Emberville Ultimate Build Guide: Classes, Weapons and Every Choice Compared

Nobody outside Cygnus Cross knows how many classes Emberville ships with.

Two are named. A fan wiki, summarising a 2025 devlog, says the studio once aimed for thirteen, and hedged even that. Early Access opens on October 27. That gap is the whole problem with writing a build guide today.

So here is what I'm doing about it. Every claim below is labelled by where it came from. Every verdict carries a confidence rating. Anything I can't back is marked as a guess instead of dressed up as fact.

A guide that crowns a best class before the class list exists is bullshit. I'd rather be honest and provisional than confident and fake.

After October 27 the guesses get replaced with tested answers, and every change goes in the log at the bottom of this page.

Provisional Picks at a Glance

Here is where I land today, on the six choices people are already asking about.

Low means I'm reasoning from previews and the shape of the systems. Medium means Cygnus Cross's own design backs the pick. High means I've tested it myself. Nothing on this page is High yet, and I won't pretend otherwise.

One disclosure before the opinions. CohhCarnage co-owns Cygnus Cross and is a creative partner on the game's design. Cygnus Cross is the developer and publisher. I'm from Cork, and so is the studio, which means I want this to be good. That's a bias, so I've left the criticism in.

What This Guide Is, and What It Isn't

This page was written on October 4, 2026. Emberville enters Early Access on Steam and GOG on October 27. I haven't played that build, because it doesn't exist yet, and I'd be an eejit to write as if I had.

What I have is a pile of sources of very different quality. Each section tags its claims so you can see which is which:

• Official: the Steam page, the Early Access FAQ on it, the official screenshots, and Cygnus Cross's July 16 press release.
• Preview: hands-on coverage of early builds, mainly Mobalytics, whose writers played under two hours.
• Footage: the demo footage behind our Prologue Demo review.
• Fan database: unofficial wikis such as meowdb and thegameswiki. These are leads. They repeat press material and add detail nobody has verified.
• Inference: my reasoning from how the confirmed systems fit together. Treat it as a bet.

Our Prologue Demo review covers what the early build is like to play. This page is about the decisions.

The Steam store page is the place to wishlist it. There is no price on it yet, with Early Access weeks away, which is bloody typical of the model.

Once a section has been checked against a real build, it gets a build stamp, for example "Tested on Early Access build 0.1.2, October 28". No stamp, no trust.

How Builds Work in Emberville

Strip the marketing away and Emberville's build system is a handful of ideas stacked together. I'll take them in the order you'll meet them.

Classes Aren't Picked at the Start

Official. There is no class selection in the character creator. The Steam page says you choose from a wide selection of classes as you gain experience and complete quests, and that you can change them at any time.

The press release is more specific about what unlocks them. Levelling up weapon and armour categories unlocks new classes. Expeditions into the Vitromotus, the underground prison the game is set in, also contribute towards unlocking classes, crafting options and NPCs.

The fan databases add that some classes are taught by rescued townspeople. That's plausible and it fits the rescue loop, but I haven't seen Cygnus Cross say it in so many words.

Weapons and Armour Level on Their Own

Preview. Mobalytics describes a Masteries section for each weapon, with the weapon levelling as you use it. That fits the official line that weapon and armour categories level up and unlock classes.

The practical point is that your class is a consequence of how you played, not a decision you made upfront. You build a class by swinging a sword for a few hours.

Swap Whenever You Like

Official. Classes can be changed at any time. There's no mention of a cost, a cooldown or a respec token anywhere in the official material, which is why a test for it sits on the day-one list further down.

Compare that with our Windrose talent tree guide, where the level 10 commitment point decides your build for good. Emberville is pitching the opposite. Nothing locks you in.

Save Anytime Makes Experiments Cheap

Official. Steam lists Save Anytime and Steam Cloud among the features. Put that next to free class swapping and you get a game that's friendly to experiments. Back up before you test a strange build, and the worst case is a few minutes lost.

