The Empty Hand

Shroud's combat is a card game. Skills are glyphs, drawn from a deck you built into a rolling hand, played, cooled down, drawn again. Focus is the school that owns that game: the pool every caster spends, and the game's only two hand verbs, a full mulligan and a hold. The numbers say almost nobody plays it. Ninety percent of all characters ever made have experience in the school's root without trying, because Focus innates train themselves while you do anything else. Four percent of players ever learned a hand verb properly. A fifth of one percent specialized, and the team's own balance notes call the result the weakest specialization in the game. Meanwhile, seventy-nine percent of everything the school does on the live server is one instant upkeep button, and three quarters of those presses come from five people. This page measures the school against the games that invented its ideas, Ultima Online's Meditation on one side and the modern card games on the other, and proposes twenty-eight changes built on one conviction: the deck school should be the school that actually plays the deck.

Executive summary

The genre has run the resource-school experiment for thirty years, and the verdict is unusually clean. Ultima Online shipped skills literally named Meditation and Focus; the parts that made them play were the frictions, an interruptible trance that revealed you to your enemies, an armor gate that priced regen against defense, and the parts that were pure percentages took two expansions of retrofits to become interesting. Guild Wars built the genre's one true mana school and proved players will spend their most valuable slot on a resource verb, then spent years nerfing the infinite engines it bred. World of Warcraft removed mana micromanagement piece by piece, the five-second rule, the Spirit stat, Life Tap, because a stat with no verbs attached is a tax, and replaced it all with a crit-proc loop players love. And the deck games Shroud quietly belongs to, Slay the Spire, Magic, Hearthstone, Balatro, spent the same decades growing a rich vocabulary of hand verbs: retain, scry, mulligan, trade, exhaust. The lesson across all of them: a resource school earns commitment when its skills are verbs with timing and prices, and never when they are percentages.

Focus already half-fits that lesson. It owns the game's only two hand verbs. It also owns the pool, the fizzle insurance, the summoner's support line, and the elixir stomach. What it does not own is anyone's commitment. Thirty days of live-server telemetry and a full census of the live database put the school's condition in three clusters:

1. The two-verb hand. The deck school has exactly two in-combat moves, and both are blunt instruments. Mind Wipe discards everything and redraws, no shaping, and quietly charges full cooldowns for any combos it wipes. Mind Lock extends every timer at once, all or nothing, at 20 seconds whether you are level 1 or level 100. The genre grew ten verbs in the same years; six dead skill identities have sat in the school's data for a decade waiting to hold them.
2. Trained by everyone, chosen by no one. Any adventuring skill you use quietly feeds your learning Focus innates, so 90 percent of every character ever made has experience in the root. From there the funnel collapses: 17 percent of accounts pushed the root to a deliberate level 40, 3.8 percent ever took a hand verb that far, 0.22 percent learned the specialization, and 89 accounts in the game's history reached specialization 80. Twenty-nine active characters carry it today. The team's own 2026 balance changeset says the quiet part aloud: Focus is the weakest specialization in the game, six mastery terms, on the school's own outlier multiplier that reads 0.8 at level 80 where every other school reads 1.0.
3. An economy without levers. The focus economy has stats with no skills attached. Sprint drain has a discount stat consumed by exactly one line of engine code, and in the whole game only a blessing and a raw banana write it. Summon upkeep rents a slice of your maximum focus and nothing reduces the rent, a complaint players have filed since 2017. Regen has been asked for as a trainable line since 2014. Not one item in the game grants Focus Mastery. And none of it is visible: no ledger, no timers on the hand, no cue when Aetheric Feedback's crit window opens.

The deal that runs through all twenty-eight proposals: Tier 1 makes the two verbs trustworthy, honest names, honest icons, honest tooltips, visible timers. Tier 2 deepens them, a shaped mulligan, a hold that scales, a stance that survives a walk, the first sprint discount, a focus ledger, real summon support. Tier 3 brings the ghost shelf back to work: six dead skill identities revived as the missing vocabulary, replay, reactive draw, motion, clarity, burn, taunt. Tier 4 asks the owner's questions in the open and gives the deck school's specialization a deck reason to exist. First the hand tells the truth; then the hand learns to play.

Twenty-four of the twenty-eight need no server work at all. The hand is simulated entirely on your machine, which is why most of this page is buildable as client data and client code.

The school we have

Fifteen skills on the Strategy tab: eleven innates and four castable runes in a tree rooted at Heightened Awareness. The innates raise your focus pool and Intelligence, cut glyph and deck-swap costs, resist fizzles and interrupts, feed summoned creatures, and improve potions. The four runes are where the school stops being a stat sheet: a full-hand mulligan, a hold that keeps your glyphs from expiring, a caster's stance that trades defense for tempo, and a crit-triggered focus surge. Two of those four are the only moves in the entire game that touch the hand itself.

[ Heightened Awareness ] ................ (1)
├─ Train Intelligence
│  └─ Focus Specialization ........... (2)
├─ Mind Wipe ......................... (3)
│  ├─ Mind Lock ...................... (4)
│  └─ Spellbinder Stance ............. (5)
│     └─ Aetheric Feedback ........... (6)
├─ Quick Swapping .................... (7)
│  └─ Inner Calm
│     ├─ Follow Through .............. (8)
│     └─ Tranquility ................. (8)
├─ Summoned Health ................... (9)
│  ├─ Summoned Toughness
│  └─ Summoned Damage
└─ Elixir Receptive .................. (10)

┌─────┐┌─────┐┌─────┐┌─────┐┌─────┐
│Fire ││Heal ││Light││Death││Mind │
│Arrow││Touch││ning ││Touch││Lock │
│█████││███░░││█░░░░││█████││████░│
└─────┘└─────┘└─────┘└─────┘└─────┘

