Twenty-Four Weapons
What Shroud of the Avatar's weapons actually differ by, what the game never tells you about them, and thirteen proposals for the ones a smith makes.
There are twenty-four craftable weapon archetypes in the game, spanning daggers to halberds to crossbows to wands. They differ in base damage, critical chance, critical multiplier, range, reach, weight and a strength-scaling divider. They do not differ in attack speed, because weapons have no such stat. And they do not differ mechanically by name: a Dagger and a Longsword are the same weapon type to every branch of combat code, separated only by the numbers on their archetype. This page is about the weapons themselves and the gear identity around them.
Three things are true about weapons here at once, and they pull in different directions.
The first is that the weapon you choose matters far more than anything you can do to it afterwards. Three masterwork slots spent entirely on damage move a crafted longsword from 11 base damage to 11.76, a gain of 6.9%, for about 6,900 gold and three permanent scars. Crafting a halberd instead moves it to 20, a gain of 82%, for the price of a different recipe. That is not a criticism of the enhancement chain so much as a statement of where the leverage actually is, and it is the reason this page exists separately from that one.
The second is that a third of the weapon classes in the game are locked out of most of the content. A Grand Master enchanting a bow, a crossbow or a shield is offered three options drawn from eleven, and all eleven are attributes, whichever of the nine gems is consumed. Every other weapon type draws from all nine elemental pools. Both of the archery-themed enchants in the game are assigned to swords, maces, spears and wands, and to nothing that fires an arrow. Wands and staves get seven masterwork options; one-handed bludgeons get eighteen.
The third is that the identity layer already exists in the build and has no door into it. Forty-eight unique weapon components ship with their bonuses fully authored: a Keen dagger blade granting extra damage and critical chance, an Executioner axe blade, Assassin, Bandit, Savage, Dragon and the rest. Not one has a recipe. Meanwhile the twenty-two weapon categories players read as identity branch nothing in combat, and eighty artifact weapons sit on a rarity ladder with its own upgrade economy that no crafted weapon can enter.
Thirteen proposals follow. Seven are pure data, three are client-only, one needs server work, and two are deferred. The tooltip fixes are filed as defects independently so they can be fixed without waiting on a vote.
| # | Proposal | Tier | Size | Server | One-line pitch |
|---|---|---|---|---|---|
| W1 | Exceptional worth wanting | 1 | M | none | a critical craft grants durability and nothing else |
| W2 | Gems that matter on bows and shields | 1 | M | none | author the nine missing element pools |
| W3 | Archery enchants on archery weapons | 1 | S | none | Ice Arrow and Obsidian Arrow are on swords, not bows |
| W4 | Bows and crossbows apart | 1 | M | none | one enhancement bucket covers both; split it |
| W5 | Masterwork pool parity | 1 | M | none | wands get 7 options, bludgeons get 18 |
| W6 | The unique component tier | 1 | L | none | 48 authored components, zero recipes |
| W7 | Weapons that wear differently | 1 | S | none | a live per-weapon wear dial, set to no-op on all 24 |
| W8 | Honest damage numbers | 2 | S | none | the tooltip prints a band twice as wide as the real roll |
| W9 | Throughput readout | 2 | M | none | charge time and damage per full swing, never shown |
| W10 | Comparison deltas | 2 | M | none | the compare panel exists and does no arithmetic |
| W11 | Weapon provenance | 3 | M | read call | make the maker's mark load-bearing, as UO did for 25 years |
| W12 | Weapon affinity | 4 | XL | per-item progress | the Episode 2 promise: make a specific weapon matter |
| W13 | The crafted-to-artifact bridge | 4 | XL | economy review | join the two ladders instead of capping one below the other |
A weapon carries base damage, a critical chance, a critical damage multiplier, a range, a strength-scaling divider, a parry contribution and a free-attack charge length. It does not carry an attack speed. Pacing comes from the ability slotted in the deck plus that charge length, which runs from five ticks on a longsword to eight on a long bow and is never displayed anywhere.
There are three separate weapon-type enumerations doing three different jobs. The functional one, which every combat branch reads, has thirteen values: unarmed, one- and two-handed blade, one- and two-handed bludgeon, shield, one- and two-handed polearm, bow, crossbow, wand and staff. The enhancement one, which decides which pools a weapon draws from, has fourteen and merges bows and crossbows into a single "ranged". And the one players actually see, with twenty-two values from Dagger to Trident, is read in exactly two places, both of which render a label.