Inherited Skills Are the Real Hook

Official. Combat skills are inherited from one levelled class to another, so you can customise your active class. Cygnus Cross lists it first under its exclusive features.

If you've played Final Fantasy V or Final Fantasy Tactics, you know the shape. Master a job, keep its abilities, carry them into the next job. Systems like that tend to produce two kinds of player. One plays a single class properly. The other spends twenty hours assembling a monster from borrowed parts.

Emberville is clearly written for the second player, and I like that. It also means the build question here is less "which class" and more "which pieces do I carry".

Two Buttons Plus Combos

Official. The base is a two-button combat system. On top of that sit weapon-bound combos, so each weapon gets its own chains, and combat skills tied to your active class.

The Steam page says combat rewards positioning, timing and knowledge. That's a polite way of saying button mashing only gets you so far. Fair enough.

Six Attributes, and a Resource Called Kymia

Official, from a screenshot. The equipment screen on the Steam page shows a level 1 character with six attributes: Strength, Dexterity, Intelligence, Wisdom, Stamina and Luck. Each has plus and minus buttons beside it.

The stats list underneath includes block amount, parry chance and counter attack chance, plus Kymia points and Kymia recovery. Kymia is the setting's life-force energy, and on that screen it looks like a resource pool. I'd guess skills spend it, but that's Inference.

What each attribute does for which weapon is not public. Whether those minus buttons mean you can refund attribute points freely is also unknown. Both are on the test plan.

Professions and Gear

Official, plus fan database. The Steam page promises a large variety of gathering and crafting professions that let you forge combat equipment, potions, food, tools and trinkets.

One fan wiki counts eleven named professions plus farming, and says four of them make gear directly: weaponsmithing, armour crafting, tailoring and jewelcrafting. Treat that number as a lead. A German preview also reports randomness in crafting, which I couldn't confirm.

What the Design Implies

Four things follow from the confirmed systems. All of them are Inference, so treat them as bets.

• Your first choice is cheap and your time is expensive. If classes swap freely, picking wrong costs you hours of levelling, not a restart.
• Breadth beats depth. Inheritance rewards anyone who levels several classes, so the sensible plan is to level more than one.
• Defensive passives are the obvious cargo. The safest thing to borrow is whatever keeps you alive while you deal damage some other way.
• Whichever class levels fastest becomes the passive farm. Every system like this develops one. Expect the community to find it within a week.

The Weapons We Can Actually Talk About

Four weapon types show up consistently in previews and demo footage: swords, bows, knuckles and magic. That is the whole confirmed list. Cygnus Cross has said nothing official about any others.

A fan wiki says a June 2025 devlog targeted sixteen weapon categories and thirteen classes, and hedged both in the same paragraph. I'd read sixteen as ambition rather than an Early Access promise. The official Early Access list is gameplay from the start to the early and middle game, with melee, magic and ranged combat all present.

What the Early Previews Complain About

Two complaints keep repeating. Mobalytics found gathering fiddly, with items picked up one at a time and mining described as tedious. And early loot is hard to read, because a weak weapon and a good one can look near-identical until you open the tooltip, which is a problem our own demo coverage raised.

Both are exactly what Early Access exists to fix. The official position is that better balance and expanded difficulty options are goals for version 1.0, not Early Access. So the numbers you see in the first month are a draft.

Expect patches to rewrite the weapon ranking more than once. Any guide that gives you permanent weapon tiers in week one is guessing.

Why the Bow Keeps Coming Up

Preview and footage. Mobalytics calls the bow versatile and great for kiting while staying safe from enemies and early bosses. The demo footage I worked from points the same way. A ranged option that lets you stand back from an early fight is hard to argue with.

That isn't the same as the bow being the best weapon. It's the best-looking one for the first few hours, and those few hours are the only window anyone has seen.

The Classes: Two Named, a Roster Still Hidden

Cygnus Cross has shown two classes so far, the Knight and the Warlock. Everything else is a gap.