• 1The accidental school. Using any adventuring skill quietly feeds every learning Focus innate, so 90 percent of all characters ever made have experience here. It is the most-trained school in the game by accident and the least-committed by choice. Cards S8, S27.
• 2The last-place specialization. Every other school's mastery stat reads 1.0 at specialization 80. Focus reads 0.8, on six mastery terms, the fewest in the game. The team's own balance ack calls it the weakest specialization outright. 89 accounts ever reached 80; 29 active characters carry it. Cards S8, S9, S27.
• 3The mulligan nobody presses. Mind Wipe is the most-learned active in the school, 4,414 accounts took it to level 40, and the least-cast, 2,009 casts by 37 players in thirty days. Players bought the verb and stopped pressing it. It redraws blind, and it charges full cooldowns for wiped combos while plain glyphs recycle free, a price printed nowhere. Cards S3, S11.
• 4The hold that became a chore. Mind Lock: 205,365 casts in thirty days, 44 players, and 74 percent of those casts from five people, roughly one press per cooldown sustained for hours. Its extension is 20 seconds at level 1 and 20 seconds at level 100. Cards S10, S27.
• 5The stance that dies to a step. Spellbinder trades 20 damage avoidance and 20 damage resistance for faster draws, faster casts, and spell crit, and all seven of its effects cancel the frame you move, at any speed. Players asked for a slow-walk allowance in 2017. It also shares a buff channel with Chaos magic's Chaotic Feedback, so the two overwrite each other, and the 2020 endgame consensus was that Chaos does the job better and lets you move. Cards S3, S12.
• 6The invisible window. Aetheric Feedback only fires in the seconds after your own magic crit. Nothing on any player's screen shows the window; the only visualization that exists is a testers' debug overlay. Card S13.
• 7The name that is not true. Quick Swapping does not make deck swaps quicker. It cuts the focus cost of the swap; the swap time is untouched, and players called the name a lie in 2016. Card S1.
• 8The bodyguards. Follow Through cuts fizzle chance and Tranquility resists interrupts, but the school's own runes never fizzle and two are instant. These innates protect your other schools' casts, which is a selling point their tooltips never make. Card S7.
• 9The summoner's tax office. Every magic summon rents a slice of your maximum focus for as long as it lives, and no skill anywhere reduces the rent, a complaint filed in 2017 and still alive last month. The three Train Summoned innates are correctly wired as of this month's fixes; the remaining problem is that their magnitudes underwhelm, on a line whose training price is four times normal. Cards S16, S17, S23.
• 10The stomach. Elixir Receptive quietly multiplies every potion you drink, and is mutually exclusive with a Tactics skill, a rule the game states nowhere. Cards S5, S26.

Who actually plays it

Two instruments, read together. The live cast log answers what people press; the character census answers what people trained. First the presses, thirty days of live server:

Live server, 30 days ending August 15, 2026. Counts are floors: the client rate-limits its own event log under load. Denominators: 8,164,352 total casts by 1,237 distinct casters that month. The school's 258,597 casts are 3.2 percent of everything, and 124 casters, one in ten, pressed any Focus verb at all. Strip the five heaviest Mind Lock users and the school is roughly 1.3 percent.

Then the census, every character ever made, against a deliberate-investment bar, enough lifetime experience in the skill to reach level 40, so the passive trickle cannot masquerade as choice:

Census of the live database, August 14, 2026: 120,337 characters on 116,361 accounts. The first bar is the bleed-through: Focus innates train while you do anything. Every bar after it is a choice. One structural note: the whole rune branch hangs off Mind Wipe, so the 4,415 accounts with any verb at 40 are, to within one account, exactly the Mind Wipe population. The gate to the school's entire active game is the button almost nobody presses.

How a glyph lives and dies

Every proposal on this page plugs into one loop. A glyph is drawn from your deck into a hand slot, where a timer immediately starts running toward a silent discard. Play it and it goes to the cooldown pile, paying focus scaled by the school's discounts; let it expire and it slips away on its own. Focus itself regenerates on two different budgets, in and out of combat, and drains through casting, sprinting, jumping, deck swaps, and summon upkeep. The school's two verbs, and nearly every card below, are interventions in this diagram:

flowchart LR
  D[("Draw pile")] -->|"automatic draw"| H["Hand slot + discard timer"]
  H -->|"play: focus cost x Inner Calm"| C["Cast resolves"]
  H -->|"timer expires: silent discard"| U[("Cooldown pile")]
  C --> U
  U -->|"cooldown done"| D
  MW["Mind Wipe: clear + redraw all"] -.->|"S3 S11 S27"| H
  ML["Mind Lock: extend all timers"] -.->|"S4 S10"| H
  SG["Second Guess: trade one glyph S25"] -.-> H
  F[("Focus pool")] ---|"pays"| C
  R["Regen: two budgets, in and out of combat S20 S24"] --> F
  F -->|"sprint S14, summons S16, swaps, jumps"| DR["Drains"]
  AF["Crit window: temporary max focus S13"] -.-> F

Lessons: Ultima Online to Balatro

Focus sits at the meeting point of two genre histories: the mana school, which mostly failed, and the hand verb, which mostly succeeded. Both histories are unusually well documented, and both have spent thirty years converging on the same conclusions.

Ultima Online: the ancestor with our skill names

UO shipped skills literally named Meditation and Focus. Meditation was two skills in one: a passive that raised baseline mana regen, and an active trance, you stop, you cannot act or speak, being struck breaks it, and meditating reveals you if you were hidden. It also refused to work in heavy armor, pricing regen against defense. Focus, added in 2003, was the warrior's version: fully passive stamina and mana regen, no armor gate, no decisions. Twenty-five years of retrofits later the pattern is legible: the parts of UO's resource skills that created play were the frictions, interruptibility, vulnerability, the armor trade, and the passive parts were trained because the math said so and remembered by nobody. Card S24 is built directly on this lesson.

EverQuest, FFXI, DAoC: the battery is a person

The classic MMOs made mana support a social identity rather than a self tree. EverQuest's Clarity line was so valued the community named it after a drug and sold it at gathering points; FFXI's Red Mage Refresh made the class permanently in demand and eventually enslaved it to the job; City of Heroes' Transference drained a foe's endurance into every nearby ally and became one of the most loved support verbs ever, because it was active, aimed, and enemy-funded. DAoC added the elegant variant: buffs there lock a slice of a dedicated concentration pool while they run, upkeep as reserved capacity, which is exactly how Shroud's summons already rent your focus. The lesson cuts two ways: aimed resource support is the one resource identity players provably commit to, and it is also the one thing Shroud's client-local focus pool cannot deliver today. The rejected list holds that honesty; card S16 takes the DAoC half.

Guild Wars: the one true mana school

GW1's Inspiration Magic remains the genre's only fully realized mana school: energy steal, interrupts that pay you, mantras that convert incoming elements into energy, whole elite slots spent on nothing but economy. It worked because energy was genuinely scarce and every economy skill was an active with a condition and a price. Its failures are as instructive: stacked engines produced infinite-energy builds that ate years of nerfs, and energy denial, the enemy-facing twin, was miserable in PvP and pointless in PvE. Cards S9 and S22 take the successes; the rejected list takes the warning.

