — and the qualifier recurs verbatim across languages: "for now", "currently", "目前", with $30 and 800₽ named as the specific sticking points.
• Three quarters still recommend, and most who raise this are deferring rather than refusing: "worth checking out either in the future when more content has been added", with early access and the solo developer cited as reasons to wait or to buy in support rather than as a warning off.
• The measuring stick is genre peers, not the game alone: "lacking content and depth that others in the genre are packed with", and "Most of the other great roguelites are throwing fun tools and powers at you from the very first run onward."
• A minority rejects the content framing entirely — "It's not really an issue of balance or even the amount of content, but the core gameplay that is a problem" — but that position does not recur.

The 99 mentions are not 99 negative verdicts: thinness is raised as a caveat inside recommending reviews roughly as often as it is a reason to refuse, which is why the theme coexists with 74.7% recommending.

FindingPlayers are not rejecting destruction-based combat; they object that staggering is the only route to damage, healing and ammo, which demotes guns to a finisher and turns fights into fetch-and-throw loops. The attack lands hardest on boss arenas, where throwing debris is the only thing that works at all.

• 54% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 1.4× more often than average (23% of their complaints)

• The recurring complaint is damage gating rather than the gimmick itself: enemies are “a massive bullet sponge, becoming marginally more killable if you stance break them,” and Chinese reviewers state it flatly as no stagger meaning essentially no damage. Zone 3's “armoured” enemies are named repeatedly as the point where the gate stops being optional.
• Stagger is also the only sustain, since health and ammo drop from staggered enemies, so players who would rather shoot still cannot skip the loop. On high difficulty the fuel runs dry — “there's no buidling left to crash and you have to rely on explosive boxes, which is fun until they run out.”
• Boss fights concentrate the complaint because the arenas have no adds to stagger: “Instead of shooting the boss like you would expect, you have to run around in circles throwing grenades at pillars and then you pick up rocks to throw at the boss.” The same circling-and-throwing description recurs in English, Chinese and Russian reviews alike.
• A specific, repeated mechanical fault sits underneath it: the carried chunk blocks the screen and grabbing one forces you to look away mid-bullet-hell, so “taking your eyes off of it to grab a hunk of debris only makes this worse.” Four separate reviews across English and Chinese raise the view-blocking debris independently.
• Chinese reviews attack the throw itself at a materially higher rate than English ones — the tether grenade has no arc and no stick, aiming is required but imprecise, and what registers as a stagger hit reads as arbitrary. English reviewers complain the loop is mandatory; Chinese reviewers complain it also misfires.
• Several of the harshest critics open by praising the concept, including one who calls the destruction and stagger combat “refreshing,” which places the fix in how hard the game forces the loop — stagger resistance scaling, armour, throwable scarcity — rather than in removing it.

52.6% still recommend and several of the most precise complaints come from recommending reviews, so this reads as a design objection inside a liked game rather than a rejection.

FindingPlayers are not complaining that the game is hard — they are complaining that there is nothing below the difficulty that is beating them. The default Normal is also the lowest setting, and the hardest thing most players meet in it arrives first, at the stage-1 boss.

• 58% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 2.2× more often than average (37% of their complaints)

• The ask is for a tier beneath the current floor, not a softer game: reviewers note Normal is both the default and the lowest option and still stops them — “it's too tryhard and punishing, even on the lowest difficulty setting.” Chinese reviews name the absence directly: “unreasonable difficulty, and no difficulty tiers at all.”
• The wall is specific and recurs across languages: the stage-1 boss. Several reviews end there outright (“This game is a-bit too hard for me and I can't get past the first bossfight”), and multiple players argue the curve is inverted rather than steep — one reports clearing the stage-3 boss at full health by standing under it throwing grenades.
• What “punishing” means here is enemy health outrunning upgrade power, not player fragility — “in the later stages the enemies health scale too high for the upgrades to catch up” — which is why the repeated description is sweaty kiting and dodging rather than dying to one mistake.
• The economy is blamed as often as the enemies: slow growth, thin early resources, and runs where the shop is never reached once, so players report the run is decided in the first two rooms before any counterplay exists.
• Chinese reviews (22 of 78) raise this at a materially different altitude than English ones — they argue the numbers (scaling at high chaos, in-run gold, the roughly 5% stage-4 boss clear rate seven months in), while English reviews mostly report the on-ramp: “it's not balanced for casuals at the moment.”
• Recommenders file this complaint rather than dissent from it — “The game is punishing, but incredibly fun” — so what is under attack is the entry tuning, not the combat or the bosses, which the same reviews praise.

57.7% of the 78 still recommend, so the count measures engaged players asking for a lower entry tier, not players writing the game off.