The official character sheet backs the Knight's identity in one respect. It lists block amount, parry chance and counter attack chance as stats, and those have to belong to someone.

Don't Confuse the Acolytes With Classes

Our demo review mentions the Acolytes as a faction. One fan wiki points out that the studio marketed four names under that heading and never described any of them as a class. I'd keep them separate until Cygnus Cross says otherwise.

How to Read Class Guides Before Launch

If a site shows you a class tier list, stat numbers or a best class today, it's inventing them. The only class information that exists is two names, one archetype each, and a handful of skill names.

At launch I'll judge classes on five things:

• Hours to unlock.
• How well they lend skills to other classes.
• How survivable they are alone.
• How clearly they play.
• How hard they are to learn.

Single-player changes the maths. Nobody needs a tank for their party, so survivability matters to you, not to a group.

What Early Access Includes, and What It Doesn't

The Early Access FAQ on the Steam page is the most useful document anyone has published on this game, because it says what's in and what's coming later. Here's the split.

In Early Access, officially: gameplay from the start through the early and middle game, single-player support, base building, gathering, crafting, farming and cooking, equipment and skills systems, melee, magic and ranged combat, the combat class system, and support for keyboard, mouse and controller.

Planned for 1.0 instead: modding support through the Steam Workshop, improved optimisation, expanded difficulty options, better balance, a more advanced creative area, formidable bosses, diverse biomes and a wider variety of monsters.

Read that list as a build planner.

• Balance is a 1.0 job. Every weapon and class ranking you see this month is a draft.
• Formidable bosses are a 1.0 feature. Early Access has bosses, because previews mention one, but expect the serious tuning later. That limits how hard a build can be tested.
• The content stops at the early and middle game. Nobody will be able to say what an endgame build looks like for a long time.
• Mods arrive with 1.0. There's no Workshop fix coming for a bad weapon in week one.
• Difficulty options expand later. The Steam page already lists adjustable difficulty, so what's coming is more of it, not the first of it.

The studio estimates about 18 months for the whole journey. That's a long time for a build guide to stay provisional, which is the strongest argument for a page built to be updated.

Three Ways Skill Inheritance Could Work

Cygnus Cross has said skills are inherited between levelled classes. It hasn't said how. The mechanism decides almost everything about builds, so here are the three shapes it could take. You'll know which one you're in within an hour of launch.

Model One: Free-Form

Any inherited skill can go in any slot, with generous or unlimited slots. Class identity fades fast, because classes turn into shops for skills. The meta becomes whatever the best set of borrowed passives turns out to be, and everyone converges on it.

The signal: you can fill every slot with skills from other classes and nothing complains.

Model Two: Capped Slots

A limited number of inherited slots, perhaps a few passives and a few actives. This is where real hybrid builds live, because every borrowed skill costs you something. The question becomes which four to six skills matter.

The signal: a slot counter on the skills screen, and tooltips that mention it.

Model Three: Gated by Gear

Skills only work with compatible weapons or armour. Shield skills need a shield. Spell skills need a staff. Classes keep their identity, and borrowing is mostly limited to universal passives. Your weapon choice constrains your build.

The signal: skill tooltips that list weapon or armour requirements.

Which One I'd Bet On

Capped slots with some gear gating. The official wording is that inheritance lets you customise the active combat class, and that combos are bound to weapons. Both suggest the weapon still matters and borrowing has limits. That's Inference, and it's the first thing I'll check.

Here is what each model does to the picks earlier on this page.

Every Choice Compared

Each pair below follows the same format. A provisional pick, what we know, the case for each side, and what would change my mind. The anchors are fixed, so you can link straight to a pair.

Sword or Bow

Provisional pick: bow for your first hours. Confidence: Low.

Previews and footage agree that the bow is the safest early option. Mobalytics says it lets you kite and stay clear of early bosses. Cygnus Cross lists better balance as a 1.0 goal, so the Early Access bow is working off an unfinished ruleset.

