The Blade Everyone Carries

Blades is the biggest school this series has ever measured. In thirty days of live play its six runes were cast 460,427 times, 5.64 percent of every skill cast in the game, by 162 different players; Thrust alone, at 145,624 casts, is the most-used rune of any sweep to date. It is the third most specialized school, the anchor of the classic fighter pairings, and the deck every new player swings first. It is also a school where a shipped stat named Riposte Defense Power and Duration has no reader anywhere in the code, where Double Slash carries a third hit that has never once fired, where the specialization has the fewest formula terms of any weapon school, and where the signature counter can only be armed by a rune that lives in a different tree. This is not a rescue page. It is a whetstone. The school everyone carries deserves an edge.

Executive summary

Every game with a sword has walked this road. Ultima Online shipped Swordsmanship in 1997 as a flat to-hit skill and spent six years watching players park it while the interesting decisions lived in Tactics and Anatomy; in 2003 the Age of Shadows revamp bolted thirteen named special moves onto that flat skill and they became the defining combat feature of the next two decades. Dark Age of Camelot proved a parry could be a key instead of a shrug: its reactionary style chains turned every defensive event into an unlocked attack, the deepest melee system any MMO has shipped. World of Warcraft spent years learning that an execute button which is dead for most of the fight becomes a slot tax, and fixed it with a proc that lights the button mid-fight. Sekiro welded defense and offense into one loop: every deflect fills the meter that opens the deathblow. The genre's verdict on sword schools is consistent and old: the blade earns its keep when defense feeds offense, when the weapon in your hand matters, and when the finisher is a moment you can see coming.

Measured against that bar, our Blades is the strongest weapon school in the game carrying a decade of unfinished sentences. The census ranks it third of twenty-one adventuring specializations, 857 players bought in and 425 finished; players comparing weapon trees on the forums crowned it outright and told the developers to buff the other trees up to it rather than cut it down. The telemetry above says the same thing. So the gaps here are not weakness. They cluster in three places:

1. The paper school. The reference doc teaches last year's Riposte and mislabels every cooldown in the tree. Coup de Grace's legacy tooltip string describes a different skill. A shipped defensive stat has a display name and no reader; the buff it names existed once, Release 23 changed Riposte to grant avoidance, and a later cleanup deprecated it away and left the stat orphaned. Double Slash's third hit multiplies a stat that nothing in the game grants, so it has fired exactly never. Three of the school's five combos have no cooldown at all in an unlocked deck, and Quad Slash's second hit lands before its first. None of this had a single test pinning it.
2. The unanswered verbs. Riposte, the school's signature, can only be armed by Parry, a Tactics rune; its numeric identity even sits in the Blades block of the skill contract, a fossil from when it was one school. When the counter lands, nothing follows it. The execute threshold is invisible on screen, moves silently from 20 to 32.5 percent with specialization, and players have logged 600-then-9 damage swings and zero-damage hits with no message saying why. Rend, the tree's crown jewel, does nothing at all against the undead the game most often asks you to farm, and says nothing about it. And the school reads only your right hand: an offhand dagger is invisible to every Blades passive.
3. The invisible craft. The specialization touches all six runes plus two combos, nine riders, more of the kit than any other weapon school, and not one of those riders is printed anywhere a player can read; the community verdict on the forums is that blade spec is not worth buying. Meanwhile daggers, one-handers, and two-handers are one undifferentiated family to every formula in the tree, and the five combos Blades feeds, the richest combo school in the game, are hinted at by nothing in any UI. Hamstring was cast 93 times in a month, by four people.

What this page refuses to do is inflate numbers. Blades is the biggest and best-loved weapon school; players who compared the trees said so and asked that the weak ones be raised instead. The thirty-four cards below spend their budget on honesty, dead-content resurrection, legibility, and loop-closure: the book corrected, the promise kept, the wound readable, the counter answered, the family distinguished, the forms revealed. The one raw damage conversation on this page, the hole left when the Torpor and Chaotic Feedback pairing was reined in, goes to the owner's desk as a native review with evidence attached, and it is explicitly not a request to revert that change.