FindingThe complaint is not that there are too many projectiles but that they can't be read: players can't judge which shots will actually reach them, how far away a shot is in depth, or when a homing shot has locked on. The density is largely accepted; what is attacked is the missing information layer on top of it.

• 56% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 1.5× more often than average (25% of their complaints)

• The recurring core is depth and threat legibility rather than volume — “Some of the enemies projectiles look like they won't hit you but they do.” Several note that drawing projectiles through walls is “one saving grace” until the screen is “pretty much 90% projectiles” and front-of-wall and behind-wall shots stop being distinguishable.
• Homing shots are the single most-named offender across english, russian and chinese, and the grievance is that they are unsignposted and cancel the defensive options the arena offers — they “trace you around the rocks anyway,” and chinese reviews list tracking fire beside “cover-piercing” and sniping fire as the set you are asked to dodge at once. The concrete ask is a marker when a tracking projectile is locked on, not its removal.
• The clearest fixable instance is the boss heal/stagger loop: the debris you must pick up and throw occludes the screen during the densest bullet phase, so the required mechanic and the required dodging fight each other. Stated most precisely as the “fragment grabbing” mechanic “blocks too much of the screen, making it hard to see the boss's movements,” and echoed by a player who says the chunks “often get you hurt more than if you were to just lock in and dodge.”
• The proposed remedy recurs independently of language and is legibility-shaped, not difficulty-shaped: fewer but larger and cleaner projectiles, summarised in russian as reducing their number and increasing their size “for better reading of the situation.” One player ties this to the game's own art direction, noting the universal ‘glitch’ effect on projectiles adds “an absurd amount of visual noise” over the bullets themselves.
• Chinese reviews report this at a materially different rate and frame it differently: nearly all of the chinese quotes here are do-not-recommend, and they describe unreadability as simultaneous demands — healers, drones, snipers and dashing hounds layered over bullet walls against a short dash cooldown and brief i-frames — rather than as a visual-clarity problem.
• A smaller but cross-language thread is telegraphs under-representing real hitboxes: the first boss's charge shows “only that small” damage radius yet still hurts outside it, matching the english report that shots which look like misses connect.

78 mentions at 56.4% still recommending, and several of the highest-voted readability complaints come from reviews that recommend the game — this reads as a fixable presentation layer, not a verdict on the combat.

FindingPlayers are not objecting to meta progression itself — they object to a double gate on it: a node unlocked is still inert until a second, scarcer currency levels it, and that currency is at its scarcest in the opening hours when nothing is unlocked yet. The word "grindy" is shorthand for runs that end too early to pay for the upgrades that would make runs last longer.

• 49% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 2.1× more often than average (28% of their complaints)

• The most precisely stated version of the shared complaint is the two-step gate: “They don’t feel great, especially since when you unlock them individually, you don’t get the benefit of them until you use a *separate* in-game currency to level it up.” Others hit the same wall on the first health tier and on the hub defensive buffs, which must be both unlocked and then upgraded.
• Scarcity is front-loaded rather than back-loaded, which players name as the structural error: “Roguelites like this should see tons of early game progression and then get grindy as you get further, this seems to be the inverse.” Concrete markers recur — six runs for a first unlock on normal, one schematic found across 39 runs in zone 1.
• When upgrades do land, the increments are too small to register against the difficulty, so the grind reads as unpaid: “the meta progression is painfully slow with tiny increments it doesn’t feel like it’s doing anything.” A first health level of “only 10% or 20%” and “the heavy grind makes the upgrades feel pointless” say the same thing.
• Simplified Chinese reviews report this at a different point in the loop — they target the in-run economy rather than the hub tree. In-run chests are spent on meta currency instead of build pieces, and in-run gold is so thin that “you might not get into a shop even once in a whole run — so what is the shop there for?”, leaving runs with no strong build to assemble.
• Players treat the grind as a consequence of difficulty, not a separate dial: “General runs last 2-3 rooms before wiping, as health / ammo / currency is scarce and enemies hit hard.” Chinese reviews state the loop directly — high difficulty, low rewards, extremely slow growth — with player scaling falling behind trash mobs at high chaos.
• Drop rates are called unreliable as well as low, and buying into better rates does not fix it: “Even with the upgrades to get more currency, everything costs way too much”, and research still “very rarely drops anything.” One player locates the flaw exactly — a currency crate can yield nothing, so “the floor of obtainable currency for individual sources is zero, which should never be the case.”

The 49.2% recommend rate splits the same evidence two ways: recommenders raise sparse cross-run currency as a caveat they expect to outgrow, non-recommenders as the reason they stopped in zone 1.

FindingPlayers are not mainly asking for more content; they report that the variety the game shows up front — mod pickups, enemy silhouettes, a weapon list — collapses into one way to play within the first two hours. The fixable core is duplicate and lower-tier mod rolls, enemies that differ by look but not by how you fight them, and a fixed stage order.