World of Warcraft: the great removal

WoW spent a decade dismantling mana micromanagement. The five-second rule, regen only if you stopped casting, was removed in 2010 because optimal play meant not playing. Spirit, the regen stat, shrank expansion by expansion and was deleted in 2016 because a stat with no verbs generates gear math and nothing else. Life Tap, the health-for-mana conversion that defined Warlocks, went in 2018 and was mourned, because conversion verbs are identity. What replaced it all is the modern loop: Evoker's Essence Burst, where crits light up a visible proc and your next spender is free. Shroud already owns that loop's skeleton, Aetheric Feedback fires on your magic crit, and hides it from the player completely. Card S13 is the missing half, and the deletion history is the warning label on every passive-regen suggestion this page declined to make.

The card games: our two verbs have famous cousins

Because Shroud's combat is a deck, the card battlers are direct precedent, not metaphor. Slay the Spire's Retain is Mind Lock exactly, and its refinements, choose what to retain, retained cards mature, are the obvious growth path. Scry turned draw randomness into planning and became evergreen in Magic because cheap smoothing makes games feel fair without warping them. Magic's London mulligan closed a 25-year design arc with the rule Shroud's mulligan is missing: keep the cost in quantity, give the player agency over quality. Hearthstone's Tradeable keyword, pay a little mana, shuffle one card back, draw another, is the single most Shroud-shaped verb in the research; The Coin proved a resource surge is best delivered as a card you time. Balatro made discards themselves a countable, buildable resource. Cards S10, S11, S25, and S22 each import one of these, sized for a real-time game: pre-commitment and one-click verbs only, no mid-fight menus.

Tabletop psionics: the discipline fantasy

D&D's psionic focus is the cleanest fiction for what this school sells: spend a quiet moment to become centered, hold that state like a banked coin, spend it for one guaranteed act of will. Concentration as a defensive stat, the caster who cannot be rattled, is Tranquility's fantasy already. The vocabulary the fiction shelf offers, the flame and the void, wells of stored power, the litany against fear, battle meditation, is the register this school's strings should be written in. Cards S24 and S7 borrow it.

The capability matrix

One honest finding closes the section. Across thirty years, no MMO made a self-directed regen tree worth committing to; every success was either a verb with timing and a price or support aimed at another person. Focus already owns the verbs, two of them, blunt and unloved but real, and a stack of dead skill identities that were clearly meant to become more. The genre's advice is unambiguous: keep the passives as the floor, and build the school on its verbs.

The twenty-eight proposals

Tier 1 makes the school trustworthy: honest names, honest icons, honest prices, visible timers, and the paper trail that licenses everything after it. Tier 2 deepens the two verbs the school already owns and pulls the levers the focus economy already has. Tier 3 revives the ghost shelf, six skill identities that have sat in the data for a decade with no skill behind them, each returning as a piece of the vocabulary the genre grew while this school stood still. Tier 4 is the capstone tier: the Ultima homage, the missing third verb, itemization, and the owner's three open decisions, stated with numbers and no promises. Every tuning number is an illustrative starting point: final values ship through the balance program's test-shard changesets and measurement windows.

The deal, stated once. The verbs become trustworthy before they become deeper; every mastery formula amplifies and never gates, meaning every skill works exactly as today for a player with no specialization; nothing on this page adds a mid-fight menu, every new interaction is a single click or a pre-fight choice; and the one-line identity the page tunes toward: the deck school should be the school that plays the deck.

Tier 1 · Trust — names, prices, timers, and the paper trail

S1 · Call the skills what they are

Tier 1Size SServer: noneplayer ask

census: Quick Swapping 2,761 active characters at level 40 or higher

Quick Swapping does not make deck swaps quicker. It cuts the focus cost of a swap; the eight seconds are untouched, and players called the name a lie back in 2016. The skill gets a display name that matches its function, Efficient Swapping or similar, with a tooltip that says plainly what it reduces. The same pass settles the school's other identity tangle: the skill the data calls Concentration displays as Tranquility, while a Taming skill also displays as Concentration, two different skills wearing each other's names in guides for years. One public name each, stated once, kept forever. The internal data names stay frozen, they are save-format, so this is entirely a display and description pass.

Seen in: Ultima Online, where Meditation and Focus described exactly what they did for twenty-five years. Requested by Xander Xavier and Agra (2016).

Implementation notes

Display names and descriptions only, in the localization data; the internal skill identifiers and the historic misspelling in one of them are serialization-frozen and stay, documented in the school reference rather than fixed. A one-release "formerly Quick Swapping" line in the tooltip eases the guide churn.

client S (loc only)server noneimpact names stop lyingtests loc-coverage check for all 15 skills

S2 · Give four skills their own icons

Tier 1Size SServer: none

census: the summon trio 1,229 / 954 / 970 active characters; Elixir Receptive 720

Open the Focus tree and four skills are wearing borrowed art. Train Summoned Pet Health, Resistance, and Damage all display the specialization's icon, three identical purple emblems in a row, and Elixir Receptive wears Quick Swapping's. A tree you cannot read at a glance is a tree you do not come back to. Four new icons, one per skill, and the tree stops looking like a photocopy.

Seen in: every card game ever printed, which gives each card its own art for the same reason.

Implementation notes

Four new icon files and four cell edits in the skill sheet. The icon pipeline resolves by filename across every icon set, so each new file lands in each set, and the test-shard bundle cache will need its usual nudge before the art shows there.

client S (art + data)server noneimpact a readable treetests no two Focus rows share an icon file

S3 · Tooltips that price the trade

Tier 1Size MServer: noneplayer ask

30 days live: Spellbinder Stance 32,365 casts by 66 players; Aetheric Feedback 18,858 by 51

Four tooltips, four missing prices. Spellbinder Stance says faster casts; it does not say you pay 20 damage avoidance and 20 damage resistance for the privilege, or that a single step cancels all of it. Aetheric Feedback says after a critical hit; it does not say the window is about four seconds, growing to six with full specialization. The specialization tooltip gestures at improvement; it does not list the six concrete things it improves. And Mind Wipe does not mention its one real trap: wiped combos go to full cooldown while plain glyphs recycle instantly, so wiping a hand you just stacked is an expensive mistake the game lets you discover alone. Every price gets printed where the purchase happens.

Seen in: Slay the Spire's energy relics, which print their downside on the card face, and are trusted precisely because of it. The stance confusion goes back to Gatsu's 2017 thread; the window confusion to Filter Bubble Algorithm's 2017 report.