The thirty-four proposals at a glance

The school we have

Eleven skills: one root, six active runes, three stat passives, and the specialization. Two limbs, the piercing line through Thrust and Double Slash and the answering line through Riposte and Rend, converge at Whirling Blades and end at Coup de Grace. Every row of the tree is implemented and none of it carries a single TODO in the code, which makes Blades that rarest of things, a finished school. Finished is not the same as polished. The numbered flags below each point at a proposal.

One school, eleven skills

              ┌──────────────────┐
              │ COUP DE GRACE     │◄─[5] feeds five combos,
              │ the finisher      │      finishes none
              └─────────┬────────┘
                        │
              ┌─────────┴────────┐
              │ WHIRLING BLADES   │  five metre blast,
              │ the circle        │  six targets
              └───┬───────────┬──┘
          ┌──────┘              └──────┐
 ┌────────┴────────┐    ┌─────────┴───────┐
 │ DOUBLE SLASH    │    │ REND             │◄─[2] does nothing
 │ two hits        │    │ the bleed        │   vs undead, in
 └────────┬────────┘    └─────────┬───────┘   silence
    [1]◄ third hit          │
        never fires   ┌─────────┴───────┐
 ┌────────┐         │ RIPOSTE          │◄─[3] armed only by
 │ THRUST │         │ the counter      │   a Tactics rune;
 │ opener │         └─────────────────┘   defense stat has
 └────┬───┘                                 no reader
      └───────┬──────────────────────────┐
              ┌──────────────────┴┐
              │ BLADED COMBAT (root)│◄─[7] no card levels,
              └────────────────────┘     no innate reward

 PASSIVES  [BLADE SPEED] [VITAL POINTS] [BLADE MASTERY]
           ◄─[4] one check for daggers, 1H and 2H,
                and it reads the right hand only
 SPECIALIZATION  [BLADE SPECIALIZATION]
           ◄─[6] nine hidden riders, zero tooltips

• 1The hit that never happened. Double Slash's third strike rolls 100×DoubleSlashBoost, a stat nothing in the game grants. Written zero, read forever. S5 makes it real.
• 2The silent whiff. Rend's every layer multiplies by a check that zeroes it against undead, the game's most farmed enemies, with no message and no fallback. S20 says it out loud, S27 gives the glyph something to do.
• 3The borrowed trigger and the empty promise. Riposte's window opens only when Parry, a Tactics rune, lands; its skill number even sits inside the Blades block of the data contract, a fossil of the school it left. And the stat named Riposte Defense Power and Duration is read by nothing. S21 arms it at home, S6 keeps the promise.
• 4One family, right hand only. No formula anywhere distinguishes a dagger from a greatsword, and the equipment check that gates all three passives inspects one hand. S23 and S14.
• 5The one-way combo school. Blades feeds five combos, more than any school, and no combo anywhere produces a Blades glyph. S25 answers back.
• 6The invisible specialization. Nine riders across six runes and two combos, none printed anywhere. The forum verdict is that blade spec is not worth it. S2, S15, S16.
• 7The undernourished root. Bladed Combat has no card levels and no innate reward; Bludgeon, Polearm, and Unarmed roots all have both. S11.

Who actually plays it

Skill casts are logged by the live game, one line per glyph. Over the thirty days ending August 14 the game recorded 8,157,099 casts from 1,236 distinct players. The Blades share:

Bladed Combat itself is the unlogged base swing; basic attacks do not emit skill events. Caster counts are exact.

Share of all skill casts, last thirty days. Blades is the largest school this series has measured.

Three stories live in this table. First, Thrust is the single most-cast rune any of these sweeps has recorded, which is exactly why its poison bug mattered (S4). Second, Riposte is the school's second widest rune, 101 casters, ahead of the execute and the whirlwind, at barely a fifth of Thrust's volume: a reaction many carry and few can fire often, which is the case for arming it in-school (S21) and paying it properly (S6). Third, Quad Slash is the only healthy combo; the cross-school forms are nearly extinct, Hamstring down to four players, which is a discovery failure rather than a power failure (S24). Concentration is real: five of the school's 162 casters produced 61.5 percent of its volume, 283,149 of 460,427 casts. The school reaches 162 of 1,236 active casters, 13.1 percent, consistent with the third most specialized school in the game and its seat in the default starter deck.