• 58% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 1.7× more often than average (22% of their complaints)

• The mod pool reads as deep at pickup and then withdraws it: chests return duplicates or lower-tier copies of what you already carry, and much of the rest is judged bad or too niche to build around. One reviewer states the shared complaint most precisely — “the illusion of choice and variety shatters pretty fast.”
• Enemies are counted as varied by silhouette and felt as identical in play: “There are very few enemy types, and in terms of how you engage with them it feels like even less.” A Chinese reviewer reduces the whole roster to three functions — the healer, the shield-bearer, the bullet-spewer — and says it makes no difference whether the shooter is a dog, a person, a spider or a flying ball.
• Elemental immunity walls are the one point Chinese reviews raise at a materially higher rate (2 of 13, against 1 of 49 in English), and both name the same moment: a frost build assembled across two stages meeting stage-three enemies immune to freeze. The objection is not difficulty but that committing to a build is punished retroactively, with no read on what to commit to — hence the request for a mod compendium.
• Part of the sameness is structural rather than a content count: the areas run in a fixed order and the same roughly identical arenas and warp rooms recur, so “every run feels more or less identical” before the mods are even weighed.
• Guns separate less than the weapon list implies — “all the weapon variety I experienced boiled down to different firing modes and ammo types… for the pistol.” Several tie this back to there being only one viable way to engage, which flattens the guns rather than the guns themselves being too few.
• The fault is in how quickly variety is spent, not in the opening: the game “makes a really good first impression,” then “isn't enough to stay interesting for more than an hour or two.” Recommenders differ from critics mostly in reading this as content the devs will add, which is what holds the recommend rate at 58%.

Most of these reviews are logged at two to five hours played, so the 62 mentions mark where interest stops rather than a veteran verdict on the full build pool.

FindingPerformance is not a general failure in this game — most reviewers volunteer that it runs better than they expected for a UE5 destruction shooter. The complaint is a narrow severe tail: a minority hit aggressive stutter and low FPS that lowering settings does not fix.

• 77% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring note across the quoted reviews is praise, often phrased as surprise: “Its rough around the edges, but no crashes, suprisingly good performance, and smooth gameplay.” Several specifically credit it for holding up when the screen fills with debris, explosions and projectiles.
• The genuine complaint recurs in a small group whose problem is settings-proof: “Even with the lowest settings possible and everything turned off, I regularly hit low FPS.” For them it is not tuning but unplayability — stutter described as aggressive and constant rather than occasional.
• Between the two sits a steady middle that runs fine and still flags headroom, best stated as “optimization is meh though, it runs better than most UE5 games for me but it's definitely a concern.” This is a reservation about margin on weaker hardware, not a report of a broken experience.
• Where otherwise-smooth players do see drops, they are localized rather than global: later-released areas, long sessions (“performance degradation after several hours of play”), and a Russian reviewer's few-second freeze when a boss fires a volley at point-blank.
• Steam Deck recurs but is not a framerate complaint — it is reported to run well there while maxing the GPU and draining battery, and the objection to its Verified badge is about difficulty, not performance.
• Non-English mentions track the English middle rather than diverging: the Chinese verdict is “average longevity, and decent optimization, but it has quite a few issues”, and both Russian reviewers report no optimization problems at all.

The 59 count overstates the complaint: most of these sentences are praise, and several are keyword false positives about gun-mod grid optimisation, weapon feel, or a rant aimed at other AAAA studios.

FindingPlayers largely like how the guns handle; what they attack sits on either end of the shot — a recoil animation that yanks the camera rather than the crosshair, and targets that absorb bullets without visibly registering them.

• 49% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 3.0× more often than average (35% of their complaints)
• English reviewers raise it 1.5× less often (8%)

• The recoil complaint is about the camera, not accuracy: the viewport is dragged up while firing and pulled back down the full distance when you stop, so anyone who compensated for the climb is left staring at the floor. A second player states the split most precisely — “the recoil animation is visually excessive even with no mods but the effect of recoil itself is negligible on your aim”.
• Impact is the other recurring word, and it concerns the target rather than the weapon: guns are called “pellet shooters” and “lackluster and non-impactful” because enemies do not react to being hit and the stagger combo damage depends on decays until it is moot. One player names the fix himself — “a small recoil stagger (not as in the mechanic but just them reacting) would fix this”.
• Shots that should land do not: hit reg is called “abysmal” on the AR and shotgun, and another's chief frustration is how often “I was shooting something, my crosshair was dead-on and the enemy just”. One review traces this to default projectile speed — bullets slow enough that you must lead enemies, each gun at a different rate — but only that one player connects the two, so treat the link as unconfirmed.
• Chinese reviews attack aiming itself at a materially higher rate than English ones, which concentrate on recoil: ADS and scopes are judged near-useless because “many fast enemies, plus bullet patterns, make aiming pointless” and scopes serve only for weak points. Controller players add a large stick deadzone and aim assist that applies only while scoped — a usability complaint about ADS fitting this game's combat, not about recoil.
• Recoil doubles as the damage currency and is priced inconsistently in both directions: one player refuses a mod granting “ONLY 20%” damage for −35% handling, while another reports recoil-increase mods are almost cosmetic for hipfire, letting a sniper stack damage downsides at no real cost.