Measured against the shipped data. The cost side of the enhancement chain, the durability budget and the failure economics are audited on the enhancement page; what follows is what is specific to weapons.
| # | Finding |
|---|---|
| P1 | An exceptional weapon gets durability, and nothing else. The critical-craft pool holds exactly one entry per weapon class. For a one-hander it is a 2.5% reduction in wear rate; for a two-hander, 5%. No damage, no crit, no speed. The single most celebrated moment in crafting produces a sword that lasts marginally longer. |
| P2 | Four exceptional effects ship and are granted to nothing. A flat +50 durability, a +100 carry weight, a generic weapon bonus and a Grand Master variant all exist as finished assets sitting in pools that no code path reads. The Grand Master one is unreachable by construction: the flag that would select it is passed into the roll and then ignored, under a comment saying it would be implemented later. |
| P3 | A Grand Master enchanting a bow, a crossbow or a shield picks three from eleven, and all eleven are attributes. Ranged and shield weapons carry only the gemless universal group. Every other weapon type carries all nine elemental groups. Burning a Fire gem, a Water gem or a Chaos gem on a longbow produces an identical menu: strength, dexterity, intelligence, health, focus, and a durability bonus. |
| P4 | The two archery enchants are unreachable on archery weapons. Ice Arrow Power and Obsidian Arrow Strength Power are assigned to one-handed blades, one-handed bludgeons, one-handed polearms and wands. The only way to enchant your bow's Ice Arrow is to enchant a sword. |
| P5 | Masterwork pool depth runs 2.6 to 1 across weapon types. One-handed and two-handed bludgeons carry 18 options each, polearms 16, ranged 14, blades 13, shields 11, and wands and staves 7. There is no shared pool smoothing this out: the arrays that would hold universal weapon effects are both empty. |
| P7 | A bow and a crossbow are the same weapon to the entire enhancement chain. They share one masterwork pool, one enchant pool and one exceptional bucket, because the enhancement type enumeration has a single "ranged" value covering both. |
| # | Finding |
|---|---|
| D1 | Weapon category is cosmetic text. The twenty-two values players read as identity are consulted in two places, both of which render a label. Every combat and enhancement branch reads one of the other two enumerations. A dagger and a longsword differ only in the numbers printed on their archetype. |
| D2 | The identity layer exists in data and has no door. Forty-eight unique weapon component archetypes ship, each bound to a finished bonus: a Keen dagger blade grants +3 damage and +10% critical chance, an Executioner axe blade grants +3 damage and a strength bonus. Not one of them has a recipe. The plumbing that would carry them onto a finished weapon is live and already used by the ordinary metal tiers. |
| D3 | The tooltip advertises a damage band twice as wide as the real one. It prints half to one and a half times base damage. The roll is three quarters to one and a quarter. A crafted longsword reads "5 to 16" against a true 8.25 to 13.75. |
| D4 | The crafted-effect tooltip sorts by English substring match. Grouping and the failure tally are decided by searching a translated display name for the words "Socket", "Masterwork" and "Enchantment", and the two section headings are hard-coded English. On a non-English client the grouping collapses. The structured data needed to do this properly is already on the item. |
| D5 | There is no throughput number anywhere, and no comparison delta. The equipment comparison panel already places a hovered weapon beside up to two equipped ones, including the dual-wield case, and computes nothing. Meanwhile charge time genuinely varies by weapon and is never shown. |
| D6 | There is no set system and no affinity. An armour-set match flag exists in the data with its only consumer commented out. Item Affinity, promised for Episode 2, is absent entirely. |
| D7 | There is a live per-weapon wear dial set to no-op on every craftable weapon. A weapon-level divider on durability damage is read at runtime and is set to exactly 1 on all 24 craftable archetypes. A free differentiation axis, authored and flattened. |
| D8 | Two "is this two-handed" implementations disagree. One counts a magic staff as two-handed; the other does not. Different call sites use different ones. |
| D9 | Item score is a real currency with no display. Every effect contributes to an item rating that the game reads for real decisions. Every failure quietly subtracts 1 from a number the player never sees. |
| D10 | A complete artifact ladder ships and is disjoint from crafting. Eighty artifact weapons exist, sixteen distinct designs across common, rare, epic and legendary, with their own upgrade currency and their own repair rules. A legendary Sword of Speed is 15.5 base damage in one hand against a crafted longsword's 11, before either has been enhanced at all. No crafted weapon can ever enter that ladder. |
| D11 | A per-scene gear cap scales every item effect and is invisible. Scenes can carry a cap that multiplies what your gear is worth. No tooltip mentions it, so a player whose weapon "feels weaker here" has no way to confirm they are right. |
Duplicate effects are excluded from the offered set, so the best damage stack is one of each damage effect. At Grand Master the slot multipliers are 1.56, 1.61 and 1.66.