How a swing happens

Combat is a synchronized simulation: the server relays which glyph fired and every client plays the result forward from authored formulas. A Blades swing passes five stations, and this page hangs a card on almost every one:

flowchart LR
  GS["Skill type bonus\nBladeAttackBonus read"] --> RH["One hit roll\nper swing"]
  RH --> CR["Crit roll\nDex / 2000 for melee"]
  CR --> MD["Damage\nweapon, skill curve, resists"]
  MD --> RS["Reaction timestamps\nyour parry stamped"]
  GS -.- P1["Counted here and again inside damage. S12"]
  CR -.- P2["Ranged divides Dex by 800 instead. S30"]
  MD -.- P3["Type set on the first hit only. Coup and Riposte carry none. S22"]
  MD -.- P4["Double Slash hit three multiplies a stat nothing grants. S5"]
  RS -.- P5["The stamp arms Riposte. The promised defense buff has no reader. S6"]
  RS -.- P6["Only a Tactics parry arms the school's counter. S21"]
  style P1 stroke-dasharray: 5 5
  style P2 stroke-dasharray: 5 5
  style P3 stroke-dasharray: 5 5
  style P4 stroke-dasharray: 5 5
  style P5 stroke-dasharray: 5 5
  style P6 stroke-dasharray: 5 5

What the record shows

The forum record on this school runs nine years deep and is remarkably consistent. In 2018, Riposte was the complaint: KenoZuki opened a thread titled Riposte Broken because the tooltip promised a hit that could not be blocked, dodged, or parried while the combat log showed parries; Nog logged ripostes doing zero damage to shield users; Alley Oop, wearing the bug moderator hat, confirmed the immunity only ever applied inside the just-parried window and closed with it IS broken right now. In 2019, the specialization: Stabbath asked what blade specialization actually does and got no answer the UI could confirm; jmcclain's advice was that blade spec is not worth it and Tactics and Life serve better; Boris Mondragon held the minority line that at high level it feels mighty. The same year Jason_M logged Coup de Grace damage swinging from the six hundreds to single digits on one target, Gravidy measured 402 followed by 9, golruul recorded zero-damage coups with no avoidance message and named the deeper problem, an excessive amount of game mechanics being hidden, and ErikRulez went seventy coups without a crit. In 2016 the combo thread: Noric observed that Blades, Fire, and Chaos hold most of the game's combos, Crazy Phil and KuBaTRiZeS proposed treating them as unlockable hidden glyphs, Time Lord never used them because the UI made it a chore. And in July and August of this year, the freshest wounds: Rinaldi's broken-skills list included Thrust, which would not apply weapon poison when every other blade skill would, and Syee Kco reported that with the Torpor and Chaotic Feedback pairing reined in, melee endgame damage fell to a level that is not good at end game, while necronut argued the pairing had been broken and its end overdue. Both positions are on this page, in S32, on the owner's desk where they belong.

One more piece of archaeology matters. Release 23's notes record that Riposte was changed to buff Damage Avoidance instead of Damage Resistance; Release 49's notes record the deprecated Riposte Defense masterwork being removed in favor of Riposte Power. Put together with the code as it stands today, where the stat named Riposte Defense Power and Duration has a display name, an initializer, and zero readers, the conclusion writes itself: the defensive half of Riposte existed, was walked back in stages, and the stat was left orphaned mid-retreat. S6 is not an invention. It is a restoration.

What players defend

The same threads that complain also draw a line around what must not change, and this page treats that line as law. Rend is the tree's crown jewel: the best damage in the whole tree in jmcclain's words, never misses in Trihugger's, an excellent cherry on top in Brass Knuckles', and Alley Oop values the charge mechanic that takes a 1 to 24 bleed to 1 to 60. The DoT identity is the school's endgame texture, rend and hamstring and poison and sawblade stacked against regenerating bosses. The crit and speed identity is why 2H blades beat polearms in every comparison thread, Casey McCormick counting Vital Points' 6.3 percent crit and the faster animations. The combo richness is an asset waiting for a UI. The offhand artifact builds, Violet Ronso's spear with a Shard of Night, Daxxe Diggler's longsword with the same dagger, are beloved exactly as they are. The stacking skill ceiling is a cherished expert game; Toular's warning that removing stacking also removes combos is honored. And Boris Mondragon's at 150 you will think you are god is the power fantasy this page builds toward rather than away from. No card below reduces any shipped number. FrostII said it for everyone in the 2017 weapon-tree comparison: buff the weak trees instead.