Implementation notes

Localization edits plus tooltip logic for the computed values, the window seconds and the stance's rider list read from the same data the effects run on, so the words cannot drift from the numbers. A small drift alarm in the test suite asserts the stance's two defensive penalties are still what the text claims.

client M (loc + tooltip logic)server noneimpact every trade priced up fronttests tooltip-vs-data drift alarm

S4 · Numbers on the sand

Tier 1Size S–MServer: none

30 days live: Mind Lock 205,365 casts by 44 players

A glyph in your hand is always dying, and the only clock the game gives you is a slow fade that begins a few seconds before the end. Mind Lock's entire product is time, and its only receipt is a thin border that says extended without saying how long. This card makes the hand readable: an optional per-slot countdown in seconds, off by default for players who like the sand, and a clearer held state that shows the extension Mind Lock actually bought. The school's most-pressed button finally shows what it does.

Seen in: every deck game with a timer, which shows the timer. The blind upkeep pattern in the live data, five players pressing the hold on cooldown for hours, is what pressing time blind looks like.

Implementation notes

The hand slot already stores an exact discard deadline and an extension stamp; this is surfacing, not simulation. One settings toggle, one small text element per slot, and a stronger treatment of the existing extended highlight. The manual right-click discard verb that already exists behind a default-off switch is noted here for the owner: if it ever ships, this display work is its prerequisite.

client S–M (UI)server noneimpact the hand becomes readabletests deadline-to-display math

S5 · Declare the elixir pair

Tier 1Size SServer: noneplayer report

census: Elixir Receptive 720 active characters at level 40 or higher

Elixir Receptive, in this school, and Elixir Extension, in Tactics, are mutually exclusive: set one to learning and the other quietly starts unlearning. It is a real and reasonable rule that the game states exactly nowhere, and players have been filing the resulting confusion as bugs for years. Both tooltips now declare the pairing in one plain sentence, and the skill window shows a small paired marker on each. The Elixir Receptive tooltip's separate accuracy defect is already filed and fixed on the Tactics page as #4486; this card adds only the declaration.

Seen in: Guild Wars elite skills, one slot, stated as a rule players plan around. Reported by Barugon (2023).

Implementation notes

Two description lines and a small marker in the advancement window near the existing toggle logic, which already resolves the companion skill. No behavior change; the exclusivity itself is tested and stays exactly as it is.

client S (loc + marker)server noneimpact the rule is finally written downtests existing pair tests stay green

S6 · Sweep the ghost shelf

Tier 1Size SServer: none

no live data. Six dead SkillIDs, no rune, no row, ever

Six Focus skill identities exist in the game's data and have never once been playable: Higher Consciousness, Consumption, Eidetic Memory, Mental Reflex, Mindfulness, and Summoned Pet Taunt. Five of their icons still sit on disk, one of them with a historic misspelling in its filename, along with an orphaned deck-swap icon no skill has ever referenced. This card is the inventory: each ghost recorded as never-implemented rather than retired, identities frozen, art catalogued. It is deliberately boring, and it is the license for Tier 3, where every one of them comes back to work.

Seen in: the Moon page's ghost-labeling card, which turned that school's dead entries into two revived spells.

Implementation notes

Documentation and a small inventory test: the ghost identities are absent from the retirement list by design, since retirement implies a refund path they never needed. A seventh identity, an orphaned second specialization once aimed at this school, stays excluded: it was formally retired this month and its fiction is spent elsewhere.

client S (docs + test)server noneimpact the revival tier stands on a recordtests ghost inventory pin

S7 · The innates that protect other schools

Tier 1Size SServer: none

census: Follow Through 1,728 active characters; Tranquility 1,584

Follow Through cuts your fizzle chance. Tranquility hardens your casting against interrupts. Here is the quiet joke: the school's own runes never fizzle, and its hold is instant. These two innates protect almost nothing in their own tree; they are bodyguards for your other schools, the fire mage's insurance, the healer's composure. That is not a flaw, it is the school's oldest identity, the discipline that steadies every art, and it is a selling point the tooltips have never once made. Both descriptions are rewritten to say who they actually serve, in the school's own voice: the calm that guards the fire.

Seen in: D&D's Concentration checks, the mind stat as armor for every other power; the litany against fear, borrowed as register.

Implementation notes

Localization only. The numbers do not move; the framing finally matches the function. This card also seeds the fiction voice the Tier 3 revivals write their strings in.

client S (loc)server noneimpact the school's oldest identity, statedtests none (editorial)

Tier 2 · The two verbs, deepened — and the levers the economy already has

S8 · A multiplier like everyone else's

Tier 2Size SServer: noneowner call

90 percent touch it, 17 percent commit, 3.8 percent learn a verb, 0.22 percent specialize

Every specialization in the game climbs its mastery stat by 0.0125 per level: 1.0 at 80, the number every formula written as one-plus-mastery was designed around. Focus alone climbs by 0.01: 0.8 at 80, permanently a fifth behind, on the school that also has the fewest mastery terms in the game. Whether that discount was ever a decision is not recorded anywhere. The team's own balance changeset from August says the conclusion aloud: Focus is the weakest specialization in the game. Parity is a one-cell change and a profiling pass. The numbers are on the table; the call is the owner's, and this page makes no promise.

Seen in: the genre's honest finding, a resource tree trained because the math requires it and specialized by nobody. The era verdict, from Violet Ronso in 2022: good, but would not recommend it to anyone.

Implementation notes

One scaling cell in the skill sheet, shipped as a balance changeset whose report profiles all six existing mastery terms at once, since every one of them strengthens together. No test pins the current curve; whichever curve the owner picks gets pinned that day.

client S (one cell, if approved)server noneimpact the spec stops being 20 percent behindtests pin the chosen curve

S9 · Give the mastery more to touch

Tier 2Size Mchangeset

census: 67 active accounts hold the specialization at level 40 or higher

Six mastery terms is the fewest of any school. This card adds four, every one of them an amplifier and none a gate: Mind Lock's cooldowns shrink with mastery, matching the grammar Mind Wipe already uses; Spellbinder Stance runs longer with mastery; and the Inner Calm and Quick Swapping discounts deepen with mastery. A player with no specialization notices nothing. A specialist finally feels the school answer back in more than two places.

Seen in: Guild Wars attribute lines, where every point moved every skill in the line, which is what made the line worth the points.

