Soulmask

Most survival games make you do everything. Chop the tree, carry the log, feed the fire, repeat until the appeal wears off. Soulmask's answer is to hand the job to somebody else — and it means it literally. You capture and recruit tribespeople, each with their own traits and aptitudes, assign them to roles based on what they're actually good at, and they keep working while you're not there.

That's a real idea, and in a genre this crowded a real idea is worth a lot. It's also the thing that has caused Soulmask the most trouble, because a game built on NPCs doing their jobs lives or dies on whether the NPCs do their jobs.

First, the date on the store page is wrong

Steam lists Soulmask as a 9 April 2026 release. It isn't. The page's own Early Access date is 30 May 2024, and the great majority of its 22,119 reviews were written during nearly two years of Early Access.

This matters because 1.0 wasn't a launch, it was an exit — and it went badly enough to generate its own wave of criticism. Sorting the hundred most-helpful English negatives by quarter shows a steady trickle of six or seven per quarter through 2025, and then 31 in the second quarter of 2026, the 1.0 window. A game that had been quietly improving got its worst reception at the moment it declared itself finished.

The AI problem

Here's the number that defines this review: 72 of the hundred most-helpful English negative reviews mention AI. Not the combat, not the price, not the bugs — the tribesman AI, which is the feature the entire game is sold on.

The complaints are the ones you'd expect from a delegation system that doesn't quite work: tribespeople failing to path, failing to pick up the job they were assigned, standing around while production stalls. When it works, the loop is genuinely special — you come back to a working settlement rather than a to-do list. When it doesn't, you're babysitting the thing you recruited to avoid babysitting.

The interface is the second-largest cluster at 62 mentions, and one 1.0-era reviewer with 147 hours reports that some help screenshots were still in Chinese at full release, with a new AI-guidance overlay whose text appears and disappears too quickly to read. That's the kind of detail that tells you a release date arrived before the game was ready.

Combat is the third front. Recent reviewers describe enemies respawning the instant they're killed and a reinforcement mechanic that produces effectively endless waves — "there's zero sense of progress," as one August 2026 review puts it.

Multiplayer has its own history

The public-server story is worth flagging separately for anyone considering official servers. A 2024 review with 478 helpful votes and 260 hours describes server transfers being introduced and a group of fifteen-plus fully-geared players arriving to wipe an entire server. More recently, a player with 6,972 hours — the most invested reviewer in the corpus — reports that enforcement falls on ordinary players while cheaters go untouched.

None of that has to affect you. Soulmask supports solo play, private servers and LAN, and the 329-vote review that goes deepest on the game's systems explicitly brackets off public servers as a separate problem. But it's a genre where "just play private" is a real recommendation rather than a dodge, and here it is.

The 90-day overhaul, which is the actual news

What separates Soulmask from a straightforward "wait for a sale" verdict is what the developers did next.

On 5 June 2026, two months after 1.0, CampFire Studio published a 90-Day Overhaul Plan — a public acknowledgement that the release wasn't good enough, with committed dev logs. They've kept to it. Patch #13 on 11 June targeted balance and tribesman AI directly. Patch #14 on 26 June brought combat optimisations and world portals. Patches #15, #16 and #17 followed through July with a rebuilt building mode and HUD improvements. That's five substantial patches in seven weeks, aimed squarely at the three complaints above.

I'd rather see that than silence, and it's more than Megabonk managed with a similar balance problem. But the honest reporting is that it hasn't worked yet. Of the 100 most recent English reviews, spanning 6 July to 4 August 2026, only 65 are positive — below the 76.4% lifetime figure, with a median playtime of 104 hours. The overhaul is real and ongoing; the audience hasn't turned.

The expansion timing

One more thing sat badly. Shifting Sands, a paid Egyptian expansion at NZ$28.99 — two-thirds the price of the base game — launched the same day as 1.0. DLC appears in 21 of the hundred negatives.

Selling an expansion on the day you leave a two-year Early Access is a defensible business decision and a poor read of the room. Players who had waited through 2024 and 2025 for polish were offered more content instead, and said so.

The verdict

Soulmask earns a 6.5 — a score for a game in motion rather than a finished one.

The idea deserves better than its execution. Mask possession is a genuinely novel traversal and combat conceit, and the tribe-automation loop is the most interesting answer anyone has given to survival-game busywork since Valheim decided building should be structural. The world is handsome and doesn't look like its competitors. When the systems cooperate, it's a management sandbox with real depth.

What keeps it here is that the systems don't reliably cooperate, that 1.0 arrived early, that combat and interface work remain unfinished, and that the developers' own overhaul plan is an admission of all three. Current sentiment is worse than lifetime sentiment, which is the clearest signal a game can send that its present state isn't its best one.

Watch it. Buy it on discount if delegation and automation are what you want from survival, and accept you're buying into a repair job. And if the specific appeal is a world that keeps producing while you're logged off, that's the whole premise of Wanderer — where the workers don't get stuck on the scenery.