Lessons: Ultima Online to modern

Sword skills are the oldest problem in the genre, and the solutions have converged hard. The short history:

What each game contributes

Ultima Online. Swordsmanship was pure hit chance for six years while Tactics multiplied damage and Anatomy added the flourish, which is precisely the take Tactics and Life instead structure players accuse SotA's blade spec of today. The 2003 Age of Shadows revamp added thirteen mana-costed special moves, each weapon carrying a primary and a secondary unlocked at skill 70 and 90: Bleed Attack, Armor Ignore, Whirlwind, Mortal Strike the heal-blocker, Feint, Disarm. A katana fighter and a broadsword fighter with identical skill points played differently for the first time. What Blades takes: depth can be retrofitted onto a finished school without touching its base progression, and the retrofit can become the identity.

Dark Age of Camelot. Every weapon style had an opening condition: anytime, positional, reactionary after your parry or evade or block, or chained after another style. Reactionary styles paid more precisely because they were conditional, and players queued a conditional with an anytime fallback. It is the closest ancestor of a riposte-window school ever shipped, and it is a quarter century old. What Blades takes: the chain. SotA's Riposte is a DAoC reactionary with no follow-up; S28 gives the counter a next beat.

World of Warcraft. Execute has been the sub-20-percent button since 2004; the Massacre talent raises the threshold to 35, validating a spec-raised execute line exactly like Coup de Grace's, and Sudden Death procs a free execute at any health because a button that is dead for most of the fight is a design bug, not a design. Overpower answered the enemy's dodge; Classic's rogue Riposte followed a parry with a counter that also disarmed, damage plus utility. Rend spent years falling out of rotations whenever it was only damage. What Blades takes: light the execute up mid-fight (S26), and let the counter carry utility (S6).

Final Fantasy XIV. The Samurai earns three named seals from three different combo finishers and spends them on iaijutsu whose shape depends on the count: one seal a potent DoT, two a cleave, three the great guaranteed-crit line. The gauge forces route variety, the exact medicine for a school whose most-cast rune outweighs its finisher three to one. The cautionary tale is Kaiten, a spend-to-buff button removed for being a tax with no decision attached. What Blades takes: S34 wholesale, with Kaiten's warning written on it.

Elder Scrolls Online. Twin Blade and Blunt: each dagger equipped adds crit rating, each sword adds weapon damage, each axe adds bleed chance, each mace adds penetration, and dual wield counts both hands. The entire sword-versus-dagger identity question, answered by one data-driven passive with four rows and zero new animations. Its known weakness is invisibility, which a character-sheet line fixes. What Blades takes: S23 nearly verbatim, and the both-hands read in S14.

Guild Wars 2. The spellbreaker's Full Counter is a prediction, a brief brace that negates the next hit and answers it, generous enough to work at network latency. The daredevil turns the dodge itself into an attack. Both weld defense into offense as one input. What Blades takes: the honesty about latency. SotA's stat-based windows are already the right architecture; the brace-stance itself is Tactics' territory and stays there, in the rejected list.

Path of Exile 1 and 2. Bleed deals triple against moving targets, a behavioral trigger that creates play instead of a flat DoT; PoE2's parry applies a debuff its counterattack consumes, and its over-specific consume condition is the community's chief complaint about it. Darkest Dungeon taught the same lesson from the other side: a bleed class facing bleed-immune enemies needs a fallback tool or entire regions turn its kit off. What Blades takes: bleeds that bite runners (S27), a Notch instead of nothing against the bloodless, and loose consume conditions everywhere.

Sekiro and Elden Ring. Posture is the genre's cleanest loop: every deflect damages the meter, the full meter opens the deathblow, and low vitality slows the meter's recovery so damage and defense are two roads to one execute. Elden Ring's guard counter and stance breaks put a rougher version in an RPG stat frame. Timing-perfect parries are unreachable at MMO latency, but the structure, defensive success filling an offensive meter that opens the finisher, ports intact. What Blades takes: S26, the school's two orphaned signatures welded into one loop.