The 54 mentions are not 54 complaints — several of the most-upvoted gun-feel sentences praise handling (“the guns feel good”, “pleasant shooting”) inside reviews that do not recommend, so the keyword count overstates how negative this theme is.

FindingPlayers are not rejecting stagger-for-health; they attack it being the only source of health and ammo, in arenas where the environment it depends on runs out and where setting up a stagger costs more health than the drop returns.

• 37% of the reviews raising this still recommend the game
• Simplified Chinese reviewers raise it 1.9× more often than average (17% of their complaints)

• The mechanic is repeatedly praised before it is criticised — “Reliance on getting health and ammo from staggered enemies is fun, until there's no buidling left to crash” — so the complaint lands on it being the sole source, not on its existence.
• The recurring failure is that healing competes directly with surviving: toppling scenery or lining up a throw means turning away from enemies and standing still, and reviewers describe being punished for it. “If you focus on trying to get health (making buildings fall on enemies so when you kill them you get health) then you get mobbed and destroyed by them all.”
• Supply is finite and arena-bound — only a few destructible structures per area, explosive boxes that run out — so the heal rate collapses exactly as difficulty rises and the run becomes, in one reviewer's words, “pretty much just a matter of time until you inevitably die.”
• Boss arenas are where it breaks hardest, and this recurs in all three languages: the only heal is throwing debris at the boss, and the debris blocks the view that dodging requires. Russian: “you shoot the boss for 10 seconds, then run in circles for another 30 throwing rocks at it so it gives you a medkit.”
• Chinese reviews more often attack the arithmetic rather than the setup cost — “掉血非常厉害，补血削弱根本补不上来” (you bleed health fast and the healing cannot keep up), with drops from kills giving “no positive feedback” at all. One Chinese reviewer states a longer gate chain most precisely — “要打失衡才能回血，打失衡又要先破盾，盾又贼多” (to heal you must stagger, to stagger you must first break shields, and shields are everywhere) — but that specific chain is one voice, while enemy healers recur in Chinese and English alike.
• Drops that do land often fail to register as recovery: orbs expire, a single barrage strips more than a pickup gave, and without upgrades there is no heal between levels. Ammo sits behind the identical gate — “Can only get health and ammo by throwing objects at them” — though several ammo complaints are plain scarcity (low magazine capacity, nothing buyable from traders), and the build players call overpowered is the infinite-ammo module that opts out of the economy entirely.

41 mentions with 36.6% still recommending — up from 39 and 33.3% — still reads as friction inside a liked system rather than a rejected mechanic, with several recommending reviewers endorsing the stagger loop itself once mastered.

FindingThis is not mainly a complaint that the movement feels bad — the traversal kit itself draws praise, and 60% of these reviewers still recommend. What recurs is the controller layer wrapped around it: aim assist with no working off switch, menus and equipping that still require a mouse, and a bind count that fights the right stick.

• 60% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• Controller support is advertised as full but is incomplete in specific, nameable ways: several settings cannot be changed with the controller, equipping weapon variants still needs the mouse, and there is no vibration. One player states the shared grievance exactly — “when a game says ‘full controller support’ I really expect the controller to actually work on the game.”
• Aim assist is the single most repeated controller complaint and the most fixable: too strong, no strength slider, and it drags the view onto enemies through buildings instead of the one in front. The German reviewer escalates it rather than merely repeating it — he turned aim assist off and felt no change, so the existing toggle may not take effect.
• “Floaty” is close to an English-only word here: five English reviewers use it, from “the movement does feel a bit floaty, though” up to “almost unplayable.” Chinese and Russian reviewers mostly praise the same kit (Titanfall 2 and Ghostrunner comparisons, grapple and slide called silky) with one Russian exception calling movement swampy, so treat floatiness as a feel complaint in one language, not a cross-language verdict.
• Where non-English reviewers do attack movement, they name mechanics rather than feel, and the same two recur: the grapple cannot be cancelled mid-flight (repeatedly overshooting into the water on the Area 03 boss map), and dash cooldown is long while its invulnerability window is too short to chain dodges against boss attacks.
• Button count is a comfort and layout problem, not a learning-curve one: players describe juggling roughly seven binds beyond the normal WASD set and cramping their hands, with Sprint Toggle only partly helping. The concrete ask is consolidation — one reviewer suggests a combo key for skills to free a slot, and asks why a controller shooter triggers the ultimate by long-pressing the grapple button.
• The first boss is where these limits become visible, named independently by a controller player and by the most-upvoted Chinese review. On controller the fight requires pressing Y, moving the thumb to aim, then leaving the stick to press Y again while the boss keeps moving — a binding conflict, not a difficulty complaint.