| Route | Base damage | Change | Cost |
|---|---|---|---|
| Crafted longsword, plain | 11.00 | — | three ingots |
| The same longsword, all three reachable masterwork slots spent on damage | 11.76 | +6.9% | about 6,900 gold and three scars |
| Craft a halberd instead | 20.00 | +82% | a different recipe |
| A legendary Sword of Speed, one-handed, before any enhancement | 15.50 | +41% | not craftable at any skill |
Three masterwork slots of damage, at the practical ceiling of the system, are worth less than a twelfth of what changing which weapon you crafted is worth. That is the crafted-versus-drop problem in one table, and it is precisely the problem Ultima Online spent 2003 to 2009 solving.
A crafted longsword at 165 maximum durability, repaired with crowns from empty each time, restores 165, then 110, 82, 66, 55, 47, 41, 36, 33, 30. Ten repairs cost 2,500 gold crowns and buy 625 points of working durability. A brand new longsword is 165 points for three ingots.
These ship as effect and recipe data plus a catalogue re-export. Magnitudes are formulas evaluated when an item loads, so a retune is retroactive: existing weapons pick up the new numbers with no data migration and no re-crafting.
The critical craft is the best moment a smith has, and its entire reward is that the weapon wears out 2.5% slower. Grow the pool from one entry to a small authored ladder: a damage grade, a critical grade, a durability grade, a maker's flourish. Wire the Grand Master flag that has been passed into the roll and ignored since launch, so Grand Master recipes draw from a distinct list. Four already-finished exceptional assets get homes in the same pass.
Seen in: UO's exceptional craft, which grants a flat damage increase a drop cannot have, plus the maker's mark; FFXIV's high-quality tier; WoW's five quality ranks.
The bonus-effect selection already picks at random from an array, so growing the array is pure data. The Grand Master branch is a two-line change on the same method against the recipe flag that is already being passed in. Keep the grades modest: on a mid-difficulty recipe a Grand Master criticals often enough that anything strong here is high-volume, and the point is a payoff worth noticing, not a second power tier.
This is deliberately the crafter's exclusive territory. The enhancement rails are shared between crafted and dropped weapons; the base quality tier is the one axis a drop cannot have, which is exactly why UO put its crafted-versus-drop answer there.
Author the nine missing elemental groups for ranged weapons and shields, so an archer's Fire gem does something an archer's Water gem does not. Roughly eighteen new groups and forty to fifty effect rows, following the melee template of a proc plus a stat per element. This is the largest missing-content hole on the page: right now a third of the weapon classes in the game treat the entire gem system as decoration.
Seen in: Diablo II's socketed jewels, which apply to bows and shields identically to swords; ESO's enchanting glyphs, weapon-agnostic by design; PoE's bow-specific mod pools.
Entirely inside the weapon enchant pool table that the enchant-effect selection reads, plus new effect assets alongside the existing enchantment set. No code. Ship stat-only variants first if proc rates need a balance review; the stat half alone closes most of the gap.
Distinct from the Frost Weapon and Shock Weapon combat skills that shipped in July. Those are temporary self-buffs cast from the deck. This is the permanent crafted layer, and the two systems do not currently share an element roster.
Ice Arrow Power and Obsidian Arrow Strength Power are assigned to one-handed blades, one-handed bludgeons, one-handed polearms and wands. Move them to ranged, and give the melee groups melee-ability equivalents from the effects that already exist. This is a data-placement fix and it rides the same export as W2 at no extra cost.
Seen in: nothing exotic. This one is just wrong.
Same asset and same accessor as W2. Existing weapons keep whatever they already carry, because effects are stored on the item by identifier rather than by pool membership, so the change is forward-only and needs a patch note rather than a migration.
One enhancement bucket covers both, so a crossbow can never enhance differently from a longbow no matter how differently they are meant to play. Add the two values, retag the three archetypes, split the pools.
Seen in: GW2, where longbow and shortbow are separate weapon types entirely; WoW's historic bow, crossbow and gun split; Monster Hunter's bowgun families.
The enhancement type enumeration is shared between client and server, so this is one line in the mirrored file plus data. Append the new values, never insert them: the enumeration is persisted by ordinal on shipped items, and inserting would silently retag every enhanced weapon in the world. Do the pool authoring once, together with W2.