The CRPG shelf. D&D's finesse rule splits the sword family by which attribute scales; Battle Brothers' dagger Puncture ignores armor entirely at a heavy action cost, arguing the armor-ignore identity belongs to daggers; DDO's Kensei deepens the chosen weapon group's crit profile as its capstone, the pattern for a spec that feels like mastery of a weapon rather than a stat dribble. What Blades takes: the dagger and sword paths of S33, and crit-forward specialization as a legitimate identity choice in S18.

The capability matrix

Ten capabilities a blade school can have, across the games that did them best, with both SotA columns highlighted: where the school stands today, and where the catalog takes it.

The thirty-four proposals

Four tiers. Tier 1 is paper and pins: defects, dead content, honesty, and the tests that hold everything else still. Tier 2 makes the school's existing verbs work the way they read. Tier 3 is identity: the family, the forms, the counter answered, the loop closed. Tier 4 is the owner's desk, where the questions that need a ruling wait with their evidence attached. Every card carries its size, its server cost, and the players who asked for it by name.

Tier 1 · Fix first — paper, pins, and dead content

Tier 2 · The verbs made right

Tier 3 · The edge and the answer

 HAND
 ┌────┬────┬────┬────┬────┬────┐
 │ T  │ DS │ RND│ WB │ CdG│ R  │
 │~~~~│~~~~│    │    │    │    │
 └──┬─┴─┬──┴────┴────┴────┴────┘
    └──┘ both rims shimmer: these two combine

 FORMS ............................. 2 of 5 found
 Quad Slash       Thrust + Double Slash      FOUND
 Sawblade Sweep   Whirling Blades + Rend     FOUND
 ???              a thrust and something cold
 ???              a thrust and something burning
 ???              a wound and a stumble

Tier 4 · The owner's desk

Considered and rejected

Sixteen ideas were weighed and set down, each with its owner named, so the next reader does not re-litigate them.

• Locked-glyph economy rework (charge-on-appear, locked-deck support skills, the fizzle and stack asymmetry): a deck-system question touching every school, argued well by Seggallion and FBohler; it belongs to a future deck page, and S6 answers the Blades-sized corner of it.
• A brace or full-counter stance: active defense real estate is the Tactics page's, and predictable enemy swings would trivialize it in PvE; the payoff lives in the riposte instead.
• First Blood, a duel-opening brand: value is nearly all player-versus-player; it belongs on the PvP page if anywhere.
• A fresh-draw opener bonus: punishes locked decks, which is the exact complaint the stacking thread already filed; a deck-page question.
• Drawn-versus-locked style split (fast steel and heavy steel): mutates the shared deck-lock math under every school; a deck-page question.
• Whirling Blades spreading bleeds natively: cannibalizes Sawblade Sweep's reason to exist while that combo is alive and healthy; revisit only as a far-future mastery term.
• Impale-style damage echo: the most bookkeeping for the least felt effect of anything on the genre shortlist.
• Tempo stacks for varied glyph play: anti-synergy with locked-slot builds; rejected on the same grounds as the opener bonus.
• Positional rear and flank openers: hostile to loose network positioning; the behind-target rules that exist already serve this.
• Timing-based parries or posture meters: unreachable at network latency; the stat-based window architecture is correct and stays.
• A weapon attack-speed stat: the weapons page rejected it with reasons; not reopened.
• Blademaster titles and doctrine naming for innates: the Tactics page's ladder card owns the title vein.
• A stealth opener: the Subterfuge page's quarry mark owns openers from hiding, and its coating cards own blade venom.
• Weapon oils and coatings: Subterfuge and Alchemy territory, per the weapons page's own routing.
• Unique blade components, compare tooltips, honest damage readouts, masterwork rails: the weapons and enhancement pages own all four surfaces.
• A blanket melee damage retune: the balance program's, with S32 carrying the school's evidence to that table.

The blade, answered

Thirty-four cards sounds like a new interface. It is not. Everything this page proposes lands on four surfaces that already exist: the target frame gains two marks, the buff strip gains one tile, the deck bar gains two glow states, and the character window gains one panel. Nothing else changes shape.