25 mentions against 38 sentences: the top-weighted review (12 helpful, Chinese) supplies five of them, so count it as one voice rather than five.

FindingPlayers are not attacking the grapple itself — they consistently call it fun and effective — but its constraints: anchors only at fixed points, no steering or cancel once attached, and no wall-running alongside it. Two thirds still recommend, and most of these read as feature requests rather than grievances.

• 67% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The hook itself is not what gets attacked: a reviewer who does not recommend the game still says “The grappling hook is effective and fun to use,” and the most-helpful mention calls the fixed-point limit “only a tiny let down, because it's still very fun to swing around.” The complaint is about where and how it can be used, not whether it works.
• Fixed anchor points recur as a wish, not a grievance — “如果钩锁不是固定绳索点而是任意连接点那就无敌了，可惜不是” (if the hook used any connection point instead of fixed rope points it would be unbeatable, a pity it isn't), while another simply asks for “a grapple point at the top of the drop downs.” Every reviewer raising this still recommends the game.
• A separate and more fixable cluster is the lack of control once attached: “the grappling hook you get later on just flings you sometimes in whatever direction you were hooking in, no control over it at all.” Others ask to cancel the grapple early and steer mid-swing, and report it bugging out or “anchoring you to the ground” with certain modules equipped.
• Wall-running is an expectation complaint imported from Titanfall rather than a gap players feel on its own terms: the one non-recommend on it says the comparison fails because “there isn't even wallrunning to be had here,” while a 129-hour recommender concedes wall-runs are “not the only defining feature of Titanfall 2s movement.” Four reviews across English and Spanish name it.
• One Russian recommender credits an update with adding it — “Добавили бег по стенам” (they added wall-running) — which single data point suggests the English and Spanish wall-run complaints may predate the patch; worth checking timestamps against the update before treating that gap as open.
• Chinese reviews raise wall-running zero times out of four; their grapple complaints are about level design and readability instead — a boss arena of “就几个小岛，距离还远到必须用抓钩” (just a few islands, far enough apart that you have to use the grapple) and the hook's debris pull blocking a quarter of the screen.

Roughly four of the 18 mentions are praise or neutral references caught by the keyword rather than complaints, so the real complaint base is nearer a dozen split across three distinct asks.

FindingCo-op is a wishlist item rather than a defect: fifteen of the seventeen mentions come from players who recommend the game and ask for multiplayer as an addition, and only two treat its absence as a reason to play something else.

• 88% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The ask is almost always phrased as a wish sitting on top of approval — “I only wish this game had co-op support, this could be fun with friends” — not as something missing from what was paid for.
• English reviews pre-excuse the developer: “I dont know how realistic that is for a solo dev”, “understandable if its not possible”, and one notes he “says he has no plans for multiplayer” yet recommends anyway.
• Correction to the previous reading that two Chinese reviews were too few to call: all four Chinese mentions now recommend and ask flatly, with none of the English hedging — one repeats “suggest adding an online mode” three times in a row, another is just “Please add a multiplayer mode”. At 4 of 17 the Chinese register is a direct feature demand, not a request carrying its own permission to be refused.
• What recurs is PvE co-op specifically, not PvP — “not PVP just the game with others and maybe harder per player”; PvP appears once, as a daydream about a sequel.
• Co-op is attached to longevity rather than to moment-to-moment play: it surfaces once the content runs out — “it becomes dull after clearing” — and as the thing that would make it “hit ror2 type of fun”.
• The two non-recommending reviews (both 1h played) use co-op comparatively, naming Mycopunk, Deadzone and Witchfire as where they would go instead — co-op decides which game gets played, not whether this one works.

The theme's loudest review is no longer a refusal: a recommending Chinese review that asks three times for multiplayer now tops it at 10 helpful, ahead of the two non-recommenders at 8 and 6.

FindingOnly a handful of named combos actually trivialize anything, and the objection to them is not that the game got easy but that they eclipse every other build. The larger share of sentences filed under this theme runs the opposite way: the zone-1 boss and high-difficulty enemies are tuned so hard that a broken build is the entry fee rather than a way around it.