The case for the bow is simple. It's the only weapon where a preview describes a clear safety advantage. If you're learning an unfamiliar system with weapon-bound combos and class skills, standing back is a gift.

The case for the sword is that it's the reference weapon. Every preview opens with it, which usually means it's the one the studio tuned first. If the Knight unlocks through sword and heavy armour levelling, as the fan databases suggest, then swinging a sword also levels the class you probably want to borrow from.

What would change my mind: a bow nerf, ammo costs I haven't heard about, or early bosses designed to punish range. Also, if the bow's mastery levels slowly, the sword wins on time.

Melee vs Magic vs Ranged

Provisional pick: don't choose. Level all three. Confidence: Medium.

Here the official design does the arguing for me. Weapon categories level independently, classes unlock from levelling them, and classes swap freely. All three combat types are in Early Access. Nothing in that setup punishes you for spreading out.

The case for specialising is depth. Focus gets you combos and masteries sooner, and if mastery caps bite early, concentrating wins.

The case for spreading is inheritance. The whole system rewards anyone who has several levelled classes to borrow from, and the cost of picking up a second weapon is time rather than a decision.

If you have to sequence them, start with whatever keeps you alive while you level the others. In the previews that points to ranged first, melee second for the armour and the Knight, and magic third. That order is Inference.

What would change my mind: a shared XP pool that splits your progress, exclusive mastery caps, or weapons gating content. Also a late unlock for magic, which would turn it from a choice into a reward.

Knight or Warlock

Provisional pick: Knight first. Confidence: Low.

We know very little about either, so this is mostly a bet on the inheritance system. The Knight is a defensive melee class, and the fan databases say its skills in screenshots include Raise Shield and Counter Attack. The official stats screen carries block, parry and counter attack chance. The Warlock is spell-driven, and nothing else is public.

The case for the Knight is that defensive passives are the best cargo in any borrowing system. In a single-player game, staying alive is the job, and a class built around it is a safe thing to level first and a useful thing to inherit from later.

The case for the Warlock is the usual one for casters in action RPGs. The damage ceiling tends to be higher, and Kymia as a visible resource suggests spellcasting leans on it. That's all I've got, and it isn't much.

What would change my mind: Knight passives that need a shield equipped to work when inherited, a Warlock that unlocks much earlier, or tight skill slots that make borrowing impractical.

Knuckles or Sword

Provisional pick: sword, but barely. Confidence: Very low.

We have exactly one sentence about knuckles. Previews call them a melee style. A verdict built on that would be fiction, so I'm not going to write one.

What I'd test is attack speed, combo length, armour pairing, and which class tracks the knuckle mastery. If knuckles turn out to be the fast, aggressive option against the sword's defensive one, this becomes a proper playstyle choice.

Right now it's a coin flip with a sword-shaped thumb on the scale. I'll replace this section the day I can swing both.

Which Profession First

Provisional pick: whichever feeds your main weapon. Confidence: Low.

The official line is that professions let you craft combat equipment, potions, food, tools and trinkets, and that crafted items improve through levelling, skills, blueprints and recipes. A fan wiki says rare materials gate the best gear. A German preview reports randomness in crafting.

The case for gathering first is that materials gate everything. You can't craft what you can't collect, and a character with good materials and a mediocre recipe still has options.

The case for crafting first is that making your own upgrade may beat waiting on drops, especially if drops are random and tooltips are hard to read. That argument only holds if crafting quality scales the way I'd expect, which is an if.

What would change my mind: heavy randomness in crafting, profession level caps, shared XP between professions, or recipes locked behind combat you can't win yet.

Farm or Fight

Provisional pick: fight first, farm as support. Confidence: Medium-low.

This is the question the game was built to ask, and the marketing doesn't help. The studio has described it as "Diablo lite meets Stardew Valley". Previews say you're fighting from the first minute.