 ZONE B · TARGET FRAME (gains two marks)
 ┌──────────────────────────────────────────────────┐
 │ Ferocious Red Spider                             │
 │ ██████████████▌│░░░░░░░░░░░░  28%           │
 │                ▲ execute line 32.5%             │
 │ REND x3  11s   GUARD CRACKED x2                 │
 └─────────────────────────────────────────────────┘
 shows the word BLOODLESS, then a NOTCH, when immune

 ZONE C · BUFF STRIP (gains one tile)
 [PARRY WINDOW 3.2s] [RIPOSTE READY] [FOLLOW-THROUGH 4s]

 ZONE A · DECK BAR (gains two glow states)
 ┌────┬────┬────┬────┬────┬────┬────┐
 │ T  │ DS │ RND│ WB │ CdG│ R  │ L  │
 │    │    │~~~~│~~~~│ARMD│    │    │
 └────┴────┴────┴────┴────┴────┴────┘
 ~~ combo shimmer · ARMD execute armed · L locked

 ZONE D · CHARACTER WINDOW, ONE NEW PANEL
 ┌─────────────────────────────────────────────────┐
 │ Blade Specialization 100                         │
 │   Coup de Grace line      20%    now 32.5%       │
 │   Riposte window          2.0s   now 4.5s        │
 │   Rend heal debuff        1.0x   now 1.63x       │
 │   Whirling Blades cost    15     now 9.2         │
 │ Weapon family: two handed sword. Leans damage.   │
 └─────────────────────────────────────────────────┘

Zone B carries S19's execute line, S20's wound readout with its Bloodless and Notch states, and S26's guard cracks. Zone C carries the existing parry window joined by S28's follow-through. Zone A carries S24's combo shimmer and the shared armed glow of S19 and S26. Zone D is the specialization panel that answers Stabbath's 2019 question in one screen: what does blade specialization do, with your own numbers, before and after. The window timeline underneath it all:

flowchart LR
  PE["Enemy swing parried\nor glancing blow taken, S21"] --> W["Window runs\n2 x (1 + mastery) seconds"]
  W --> IN["Riposte inside\n3x, ignores armor"]
  W --> OUT["Riposte after\n0.75x"]
  IN --> C["Counter lands\navoidance buff granted, S6"]
  C --> F["Follow-through\nnext glyph empowered, S28"]
  C --> G["Guard cracked\nthree cracks expose, S26"]
  G --> X["Coup de Grace\ntreats exposed as under the line"]
  style F stroke-dasharray: 5 5
  style G stroke-dasharray: 5 5
  style X stroke-dasharray: 5 5

Sequencing

Five waves, honesty first. Wave zero ships with this page: the corrected book and the golden pins, plus the poison fix already built in the migration branch. Wave one is the fix-first sweep, S2 and S5 and S9 and S10 and S13, with the S12 ruling and the S7 retest requested up front and the S8 audit started, because its findings steer two later cards. Wave two makes the verbs right: the promise kept, the root fed after S12 rules, the other hand counted, the gradients, the guard, the visible execute and wound, the counter armed at home, the typed steel after S8 reports. Wave three is identity: the wider mastery once S6's buff exists, the family passives on S14's helper, the forms page after S9's cooldowns are honest, the never-dead glyph, the answered strike once ripostes fire natively, the first hour. Wave four closes the loop and clears the desk: the opened guard, then Misericorde on top of it, the ledgers and the native-edge review with their evidence packs, and the far pair, the doctrine fork and the three edges, waiting on their rulings.

flowchart TD
  subgraph WV0["Wave 0, with this page"]
    S1c["S1 the book, built"]
    S3c["S3 golden pins"]
    S4c["S4 poison fix, built"]
  end
  subgraph WV1["Wave 1, fix first"]
    S5c["S5 third hit"]
    S8c["S8 coup audit"]
    S9c["S9 combo cooldowns"]
    S12c["S12 double-count ruling"]
  end
  subgraph WV2["Wave 2, the verbs"]
    S6c["S6 promise kept"]
    S14c["S14 other hand"]
    S19c["S19 execute line"]
    S21c["S21 armed at home"]
    S22c["S22 typed steel"]
  end
  subgraph WV3["Wave 3, identity"]
    S16c["S16 wider mastery"]
    S23c["S23 two edges"]
    S24c["S24 forms revealed"]
    S27c["S27 never dead"]
    S28c["S28 answered strike"]
  end
  subgraph WV4["Wave 4, the loop and the desk"]
    S25c["S25 Misericorde"]
    S26c["S26 opened guard"]
    S30c["S30 crit ledger"]
    S32c["S32 native edge"]
    S33c["S33 doctrine fork"]
  end
  S3c --> S5c
  S3c --> S8c
  S12c --> S6c
  S8c --> S22c
  S8c --> S30c
  S6c --> S16c
  S14c --> S23c
  S9c --> S24c
  S21c --> S28c
  S21c --> S26c
  S19c --> S26c
  S26c --> S25c
  S24c --> S25c
  S16c --> S33c
  S23c --> S33c
  S30c --> S32c