Implementation notes

The rune-side terms are one changeset of formula edits, cloning the grammar the school already uses elsewhere. The innate-side discounts depend on whether innate rewards accept formula terms; if they do not, the fallback adds the deepened discounts to the specialization's own reward rows, same result, different cell. The crit window's formula is pinned by tests and deliberately untouched here.

client M (formulas)server noneimpact six terms become tentests every new term positive at mastery zero

S10 · Mind Lock learns its level

Tier 2Size Schangesetplayer ask

30 days live: 205,365 casts by 44 players, three quarters of them from five

Mind Lock's extension is 20 seconds at level 1 and 20 seconds at level 100. The amount has no training term at all; only specialization moves it, which is one of exactly two things specialization does for it. Now the hold grows with the skill: 20 seconds at level 1 climbing toward 40 at grandmaster, doubled again by full mastery. The existing safety already in the engine, extensions cannot push a glyph past one full lifetime beyond now, keeps the number honest. Fewer presses buying more time is also the only humane answer to the five players currently pressing this button several hundred times an hour.

Seen in: Slay the Spire's Retain refinements, where holding scales with investment. Requested in spirit by Cinder Sear and Smalls (2016), who found the hold too weak for anything but pre-fight stacking.

Implementation notes

One amount formula gains a training term, shipped as a changeset with before and after tables at levels 1, 50, 100, and 200, at mastery zero and full. The clamp behavior gets its own pin.

client S (one formula)server noneimpact the hold finally trainstests amount pin at four corners + clamp pin

S11 · A mulligan you can shape

Tier 2Size MServer: none

census 4,414 accounts at level 40; live 2,009 casts in 30 days

Mind Wipe is the school's signature and its saddest number: the most-learned active, the least-cast. A blind full redraw is simply not worth a slot to most hands. Magic: the Gathering spent twenty-five years iterating exactly one verb and landed on the answer: keep the mulligan's cost fixed, give the player one shaping choice over the result. Mind Wipe gets the same: for five seconds after the redraw, you may right-click one glyph to swap it for the next card of your deck, free, once. One click, no menu, no pause. The panic button becomes a play.

Seen in: the London mulligan (2019), the endpoint of Magic's longest-running design iteration; requested in spirit by A'chelata (2017), whose wishlist asked for smarter draws.

Implementation notes

Client engine work in the deck controller: a short grace state armed by the wipe, and a replace-one-slot verb that retires the old glyph without cooldown and draws directly into the vacated slot, sidestepping the engine's one draw hazard, a drawn card whose slots are all occupied is silently lost, by never letting the situation arise. Locked slots and combos are skipped. The armed right-click interaction built here is reused wholesale by S25.

client M (controller verb + input)server noneimpact the mulligan becomes a decisiontests grace-window and replace-verb logic

S12 · Let the stance survive a walk

Tier 2Size SServer: noneplayer ask

30 days live: 32,365 casts by 66 players

Spellbinder Stance ends the frame you move, at any speed, all seven of its effects at once. Gatsu asked for the obvious fix in 2017: let a careful walk keep the stance and let running break it. That is this card. The stance tolerates walking pace; running and sprinting end it exactly as today. The 2020 endgame verdict, that Chaos magic's rival buff does the job better and lets you sprint, gets its answer not by nerfing the rival but by letting the binder move like a person. While it runs, the stance also finally announces itself: a buff bar entry naming its riders, including the two defensive penalties S3 printed.

Seen in: World of Warcraft's five-second rule, removed by its own designers in 2010, because rules that make not-playing the optimal play do not survive. Requested by Gatsu. (2017).

Implementation notes

The movement-cancel check compares animation speed against an epsilon; it learns to compare against a per-effect threshold instead, default zero everywhere so nothing else in the game loosens, set to walking pace on the stance's seven effects, with the value behind a tester-tunable tweak. An orphaned stealth threshold constant from years ago suggests someone meant to build exactly this once.

client S (one check + data)server noneimpact the binder can repositiontests threshold pin on all seven effects + zero elsewhere

S13 · Show the feedback window

Tier 2Size MServer: none

30 days live: 18,858 casts by 51 players

Aetheric Feedback is the school's best-designed loop and its best-kept secret: your magic crit opens a roughly four-second window, and casting the skill inside it surges your maximum focus and combat regen. Nothing tells you the window opened. The only visualization in existence is a testers' debug overlay. The fix is the modern proc pattern: while your window is open, the Aetheric Feedback glyph glows in your hand, the same visual grammar the hold's extended border already uses. Nothing about the skill changes; you can finally see it. For the owner's desk, stated without a promise: the skill's regen effect stays at a deliberate flat sixty seconds by an August ruling, and if visibility changes how that feels, reopening it is a decision, not a bug.

Seen in: WoW's Essence Burst, the proc loop that works because the proc is visible. The window itself shipped its scaling fix in August for Barugon's long-standing ask; this card is the visibility half.

Implementation notes

The reaction-window catalog already evaluates these gates for the testers' overlay; the same evaluation drives a hand-slot glow for the local player, throttled to the UI tick. No gate changes, no new HUD element, and the pinned reaction tests stay untouched.

client M (UI + eval reuse)server noneimpact the loop becomes playable on purposetests window-eval helper pin

S14 · The first sprint discount

Tier 2Size Schangeset

census: Tranquility 1,584 active characters at level 40 or higher

Sprinting drains ten focus a second, and a discount stat for it has existed in the game for years, consumed by exactly one line of engine code and written, in the entire game, by one blessing and one raw banana. No skill has ever touched it. Tranquility becomes the first: up to 12.5 percent off sprint's drain at level 100, the discipline innate finally covering the discipline of breathing while you run. The engine line gains a clamp at the same time, so stacked discounts can never turn sprinting into a focus fountain.

Seen in: UO's Focus, the mover's regen skill, no armor gate, built for people who cannot stand still.

Implementation notes

One innate reward line on Tranquility's row, and one clamp on the single consuming line, since the stat is additive and unclamped today. The clamp ships first; S20's revival stacks a deeper version under the same ceiling later.

client S (data + one clamp)server noneimpact the first skill in history to cheapen sprinttests clamp pin at three stack levels

S15 · The focus ledger

Tier 2Size MServer: none

1,237 casters spent this pool last month; the ledger serves every one of them

The character sheet tells you your maximum focus and, if you hover, where the maximum came from. It tells you nothing about the economy: what your glyphs cost after Inner Calm, what a deck swap costs after the swap discount, what sprint drains, what your summon rents, what your regen actually is in combat versus out, what Aetheric Feedback is temporarily lending you. One breakdown panel shows all of it, in the same format the sheet already uses for maximum health and weight. The resource school gets a resource statement.

Seen in: Guild Wars, where energy legibility was the caster's whole game and the pips never left the bar.