• 56% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker
• Simplified Chinese reviewers raise it 4.5× more often than average (15% of their complaints)

• The trivialization is specific, not general: four reviews name combos — scatternet/orb-e with adds, the mini-rocket, sniper damage mods stacked with the infinite-ammo skill, and grenade-spam from under the zone-3 boss. No one describes the game as broadly too easy.
• The shared grievance is dead options rather than lost challenge, stated most precisely as “meta remains the same with scatternet/orb-e/+adds being able to instakill bosses and it overshadows all other builds, especially on higher difficulties.”
• Reviewers read the floor as the problem as much as the ceiling: most weapon variants are worse than the base weapon, so “picking one is the same as raising the difficulty,” and one legendary mod is “over-tuned compared to the other legendary mods in the game.”
• Against the theme title, more sentences here describe over-tuned enemies than over-powered players — the three-phase zone-1 boss faced with almost no gear, later difficulties “completely over-tuned,” and the flat statement that “you need a broken build to be able to kill the boss.”
• Every explicit nerf request in this set targets enemies, not builds — the zone-1 boss, “nerf the monsters” — and one reviewer asks the developer outright to leave the strong builds alone and add more module combinations instead.
• Chinese reviews carry all but one of the seven non-recommends and frame the issue as an inverted curve (hardest at zone 1, cleared at full health by zone 3), while the English and Japanese reviewers who name the overpowered build mostly still recommend and call it fun.

Of the 16 mentions, roughly four describe actual trivialization; the keyword sweep also pulled in an aim-assist “too strong” complaint and a healing-nerf complaint that are not about mods at all.

FindingPlayers do not object to corrupted mods as a gimmick — the objection is that these mods buy their power with the handful of settings people hold fixed for comfort (motion blur, FOV, resolution, sensitivity), so the drops that pay most are the ones they either cannot use or are made miserable using. The concept is repeatedly praised in the same breath as the implementation is rejected.

• 46% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The mechanic itself reads as a highlight, not a flaw: recommending reviews call the settings-scaling mods “the crazier ones like damage boosts from lowering resolution, what the motion blur value, or your sensitivity setting is are cool as hell” and “the corrupted perks are a standout,” and even a negative Chinese review grants the sensitivity-damage module is an interesting idea.
• What recurs is the same four settings every time — motion blur, FOV, resolution, mouse sensitivity — and the complaint is that these are comfort settings, not strategy knobs. The Japanese review states the shared point most precisely: settings exist to be optimized to each user's own preference to improve the user experience, so making them determine in-game strength inverts what they are for.
• Motion blur is the most-named offender and the only one with a physical cost rather than a taste cost — “Remove the motion blur corrupted item, causes headaches and feels bad if you don't use it, so it's pretty useless.” That one sentence carries both halves of the theme: unusable for some players, and worthless precisely because it is unusable.
• The practical result is a dead drop rather than a real trade-off, which is why the complaint lands as reward-pool contamination: a Chinese reviewer asks why such mods are put in the run to “污染奖励池” (pollute the reward pool) instead of being unlocked outside it.
• The pressure concentrates at the top rarity, where the conditions stack rather than appear singly — one review lists a single tier demanding constant jumping, lowered FOV, lowered resolution, raised motion blur and raised sensitivity at once, so the best-paying tier is the least usable one.
• Only the Japanese review extends the objection to real-time conditions, saying it half-forces late-night play alongside the unwanted settings; one reviewer is not recurrence, so treat it as an adjacent edge of the same design rather than a second complaint.

13 mentions overstates the cohort: at least four quoted sentences are keyword matches about graphics settings, NVIDIA-only upscaling or performance troubleshooting with no bearing on corrupted mods, and the 46.2% recommend rate reflects players who praise and resent the mechanic within the same review.

FindingThe complaint is not that the game is random but that module drops alone decide whether a build ever comes together, and the weapon is locked in before the run starts so mismatched mods cannot be steered back toward it. Players like the build ceiling; what they attack is having no influence over which modules they are offered.

• 83% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The weapon-modding system is the praised part, which is why the mod rolls sting: "Being able to so heavily spec out your weapons into genuinely insane combinations is fun, though you do admittedly sometimes get shafted by RNG with terrible mods that don't synergize whatsoever." Another puts the split plainly — the guns feel satisfying "but you need to luck out on mods to make em work."
• One review names the mechanism most precisely: "武器只能在开始游戏前选择，游玩中如果运气不好很容易拿到一堆与当前武器不匹配的道具" (the weapon can only be chosen before the run, so bad luck leaves you holding a pile of items that don't match it). The grievance is the locked weapon plus unsteerable drops, not drop variance by itself.
• What is asked for is direction over the pool rather than less randomness, and the existing mitigation is named and judged insufficient: "I also wish there was a greater sense of player direction when it comes to what modules you get, rather than complete RNG; even with the reroll chest upgrade, it's too much RNG."
• Chinese reviews measure the problem as run outcome at the first area rather than as build feel — three of four say bad rolls mean you can't reach or beat the zone-1 boss, one contrasting a lucky second run that cleared the whole game with following runs stuck there. Both non-recommends in the theme are Chinese.
• The frustration is shaped as waiting, not quitting: several describe grinding runs "until the RNG gods finally favor you," framed as the accepted roguelike bargain pushed one notch too far, which is consistent with ten of twelve still recommending.