Wands and staves have seven masterwork options each; one-handed bludgeons have eighteen. Bring the thin pools up, and use the two universal weapon arrays that ship empty as a floor every weapon type inherits, so the next weapon type added does not start at zero.
Seen in: WoW's secondary-stat model, where everything draws from a shared pool and only the flavour is type-locked.
Pool size changes the odds of seeing any particular effect in a three-of-N draw, so state the intended draw rates before authoring rather than after. This proposal deliberately does not change any effect's magnitude coefficient: which pool an effect lives in is itemization, what it is worth is balance, and the balance work owns the second one.
Forty-eight unique weapon components ship with their bonuses fully authored and no recipe makes any of them. Keen, Slender, Assassin, Bandit, Executioner, Savage, Sturdy, Heavy, Dragon and the rest are sitting in the build waiting for a door. Add recipes at a high difficulty band behind rarer stock, and weapon category stops being flavour text: a dagger blade can be Keen and a longsword blade cannot.
Seen in: UO's runic hammers producing named property sets; Monster Hunter's weapon trees; Diablo II's superior base items.
| Recipe | Diff | Chance |
|---|---|---|
| Keen Dagger Blade | 11 | 42% |
| Keen Shortsword Blade | 11 | 42% |
| Executioner Axe Blade | 12 | 28% |
New recipe assets pointing at the existing component archetypes. The effects propagate onto the finished weapon through the same descriptor path the copper, iron and obsidian tiers already use, so nothing new is needed to carry them.
One decision to make first. These effects are flat constants, not proficiency-scaled formulas like the rest of the chain, so a Keen blade made by a novice is identical to one made by a Grand Master. Either convert them to the standard expression or accept them deliberately as a flat tier and say so, but do not ship the inconsistency by accident.
There is a live per-weapon divider on durability damage, read every time a weapon takes wear, and it is set to exactly 1 on all twenty-four craftable archetypes. Author it. A dagger wears fast, a war hammer wears slow, an obsidian blade holds an edge. Twenty-four field edits and no code at all, on a system where durability is the scarcest resource in the game.
Seen in: UO's material-driven durability; Mount and Blade's per-weapon condition; the general principle that if you ship a dial you should turn it.
Needs the structured effect tooltip from the enhancement page or it is an invisible buff. Sequence it after that page's failure-scar card, because it interacts directly with the durability economy that proposal is already moving, and shipping both in the same window without modelling them together would make the result unreadable.
All three read data the client already holds. The shared enhancement surfaces (the effect tooltip, the pool browser, the planner) live on the enhancement page; these three are the weapon-specific half of the item surface.
The tooltip prints half to one and a half times base damage. The game rolls three quarters to one and a quarter. Print the band the game actually rolls, and print the mean alongside it. Two constants.
Seen in: every modern action RPG prints the true roll; WoW prints exact minimum and maximum damage.
Players will read a narrower band as a nerf even though nothing changes mechanically, so the patch note has to carry it: the weapon is not weaker, the label was wrong. Filed as a defect separately from this proposal so it can be fixed without waiting on the vote.
Charge time varies from five ticks on a longsword to eight on a long bow, it is affected by the attack speed effects a crafter can roll, and it has never been displayed anywhere. Add one line: time to a full charge, and the damage a full-charge swing does. It is the only number that lets a player compare two weapons honestly.
Seen in: EverQuest players inventing "ratio" literacy in 1999 because the game showed damage and delay but never the quotient; every ARPG since printing damage per second on the item.
Charge length, the tick constants and the attack speed stat are all local. The reference calculation already exists in the combat maths for the free attack itself.
Label it as a per-swing estimate, not a sustained rate. Real pacing is dominated by the cooldowns of whatever is slotted in the deck, and those belong to the combat balance work. Promising a damage-per-second figure this system cannot honestly compute would be worse than showing nothing.
The comparison panel already places a hovered weapon beside up to two equipped ones, including the dual-wield case, and does no arithmetic at all. Add the subtraction: damage, critical, charge time, and every stat the effects contribute, with an honest note about which hand is being replaced when a two-hander displaces a pair.
Seen in: WoW and Diablo's green and red stat arrows; Destiny's compare overlay; New World's gear score band.
The composition is already built; the deltas come from the same stat totals the panels already compute. Coordinate with the inventory page's shared tooltip work rather than forking a second builder, and take the throughput number from W9 so the compare and the tooltip never disagree.