Implementation notes

Extends the existing vitals-breakdown formatter, a pure text builder with reserved localization space, fed from values the sheet already tracks. Fully testable as a formatter: fixed inputs, exact strings. The one design rule carried over from its siblings: the total is never recomputed in the tooltip, so the breakdown cannot drift from the real number, and any residue prints as its own honest line.

client M (formatter + panel)server noneimpact the economy becomes visibletests formatter fixture suite

S16 · Lighter upkeep for summons

Tier 2Size Menum mirrorplayer ask

census: Summoned Health 1,229 active characters at level 40 or higher

Every magic summon rents a slice of your maximum focus for as long as it lives, a wolf at over a hundred, the big names at two hundred and more, and in the entire game nothing reduces the rent. spainchaud asked in 2017 whether the drain ever goes down; the honest answer then and now is no, and the same complaint was alive again last month. A new stat answers it, written by the Summoned Health innate: up to a quarter off summon upkeep at grandmaster, deepened further by mastery, never gated by it. The summoner's third school finally pays for its own pets' room and board.

Seen in: DAoC's concentration pool, upkeep as reserved capacity, made into something you could build around. Requested by spainchaud (2017).

Implementation notes

One new character stat, appended at the tail of the shared enumeration and synchronized to both halves of the codebase in the same change, which is the entire server footprint: the enum exists there too. The upkeep effect multiplies its reservation by one minus the clamped stat; the innate writes it. Ships in the same enumeration append as S20's stat so the tail moves once.

client M (stat + consumer + data)server enum mirror onlyimpact summon rent finally negotiabletests clamp + floor-at-mastery-zero pins

S17 · Summon training that moves the needle

Tier 2Size Schangesetplayer ask

census: the summon trio 1,229 / 954 / 970 active characters at level 40 or higher

Hemswal The Descended ran the numbers in 2019: grandmastering Train Summoned Pet Damage takes a fire elemental from 37.5 average damage to 46.8, his word for it was pitiful, and the developer on the thread said he would investigate. This month's fixes rewired the line so the stats actually reach every summon correctly; what remains is magnitude, on a line whose training price has been four times normal since it shipped. Proposed plainly: double all three coefficients, with before and after profiles for every summon in the changeset report, and let the line's price finally buy something structural.

Seen in: requested by Hemswal The Descended (2019), seconded by Barugon in the same thread; Perashim's 2018 question about specialization helping summons gets its real answer in S27.

Implementation notes

Three reward-scaling cells in the skill sheet; the creature-side formulas are pinned by this month's tests and deliberately untouched, so the doubling flows through the already-correct wiring. Ships as one changeset with S16 profiled alongside, since the same players feel both.

client S (three cells)server noneimpact the trio starts earning its 4x pricetests existing pins stay green; profile in report

Tier 3 · The ghost shelf comes back — six revivals, each wearing its own old name

S18 · Eidetic Memory: play it again

Tier 3Size LServer: nonerevival

no live data. Dead SkillID 176, no rune, no row, ever

The perfect recall of the practiced mind. A new rune on the dormant identity: replay the glyph you last cast, without holding its card, at half again the focus price. The spent hand becomes, once, a second hand. It is the school's answer to the oldest card-game trick in the book, the spell that comes back, and it makes memory itself the resource the school's name always promised.

Seen in: Magic's flashback and buyback, the discard pile as a second hand at a premium.

Implementation notes

A new effect type, client-only like all of them, plus last-cast tracking at the point where a played card resolves. The replay honors the original's targeting, pays the premium, and touches neither the deck nor the original's cooldown. Version one excludes combos, channels, and reactions, and it cannot replay itself. The real engineering is replaying a rune outside the hand context safely; the exclusion list is the fence while that is proven.

client L (new effect + tracking)server noneimpact the school gains a memory verbtests exclusion table + cost math

S19 · Mental Reflex: draw on instinct

Tier 3Size MServer: nonerevival

no live data. Dead SkillID 178, no rune, no row, ever

The trained mind answers pressure with readiness. A revived innate: when you dodge, parry, block, or take a critical hit, a glyph is drawn into an empty hand slot, on an internal cooldown of a few seconds so it stays a rhythm and never a firehose. Defense refills the hand. It is the school's first passive that plays the deck for you at exactly the moments you cannot.

Seen in: Guild Wars' elemental mantras, incoming pressure converted to resource; the whole reactive-defense genre in one innate.

Implementation notes

The combat log already stamps every one of those defensive moments for the reaction system; this adds a listener at the stamp site and a guarded draw that only fills empty, unlocked slots, so it can never burn a card against a full hand. Deliberately built as an event listener rather than a reaction-gated rune, so the pinned reaction catalog does not change shape.

client M (hook + guarded draw)server noneimpact defense feeds the handtests cooldown + empty-slot-only logic

S20 · Mindfulness: focus in motion

Tier 3Size Menum mirrorrevival

no live data. Dead SkillID 179, no rune, no row, ever

Spellbinder rewards the caster who stands still. Nothing in the school, or the game, rewards the one who keeps moving. Mindfulness, revived as an innate, is the anti-stance: a modest combat focus regen that ticks only while you are in motion, and a second, deeper sprint discount stacking with Tranquility's under the same hard ceiling. Together with S12 it completes a spectrum: stillness pays through the stance, motion pays through Mindfulness, and the school finally answers its oldest recorded request, trainable regen, first asked for in 2014.

Seen in: walking meditation, the oldest mindfulness practice there is; and the graves of DAoC's Serenity and WoW's Spirit, which teach the fix, regen attached to a behavior instead of a percentage. Requested by Satan Himself (2014).

Implementation notes

One new character stat in the same enumeration append as S16's, consumed by the regen tick only while the engine's existing moving check is true. An adjacent piece of archaeology gets settled in the same pass: the engine contains a fully stubbed but never-connected regen-delay mechanic from years ago, and this card's changeset either wires it or removes it, but stops carrying it.

client M (stat + regen branch + data)server enum mirror onlyimpact motion earns regen; the 2014 ask landstests moving and still branch math

S21 · Higher Consciousness: the clarity window

Tier 3Size LServer: nonerevival

no live data. Dead SkillID 171, no rune, no row, ever

The elevated state, held for one deep breath of a fight. A revived active with an elite's silhouette: open the window, and for its duration every glyph you draw arrives without an expiry timer and your glyph costs drop. When the window closes, every held glyph's timer restarts at full, and the sand runs again. A long cooldown makes it the fight's one moment of total clarity rather than a rotation button, and the closing rule is the fence the genre demands: hands that never expire are how deck games die of clog.

Seen in: Guild Wars' elite economy skills, a whole slot for an engine; Slay the Spire's Runic Pyramid, the warning about hands that never let go. Requested in spirit by A'chelata (2017), longer glyph availability.