Appendix

A1 · Why almost everything here is a data change

Combat in this game is a synchronized client simulation; the server relays which glyph fired and holds the shared identifiers. That is why thirty of thirty-four cards are client data, formulas, and interface: rune behavior lives in authored expressions that a balance changeset can edit, and only the cards that add new shared identifiers or persistent records, the guard, the marks, the fork, the new combo, touch both code trees, always in one synchronized change.

A2 · Constraints and fences every card honors

No card reduces any shipped number, per the philosophy players set themselves. Rend's base bleed is untouched everywhere. The artifact offhand builds are untouched; S14 only adds eligibility. Stacking play is untouched; the combo cards ride it. Specialization changes are strictly upward for every existing specialist. Territory owned by other pages stays theirs: the Parry window, reaction glyphs, the maneuver pool, and counter-stances belong to the Tactics page; openers from stealth and blade venom to Subterfuge; the dual-wield compare tooltip, unique components, and damage readouts to the weapons page; masterwork rails to the enhancement page; and the wider melee retune to the balance program, with S32 as this school's evidence brief to that table.

A3 · Test surface

One new golden suite pins the six rune expression sets, the nine specialization riders, and all five combo cooldown pairs before any number moves. New suites arrive with their cards: poison eligibility per rune, the Double Slash hit sequence, the riposte buff, both-hands eligibility, the family matrix, the mask-to-Notch path, the follow-through boundaries, and the guard stacks with their decay and cap. Every suite runs scoped, one class at a time, in the project's ported editor test tree.

A4 · Player asks answered

Rinaldi, the Thrust poison report, S4 built. Stabbath and jmcclain, the invisible specialization, S2, S15, S16, and zone D. Barugon, spec numbers in tooltips, S2. KenoZuki, Nog, and Alley Oop, the Riposte record, S6 and S21. Jason_M, Sarek, Gravidy, golruul, Ryodin Stormwind, FrostII, Cora Cuz'avich, Boris Mondragon, and ErikRulez, the coup's accountability, S8, S19, and S30. Crazy Phil, KuBaTRiZeS, Smalls, Time Lord, Noric, and agra, the hidden forms, S24. Toular, the stacking constraint, honored throughout. Seggallion and FBohler, the locked-deck economics, S6's rent and the rejected list's honest routing. Khalest, Kain Darkmoor, Ben_Hroth, and Warrior B'Patrick, the other hand, S14. Violet Ronso and Daxxe Diggler, the protected offhand builds, fenced. Yipp, redfish, Curt, and By Tor, the family question, S23. Casey McCormick, Trihugger, null2, Toadster, Brass Knuckles, and Bow Vale, the school's crowned identity, protected. FrostII, buff the weak instead, the page's law. Syee Kco and necronut, the endgame question, S32, both positions on the record.

A5 · Data sources and method

Live telemetry: per-cast skill logging over the thirty days ending August 14, 8,157,099 casts from 1,236 distinct players, exact per-rune counts and exact caster cardinality, with top-caster concentration reported anonymously. Census: the specialization popularity study of 120,337 characters, same date. Code truth: every formula quoted was read from the shipped rune data and combat code, and the seven errata in S1 were verified against the data before the doc was corrected. Forums: eleven threads from 2016 to August 2026, every requester credited by exact handle; Steam threads were read for background only and are never credited. Genre: primary wikis and design retrospectives for the nine games in section 2. All proposals here are proposals; nothing on this page is a commitment until its card ships, and the two cards marked Built are the exceptions that already did.

General comments

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.