The smith's name is already stamped on every non-stacking craft, and the proficiency each individual effect was rolled at is already stored with that effect. Almost nothing else in the genre stores that second thing. Surface both properly: who made it, how good they were when they made it, the engraving, the pattern. Then add the lookup that lets a buyer see a smith's body of work, which is what turns a name into a brand.
Seen in: UO's maker's mark running a named-smith economy for twenty-five years, gated behind both Grand Master skill and an exceptional roll so a signature certified two facts at once; Star Wars Galaxies, where the crafter named the item and the name stayed on it permanently, and reputations were built on exactly that.
The display half is client-only over data already synchronised, and could ship on its own inside the enhancement page's effect tooltip. Only the lookup needs server work: a read call returning a crafter's public output, or a small registry if it should be queryable at scale. Name the registry in the migration ledger if that route is taken.
Coordinate with the crafting page's proposal to make some low-value commodity crafts signature-free, so weapons are never in that set. A signature that appears on iron nails is not a brand.
Two big swings, stated with their dependencies rather than costed, because neither should start before the tiers above and the enhancement chain work have landed.
The Episode 2 promise, still absent, and the only mechanic here that would make a specific weapon matter rather than a weapon archetype. A weapon accrues affinity with the character carrying it through use, unlocking a small track that differs by weapon category, stored on the item. It is the natural home for the twenty-two categories that currently branch nothing, and it is the only proposal on this page that gives a crafted weapon a reason to be repaired rather than replaced.
Seen in: Dark Souls upgrade paths and infusions; Final Fantasy XI's weapon skill points; Warframe's per-weapon mastery; Monster Hunter's weapon trees.
Needs a new per-item counter and a new effect family, with a migration-ledger entry for the added field. The open economy question is whether affinity travels with the weapon when it is sold, and the answer changes what the feature is: bound affinity makes weapons personal, tradeable affinity makes them a commodity with a grind attached.
Depends on W6 (a real category identity to hang it from), the enhancement page's re-tempering card (a weapon has to be survivable before bonding with one is a good idea), and the combat balance work landing first.
Eighty artifact weapons ship with their own rarity ladder, their own upgrade currency and their own repair rules, and no crafted weapon can enter that ladder at any skill level. W4 prices the gap: a legendary one-hander beats a fully masterworked crafted longsword before either is enhanced. The sketch is a top-tier crafting path that consumes artifact upgrade currency and a fully enhanced crafted weapon to produce a crafted artifact, joining the ladders instead of leaving one permanently capped below the other.
Seen in: UO's runic-crafted items competing with artifacts, which was the eventual answer to Age of Shadows; WoW's crafted best-in-slot in Dragonflight; FFXIV's relic weapons as a crafting and content hybrid.
New archetypes, new recipes, and a decision on repair policy, since artifacts today are crown-repairable only. The economy review is the real cost: this changes what the top of the crafting ladder is worth, and it should not be attempted while the top three slots of that ladder are unreachable.
Depends on the whole of Tier 1. Bridging to artifacts while the crafted chain is capped at three slots by a durability floor would widen the gap it is meant to close.
The anti-pattern is thirteen features becoming thirteen widgets. Everything client-side above composes into the weapon tooltip, which should answer "what is this weapon and what does it do per swing" before the player has to open anything.
| Zone | Answers | Built from |
|---|---|---|
| Header chips | What kind of weapon is this, and what is it made of | W6 unique components, and the category finally meaning something |
| Stat block | What does it do per swing | W8 honest band, W9 throughput, W7 wear identity |
| Compare panel | Is it better than what I am holding | W10 deltas, honest about which hand is replaced |
| Footer | Who made it, and where do I go to change it | W11 provenance; the button opens the shared enhancement bench |
| Phase | Items | Why here |
|---|---|---|
| 1 · one data export, one client patch | W2, W3, W8 | The two biggest content holes close in a single export, and the damage band is two constants and the oldest credibility gap on the item |
| 2 · finish the item surface | W9, W10, W1, W5 | Throughput and compare complete the tooltip; the exceptional and masterwork pools get worth drawing from |
| 3 · identity | W4, W7, W6 | W7 needs the enhancement page's effect tooltip shipped to be visible at all; W6 is the largest content item on the page |
| 4 · provenance and beyond | W11, then W12 and W13 | After the balance work lands and the crafted ladder is actually climbable |
Comments and votes below are per proposal. If you only read one card, read W2: for a third of the weapon classes in this game, the entire nine-gem enchant system currently does nothing.
No account needed. Every proposal above also has its own Discussion box for talking about that one alone; this thread is for the plan as a whole. Comments are plain text, held to basic decency, and may be trimmed.