One of the 12 is a sarcastic complaint about enemies spawning outside the arena that matched only on the word "luck", so the real theme size is 11.

FindingThis is not a balance complaint but a sequencing one: players who clear the final boss on their first or second normal run want the harder modifiers available and locked on from the start, instead of earned after a clear. The modifier system itself is accepted as sufficient.

• 91% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring specific behind “too easy” is first-clear speed, not moment-to-moment combat — several players report finishing the final zone on run one or within “2-3 runs,” one “on my first run with no upgrades.”
• The ask is a locked-in hard start rather than a rebalance, stated most precisely as “I hope you add a ‘hard start’ mode similar to the original Hades — ... certain difficulty modifiers are basically forced and locked on from before you can even adjust the others,” and echoed by “a more difficult setting that’s accessible from the start.”
• Players defend the existing modifier system rather than attack it: the most-helpful review in the theme argues “the dev shouldn’t be afraid of making the game too easy because the player can scale the difficulty,” and another notes everything above base is “hard as balls.” The ceiling is fine; the entry point is what’s contested.
• A distinct second strand is difficulty sagging as a run progresses — “Act 2 (Zones 4-5) ... feels noticeably less challenging than Act 1 (Zones 1-3),” attributed elsewhere to builds that “scale effortlesly,” leaving the final boss “a walk in the park on normal.” That is a within-run curve problem, not the same request as a harder start.
• Tone is requests from satisfied players, not churn: 10 of 11 recommend, and several volunteer that difficulty is “perfect” or “very fair, even though the default option was a bit easy.” Only one player links the easy clear to not returning, and that review also blames enemies and areas that “looked the same.”

At 11 mentions this is a thin theme, and its single non-recommending review is defending the easy default rather than complaining about it.

FindingThis is not a complaint about sound design quality — players praise it unprompted — but about level and high-frequency content: specific sounds are piercing enough to cause headaches over a session, especially on headphones. It reads as a mixing problem sitting on top of work people otherwise admire.

• 50% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring shape is praise followed by a wince: “I do want to mention that the sound design in this game is top notch, but it is HARSH as hell.” The objection is to loudness and treble, not to the design itself.
• Two triggers are named specifically rather than in general: hit markers when damage stacks — “when you're hitting multiple enemies with different sources of damage it's too loud” — and the shared UI/menu sound texture, which one player extends to the hound death sound.
• The harm reported is cumulative and physical, not an immediate dislike: headaches in long sessions and roughness on headphones, which is why it surfaces from players who are recommending the game anyway.
• Players volunteer the fix themselves and scope it small — “a very small and fixable issue” — treating it as a mix and EQ pass on a few high frequencies rather than a rebuild.
• The ceiling is worse than the average: one reviewer reports they “can't actually play the game unless it's fully muted,” so for a minority this is a hard blocker, not an irritant.
• Only English reviews mention it, so there is no cross-language signal to weigh; the 50% recommend split is not attributable to sound alone, since the harshest wording comes from both recommenders and non-recommenders.

Three of the ten mentions match on “grating”/“headache” but are about the companion VO, the motion-blur item, and the overall experience — the audio-harshness theme is really about six reviews.

FindingThe complaint is about quantity, not craft: the campaign ends in roughly two hours and stops on a "to be continued", and players who liked what they played want more of it before early access closes. Nobody attacks how the story is told.

• 88% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• Shortness is logged as the lone exception to an otherwise positive verdict — "Only bad thing i really can say is i wish there would be more story" and "The only thing is the current length of the game" — which is why 7 of 8 of these reviewers still recommend.
• The writing itself is explicitly spared: "Story is short, but well told", and the one Russian review says the same thing, that the plot is short but there, with an ending that is not quite "WOW" but fine. The attack is on the amount, not the telling.
• Most reviewers date-stamp the complaint — "Quite a short story (for now)", "the current length" — treating the length as an early access state they expect to be filled in rather than a finished game's failure.
• The cliffhanger is a separate irritant from the runtime: the ending is read as a setup for a sequel, and one player calls that "to be continued ... kinda mean" while still saying they would buy the follow-up.
• The only recurring concrete ask is more content ahead of the existing ending: "Please extend the game before the twist at the end of early access, that would be nice, one or two areas more would not hurt."
• The single hard number — 106 minutes for a first, uninterrupted completion — comes from the one reviewer here who does not recommend, so length alone is rarely enough to flip a verdict.

At 8 mentions this is a consistent minor footnote rather than a groundswell, and a single review carries all of its negative weight.