The Steam page says you progress through combat, exploration, rebuilding and the choices you make, with farming, cooking and base building all in Early Access. The director told a Spanish outlet the game is meant to get darker as you go and isn't trying to be Stardew. Mobalytics called the early game combat-heavy and the mining basic.

The official screenshots show a day counter and a clock on the HUD. Time matters somehow. How it matters is the unknown that decides this pair.

The case for farming is that food and potions are usually the cheapest power in games like this. A farm is steady income, and the estate is described as a place that improves skills.

The case for fighting is that levelling weapons unlocks classes, and the classes are the build. Everything else supports that.

What would change my mind: time pressure that forces trade-offs, food buffs that bosses effectively demand, or farm output that scales with in-game days. Any of those turns farming from a side dish into a main course.

Four Builds We Want to Test on Day One

These are hypotheses, not recommendations. Each one rests on the confirmed systems plus a bet. Each has a kill condition, so we know when to bin it.

The Wall

The hypothesis is a Knight carrying the toughest passives you can inherit, using a high-damage weapon for offence. For it to work, defensive skills have to carry across classes, and block, parry and counter stats need to scale hard enough to matter.

The kill condition is shield skills that only work with a shield equipped. That would make the borrowing worthless.

The Kiter

The hypothesis is a bow plus defensive passives, staying away from everything. It needs the bow to keep its early strength after the first balance patch, and no punishing ammo cost.

The kill condition is bosses that close distance fast, or arenas too small to kite in.

The Borrowed Caster

The hypothesis is Warlock spells with Knight survivability grafted on. It needs strong spells, Kymia recovery that keeps up, and melee passives that still work with a staff.

The kill condition is Kymia starvation, or skill slots too tight to fit both.

The Homesteader

The hypothesis is investing early in gathering and crafting so your gear tier stays ahead of the enemy tier. It needs gear quality to scale with crafting level, and materials that aren't locked behind fights you can't win yet.

The kill condition is crafting randomness that makes drops better than anything you can make, or a time system that turns farming into a trap.

The Day-One Test Plan

When the build lands, I'll check these in this order and write down build numbers. Anything I can't verify stays as it is, with its Low confidence tag.

• How many classes exist at launch, and what each one needs to unlock.
• Whether any classes unlock through rescued townspeople, and which ones.
• Whether swapping classes has a cost, a cooldown or a restriction.
• How many inherited skill slots there are, and whether any skills can't be inherited.
• What each attribute does for each weapon type, and whether attribute points can be refunded.
• How fast each weapon category levels, and whether mastery has an early cap.
• Whether the bow is still the strongest option after the launch patch, and whether it uses ammo.
• What knuckles and magic actually do compared with the sword and bow.
• How crafting works, including whether randomness affects quality.
• What the day and clock system does to farming, enemies and shops.
• Which early bosses exist, and whether any of them force a weapon type.
• How it plays on a controller and on Steam Deck, since Valve hasn't rated it.

The Steam listing for the free Emberville Demo gives an October 19 release date. I'll rerun the relevant checks on that build first, because it lands eight days before Early Access and it's the first public Steam build. This page gets updated the same day.

Questions People Are Already Asking

When does Emberville come out?

October 27, 2026, in Steam Early Access, with a GOG release the same day according to the studio's press release.

How much does it cost?

Steam shows no price as of October 4. Cygnus Cross says it's a one-time payment with no microtransactions, that the price stays the same after Early Access, and that it may discount during Early Access. Until a number appears, anything quoting a price is guessing. I'll update this page the same day one is announced.

How long will Early Access last?

Steam's Early Access FAQ estimates about 18 months, and the studio adds that community feedback could stretch it. Earlier coverage, including several previews, said roughly a year. The Steam page is the current word.

Is Emberville single-player or co-op?

Steam lists it as single-player, and the Early Access content list says single player support. No co-op has been announced.

Can you change class?