Implementation notes

A new effect type and a window state on the deck controller: while open, drawn glyphs are stamped with a far deadline; on close, every affected slot restamps to a fresh full lifetime, wearing the same extended highlight the hold uses. The cost discount rides the existing stat machinery. Interaction with Mind Lock is last-writer-wins on the deadline, documented on both tooltips.

client L (new effect + window state)server noneimpact the school gains its elitetests deadline stamp and restamp math

S22 · Consumption: burn a glyph for focus

Tier 3Size MServer: nonerevival

no live data. Dead SkillID 174, no rune, no row, ever

The mind that feeds on its own thoughts. The old identity returns re-purposed as the school's conversion verb: cast it, and your oldest unlocked glyph burns, going to cooldown and refunding focus in proportion to its cost, amplified by mastery. One press, no aiming, no choosing, the target is always the glyph you have been ignoring longest. It is the card-as-fuel archetype, priced for a real-time game, and the refund carries a hard cap per stretch of seconds, because every focus-refund loop in genre history that shipped uncapped got famous for the wrong reason.

Seen in: Balatro, where discards fuel whole builds; Slay the Spire's exhaust, a price you choose to pay for tempo. The health-for-focus version of this idea is deliberately absent: that verb belongs to Death magic and stays there.

Implementation notes

A new effect type and a small controller verb: find the oldest non-locked, non-combined glyph, retire it on its normal cooldown, refund a fraction of its cost. Combos are excluded and say so on the tooltip, since their cooldown rules make them terrible fuel anyway. The refund cap is a tunable, pinned by test.

client M (new effect + verb)server noneimpact the school gains a conversion verbtests refund cap window math

S23 · Summoned Pet Taunt: the pet takes the heat

Tier 3Size MServer: nonerevival

no live data. Dead SkillID 13523, no rune, no row, ever

The newest ghost on the shelf names its own design. A revived active for the summon line: command your summoned creature to taunt, pulling nearby enemies onto it and off you. The summoner's sub-tree today is three percentages; this is its first verb, and it is the verb every summoner in every game reaches for first. The taunt primitive already exists in the creature toolkit; the work is proving the path from a player's command to a pet's provocation runs clean.

Seen in: every pet class worth playing since 1999; in this game's own creature toolkit, where taunt already exists as a primitive.

Implementation notes

A rune on the dormant identity that routes the existing aggro machinery through the player's summoned creature. The verification work is the card: confirming the threat write lands from a pet-directed cast, and documenting the known interaction where a pet returning to leash range clears its threat. No surviving icon for this one; it gets the first genuinely new art in the school in years.

client M (rune + threat-path proof)server noneimpact the summon line gains its verbtests asset shape; behavior by test shard

Tier 4 · Capstones and the owner's desk

S24 · Meditation

Tier 4Size Menum mirrornew skill

new skill. The hidden stillness bonus it builds on is already live in the engine

The Ultima homage, built on what actually made Meditation work there: not the size of the regen but its frictions. Kneel, and focus returns fast, tick by tick. Moving ends it. Playing a glyph ends it. Taking a hit ends it, and while you kneel you are revealed, no stealth holds through a trance. Out of combat it erases downtime; in combat it is a gamble you kneel into with your eyes closed. And here is the secret this card finally tells: the game already pays a hidden stillness bonus, stand still five seconds and your spell crit and glyph costs quietly improve, stacking five times, and no surface in the game has ever mentioned it. Meditation claims that machinery as its own, names it, and at high skill banks one Centered moment: rise from a full trance and your next cast cannot fizzle.

Seen in: UO's Meditation, the trance that revealed you; D&D's psionic focus, the banked coin of calm; the flame and the void, the well of stored power, the whole fiction shelf this school's strings should be written from.

Implementation notes

Version one is mostly data: a channel with a persistent self effect ticking focus restoration, movement-cancel on, casting interrupting it economically, stealth-break on entry. The hidden stillness ladder it surfaces is real, live, and tester-tunable today. Version two adds the one hook the engine lacks, break-on-damage for a character effect, and the Centered no-fizzle bank. One new skill identity, appended to the shared enumeration on both sides of the codebase, plus a sheet row, strings, and art.

client M (rune + v2 hook)server enum mirror onlyimpact the genre's oldest resource skill, done righttests tick math + cancel flags + ladder claim

S25 · Second Guess: trade one glyph

Tier 4Size Menum mirrornew skill

no telemetry yet. New rune; the verb it completes was cast 2,009 times last month

The missing verb between the full wipe and nothing at all: pay a little focus, shuffle one glyph back into the deck, draw its replacement. One right-click inside a short window after the cast. It is the single most Shroud-shaped verb in three decades of card game design, Hearthstone shipped it as Tradeable in 2021 and players never stopped praising it, and it wears a resurrected name: Second Guess was Mind Wipe's own title in the game's earliest releases, before the wipe grew into a whole-hand verb. The school's first verb lends its old name to its newest. Even the icon is ready: an orphaned deck-swap icon has sat unreferenced in the game's files for years, waiting for exactly this skill.

Seen in: Hearthstone's Tradeable (2021); requested in spirit by A'chelata (2017). The name, from this game's own history, release 9 and earlier.

Implementation notes

Reuses S11's armed right-click machinery and replace-one-slot verb wholesale; the difference is the destination, the traded glyph shuffles back into the draw pile rather than retiring to cooldown. Costs focus per trade, cooldown shortened by mastery, and a gently rising per-fight price to keep the machine-gun build honest, reviewed for PvP before ship. One new skill identity on the enumeration tail, one sheet row, strings, and the orphan icon finally referenced.

client M (S, once S11 ships)server enum mirror onlyimpact the third hand verbtests cost and cooldown expressions

S26 · Gear and elixirs that know the school

Tier 4Size Lchangesetowner call, in part

itemization scan: Focus Mastery on zero items; sprint discount on zero; elixir receptiveness on zero

The itemization scan reads like a missing floor: not one item in the game grants Focus Mastery, or the sprint discount, or elixir receptiveness, and the focus-economy stats that do exist on gear live on a handful of artifacts. The buildable half of this card ships the safe lines: masterwork and enchant options for glyph cost, draw speed, and combat regen, on the pattern the game's existing focus masterworks already use. The other half is stated for the owner's desk, both parts of one linked decision, exactly as the specialization register requires: the specialization elixirs write whole points into a stat measured in fractions, a factor fix was tried once and reverted, and gear-granted mastery inverts the training gate in ways still classified as an open design question. No mastery appears on any item until that question is answered. Stated, sourced, and left on the desk.