FindingTwo different complaints sit under one title: Chinese players, who already have a translation, fault its completeness and typography, while Russian and Japanese players are simply asking for a language that does not exist yet. The cost is concentrated in the first group — both non-recommending reviews in the theme are Chinese, and both are about the quality of the shipped translation rather than its absence.

• 86% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The shipped Chinese is incomplete rather than wrong-headed: one player reports that “part of the in-game text is not fully translated, some of the translated text has errors in meaning, and the simplified Chinese font layout is offset”, and another confirms that “some places in EA still seem untranslated”.
• Presentation is a second, separate Chinese complaint: the menu title in Chinese is described as five small characters shoved to the left of the character, which “gives a cheap impression, out of keeping with the quality the game itself shows”.
• The translation gap carries a playability cost, not just a cosmetic one — the same reviewer notes that “the guidance is not clear enough … because the text translation is not up to standard, the guidance also misleads players”.
• Requests for absent languages are insistent but not blocking: “Main problem: RUSSIAN LANGUAGE IS NEEDED!” — yet every reviewer asking for a new language still recommends the game. A second Russian treats it as temporary and notes a fan-made Russian patch already circulates, and the Japanese reviewer says it is “no problem at all for anyone used to playing games in English”.
• The split runs by language, not by intensity: all four Chinese mentions concern the quality of a translation that exists and contain both of the theme's negative reviews, while the Russian and Japanese mentions concern absence only and all three still recommend.
• The loudest voice in this theme is credit rather than grievance — the 56-helpful Chinese review says that “for an indie developer, being able to add Chinese beyond English at the Early Access launch deserves big praise”.

Seven mentions is a thin base, and the single most-helpful one is praise, so the theme's count overstates the dissatisfaction behind it.

FindingPlayers are not attacking the module system — they are asking for an exit from it. Unwanted modules accumulate with no way to destroy, convert, or reduce the odds of getting them again.

• 100% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The clearest recurring want is a sink for surplus mods: "I wish there was a way to recycle some of the mods you get though because on my winning run I had like 10 that just sat in my inventory." A second reviewer asks for the same thing by name, "scrap down modules that you don't want (similar too gunfire reborn)."
• The other recurring strand is steering the pool rather than the inventory — "limited ability to influence your mod pool" and a wish to "manipulate drops a little bit." Both are the same underlying complaint as scrapping: no lever on which modules you end up holding.
• One reviewer reports unequip simply not working, but hedges it himself — "Might be missing something here but I tried to unequip and it didn't work." On a single hedged mention this cannot be separated from a discoverability problem, so it is logged, not concluded.
• All six reviewers still recommend the game and frame this as a future-update wish, not a defect; one request for a collected-modules compendium is adjacent wishlisting rather than the same complaint.
• Nothing here says the modules themselves are bad — the surplus exists because the drops keep coming and nothing consumes them.

Six mentions, all within the last day and every one at zero helpful votes, so this is an early, unvalidated signal rather than a settled theme.

FindingPlayers aren't asking for damage nerfs — they're saying the non-rifle weapons have no legible role, so the assault rifle becomes the default answer rather than one choice among several. The sharpest version of the complaint points at the shared upgrade pool, not the guns' raw numbers.

• 100% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The assault rifle as the only viable pick recurs independently in both languages: "the only one I see worth using is the assault rifle" and "只有步枪能玩" (only the rifle is playable). This is the spine of the theme — every other point attaches to it.
• Shotgun and pistol are the two named casualties, each cited twice across English and Chinese: "shotgun could use some love," "The pistol is not great," and "喷子手枪射手全是废物" (shotgun, pistol, marksman are all garbage). No reviewer names a weapon they want weakened.
• The most precise diagnosis is structural rather than statistical — one Chinese reviewer expected per-weapon upgrade pools and found a shared one: "我还以为会像塔那样有专属卡池，结果是共通的" (I thought there'd be dedicated card pools like in the Tower, but they're shared). That reframes the issue as weapons having nothing of their own to build toward.
• Both English reviewers hedge toward self-blame — "Maybe I just suck and making it work" and "maybe im just not understanding the power behind the DMR" — which reads as weapons failing to communicate their role rather than players giving up on them.
• Only the Chinese reviews escalate beyond balance: one ties the shared pool to run longevity ("让游戏寿命少了不止一半"), the other to feel and spread ("喷子那个散射更是大的吓人" — the shotgun's spread is frighteningly wide). English mentions stay at the level of one weak gun.
• All five reviewers still recommend the game, and one frames the cost as early pacing — the first zone is "a slog until you can get the auto rifle" — so this is a fix request from people who are enjoying it, not a verdict against it.

Five sentences, four of them with zero helpful votes and all from reviews under 24 hours old — the pattern is consistent but thin, and only the 9-helpful Chinese review has been read widely.