Yes, at any time, according to the Steam page. What it costs, if anything, isn't stated. You don't pick one in the character creator at all.

Is there a demo?

Steam lists a free Emberville Demo ($0) with an October 19 release date. Earlier previews and footage, including what our Prologue Demo review draws on, come from pre-release builds.

Does it work on Steam Deck?

Valve hadn't rated it as of October 4, and the status reads unknown. The store page lists requirements for SteamOS and Linux and full controller support, so it's a reasonable bet, but I won't call it confirmed.

What are the system requirements?

The minimum is Windows 10, an Intel Core i5 at 3.2 GHz or equivalent, 4 GB of RAM, a DirectX 11 graphics card and 4 GB of storage. That's low. Recommended specs aren't listed yet.

Who voices the characters?

The July 16 press release names Doug Cockle, Amelia Tyler, Alex Jordan and Abubakar Salim, then adds Matthew Mercer, Jennifer English, Ben Starr, Nick Apostolides, Elias Toufexis, Anjali Bhimani and Briana White. Jane Perry narrates. LilyPichu and the DJ Dillon Francis appear too, and the release says there are more. Which actor plays which character is mostly unannounced.

Is it hard?

Steam lists adjustable difficulty. The demo footage behind our review found normal settings forgiving. Expanded difficulty options are a 1.0 goal, so Early Access may feel gentler than the final game.

Which languages does it support?

English has full audio. Interface and subtitles cover Spanish (Spain), French, German, Japanese, Korean, Simplified Chinese, Traditional Chinese and Brazilian Portuguese.

Does it have Steam achievements?

Steam doesn't list any yet. That could change before October 27, and I'll update this page if it does.

Is there a Linux version?

The store page has a SteamOS and Linux requirements tab. I don't see a Mac listing.

Is it Stardew Valley with combat?

Closer to Diablo with a town attached, going by the previews. Farming, cooking and base building are all in Early Access, so the town matters. Treat it as an action RPG first.

Glossary

The game has its own vocabulary, and a few terms get misused already. Here's how I'm using them.

• Vitromotus: the vast underground prison, built as a magical dungeon, where the game starts and where expeditions go.
• Kymia: the setting's life-force energy, said to bring those trapped in the Vitromotus back to life. It also appears as a stat and resource on the character sheet.
• Emberville: the ruined town above the prison that you rebuild by rescuing its people.
• Estate: your personal area, where you build, craft, farm and decorate, and which the studio says can improve skills.
• Masteries: the per-weapon progression sections described in Mobalytics' preview.
• Weapon-bound combos: combo chains that belong to a specific weapon, sitting on top of the two-button base.
• Inherited skills: combat skills carried from one levelled class into another so you can customise your active class.
• The Acolytes: a name the studio has used in marketing. It has not been described as a class.
• Early Access: Steam's work-in-progress release model. Emberville's runs from October 27, with an estimated 18 months to 1.0.

Sources

• Emberville on Steam, including the Early Access FAQ, features list and system requirements.
• Cygnus Cross press release, July 16, 2026, covering the feature list and full voice cast.
• Mobalytics' early-build previews: weapons and skills and first impressions.
• PCGamesN on the Early Access date and cast, and 3DJuegos on the director's comments.
• Fan databases, used as leads only: meowdb and thegameswiki.
• Our own Emberville Prologue Demo review.

What Happens to This Page After October 27

This page is the hub for everything we publish on Emberville. Between now and launch I'll add standalone pages for the release date and price, classes, weapons, the voice cast and how it compares with Stardew Valley. At launch the class roster, unlock guide, beginner guide and professions follow within hours, and this page absorbs the tested answers.

Every section that gets checked against a real build is stamped with the build number and the date. A section without a stamp is still a bet.

Here's the bet. Early Access guides go stale within a week, so the useful ones are the ones that say what they don't know and fix it fastest. This page will be wrong about something by October 28. I'd rather you read it knowing which parts.