Seen in: EverQuest's Flowing Thought, worn regen as a real channel; for bards, nearly the only one. Itemization can carry a school's economy when skills alone cannot.

Implementation notes

New item-effect assets on the existing masterwork template, including its durability-cost rider, for the three safe stats; crafting-table wiring; one changeset. The two owner items cite the specialization register's open entries and this page adds no third opinion, only the inventory: zero, zero, zero.

client L (item assets + wiring)server noneimpact the economy reaches the crafting benchtests item-effect stat and value pins

S27 · A deck reason to specialize

Tier 4Size MServer: none

90 percent touch it, 17 percent commit, 3.8 percent learn a verb, 0.22 percent specialize

Today the deck school's specialization buys percentages: a sliver of regen, a shade less fizzle. This card makes it buy a play. A committed specialist's first Mind Wipe of every fight refunds its focus cost and most of its cooldown, scaled by mastery, so the school's signature verb becomes the specialist's opening move instead of the button nobody presses. At mastery zero nothing changes; the refund grows from nothing, the grammar every formula on this page obeys. Two variants sit on the owner's table beside it: a hold that extends stacked glyphs double, and a specialization rider for the summon line, which is the honest answer to Perashim's question from 2018.

Seen in: Guild Wars' elite slots, the most valuable real estate in the game, spent gladly on economy verbs. This card exists to move one number: 89 accounts ever, 29 active today. Variant credit: Perashim (2018).

Implementation notes

A per-fight flag on the deck controller, reset where combat exit already reshuffles the hand, read by the wipe: refund focus scaled by mastery, shorten the booked cooldowns by a mastery fraction. Client-only, continuous scaling, no threshold gates. The refund is once per fight by design, so it feeds openings, not upkeep chores.

client M (flag + refund logic)server noneimpact the spec buys a play, not a percentagetests flag reset + refund math

S28 · The composed school

Tier 4Size S–MServer: none

school today: 258,597 casts, 3.2 percent of everything, 124 of 1,237 casters

The finish line, in one picture. A tree where all six ghosts are alive and every skill wears its own name and its own art. A character sheet that shows the whole focus economy on one panel. A hand that shows its timers, marks its holds, and glows when its windows open. A stance you can walk in, a trance you can kneel into, three hand verbs instead of two, and a specialization that buys an opening play. Nothing in this catalog adds a window, a menu, or a second resource bar; the school's whole upgrade lands on three surfaces that already exist. Focus stops being the tax everyone pays by accident and becomes the school that plays the game the combat was always secretly built on.

Seen in: the genre's one honest success pattern, verbs with timing and prices, passives as the floor, and everything visible.

Implementation notes

An integration card: the tree layout derives itself from the skill sheet, so the revived nodes cost rows and nudges, not layout art; the school reference gets its full rewrite after the dust settles; and the existing skill-identity validation catches any wiring omissions on the new rows.

client S–M (integration)server noneimpact the thesis, shippedtests identity validation + layout smoke

Considered and rejected

• Enemy focus burn or steal. Guild Wars ran this experiment at scale and the verdict was decisive: worthless in PvE, where enemies die before their pool matters, and miserable in PvP, where being denied your resource means not playing the game. The school's identity is self-mastery, not denial.
• A party focus feed. The one resource identity MMO history proves players commit to, EverQuest's Clarity, FFXI's Refresh, City of Heroes' Transference, and the one this page cannot honestly propose: focus adjustments stopped crossing the network years ago, the old pathway sits commented out, and a cross-player grant would need that wire rebuilt and secured. Logged for the owner with the genre's warning attached, mandatory-buff roles can enslave the class that carries them, and left off the cards.
• Health-to-focus conversion. A beloved identity verb, and in this game it is Death magic's identity: the endgame already runs on Tap Soul. Building a rival inside Focus starts a fight Focus loses.
• A hand that never expires. Slay the Spire's Runic Pyramid teaches both halves: the fantasy is real and the clog is the price. Mind Lock is the priced version, and S21's clarity window delivers the dream with an off switch.
• Choose-your-cards menus, scry previews, pick-one-of-three conjures. Turn-based verbs that cost attention a real-time fight does not have, and the conjure additionally cuts against deck authorship, which players here treat as sacred and which the Tactics page owns. Every interaction that shipped in this catalog is a single click or a pre-fight choice.
• Cooldown refund verbs. The cooldown pile's timers are immutable once booked, mutating them is genuine engine surgery with runaway-engine risk, and S27 delivers the payoff, a specialist's refunded opening, without touching the pile.
• Rewarding glyphs for staying held. Three quarters of the hold's volume is already five players pressing it on cooldown for hours; paying held glyphs to mature would reward the chore this page is trying to retire. S10 deepens the hold by making each press worth more, not by making holding a job.

                                  ┌─ FOCUS  312 / 380 ────────────┐
  ┌─ Centered ────────┐        │ regen, in combat   +1.4 / s      │  S15
  │ next cast cannot  │  S24   │ glyph cost         x0.88         │
  │ fizzle            │        │ sprint             -8.8 / s      │  S14
  └───────────────────┘        │ summon upkeep      -2.4 / s      │  S16
                                  │ Aetheric maximum   +140, 38 s   │
                                  └─────────────────────────────────┘
   your spell struck true                        S13 window cue
  ┌──────┐┌──────┐┌──────┐┌──────┐┌──────┐┌───────┐
  │Fire  ││Heal  ││Light ││Death ││Aeth. ││Second │
  │Arrow ││Touch ││ning  ││Touch ││Feedb.││Guess  │
  │█ 12s ││█  6s ││█  2s ││█ 31s ││██ go ││█  9s  │
  └──────┘└──────┘└──────┘└──────┘└──────┘└───────┘
     S4 timers      S10 held, 31s   S13 glowing  S25 trade
  Spellbinder Stance: active, walking pace            S12

flowchart TD
  P1["Phase 1: Trust\nS1-S7 names, prices, timers"] --> P2["Phase 2: Formulas\nS9 S10 S14 S17"]
  P1 --> P3["Phase 3: Small engine\nS11 S12 S13 S15"]
  P3 --> P4["Phase 4: Stats + revivals\nS16 S20, then S18 S19 S21 S22 S23"]
  P2 --> P4
  P3 --> P5["Phase 5: Capstones\nS24 S25 S26 S27 S28"]
  P4 --> P5
  S11x["S11 builds the armed right-click"] -.-> S25x["S25 Second Guess reuses it"]
  OD{{"Owner's desk: S8 parity, S26 elixir pair, S13 sixty-second note"}} -.->|"decisions, not code"| P5