the loop that WANTS a contested board, and what its seven bodies cost

Issue #200, 2026-09-13. Constructed, Duel (485). Rules version 2026-07-16. Card text verbatim from data/corpus_flat.txt; rules pasted from data/Riftbound-Core-Rules-2026-07-16.txt. Read the unopposed clock for the baseline and the catalogue's fastest win for why this subject was chosen over the faster-looking one.

Subject: jhin-fiora-facebreaker-recall, an INFINITE the turn clock reads at T5 against a T6 baseline. The companion play found that of the rows beating their baseline, several carry an unpriced empty-Main-Deck cost — five of nine when it was written that morning, four of twenty-one once the ALT_WIN class was added to the clock later the same day. This is one of the ones that do not: a combat-recall loop that spends zero cards a pass, so its number is honest to begin with. The ratio is perishable and the property is not.

Walked by hand it lands on T4, one turn earlier than the clock says, and the reason is a blind spot in the clock rather than an error in the entry.

1. Why this one is the shape a finisher should have

npm run adversarial -- --stalled classifies every finisher by the board it needs. This line is in the ATTACK bucket, and that bucket's label is the only one that reads as a credential:

"ALIVE WHERE THE CURVE STALLS. The printed text needs the Attacker designation, which 807.1.d and 323.9 make impossible without an enemy garrison. Dead on an empty board, alive on a contested one."

Every pass of this loop is a combat, and 466.1.a.2 recalls the attackers because defenders remain. So the opponent's garrison is not an obstacle to the line — it is the line's fuel. That is the exact inverse of the Renata engine in the companion play, whose payoff needs a battlefield you already control and therefore dies with the curve it exists to rescue.

2. Needs

Domain identity: Jhin and the Dirk and the Amplifier are mono-Fury, Fiora and Facebreaker and the Eye and Renata are mono-Order, the Plaza is Colorless. Union {fury, order} under 103.1.b. None is Signature and data/legality.json names none of them.

3. The loop, and what one pass actually costs

Jhin Standard-Moves base -> the enemy battlefield        144.4.a, cost = his own exhaust (144.2)
  his own trigger:  "When I move, [Add] 1 Energy + 1 rainbow"          +1 Energy, +1 rainbow
  the Eye's effect: "When I move, play a 1 Might Recruit unit token here"   +1 Recruit
  464.2.c.3 designates him Attacker; the Dirk's [Assault 2] makes him Might 6
  709 "becomes Mighty" fires -> Fiora: pay 1 Order Power to ready him       -1 Power
Combat cleanup: 466.1.a.1 heals, 466.1.a.2 recalls the attackers because defenders remain,
  466.3.d is No Result. 458 leaves Jhin as he was, i.e. READY.
Repeat. 144.1.a allows the Standard Move any time in your Main Phase.

Per pass: +1 Energy, +1 Recruit, and Power breaks exactly even — Jhin's rainbow pays Fiora's Order cost, because 135.2.e.5.a makes [A] payable for a Power cost of any Domain. So the loop is an Energy faucet and not a Power faucet, which decides everything below: every Power in this deck comes off the rune curve, and every Power spent takes a rune off the board (161.2.b).

OGN-220 Facebreaker is first-pass-only housekeeping: "Stun a friendly unit and an enemy unit at the same battlefield", so neither Jhin nor the defender deals combat damage, and 466.1.a.1 heals the Recruit's 1 damage at every cleanup anyway.

4. The turns, going first (485.7 gives the extra rune to the player going second)

R = runes on the board after that turn's Channel Phase.

Turn 4. The do-nothing Hold curve in fury/order is turn 6, and this beats it by two.

5. Why the clock says T5, and it is the clock that is wrong

Two blind spots, and the first is the interesting one.

An INFINITE pays for its own remaining deployment, and deployTurn cannot know that. The clock prices every card in a closure against the rune curve — 2 runes a turn, 12 maximum — which is exactly right for a BURST or a CHAIN, where every Energy comes from a rune. It is wrong for an Energy engine: from the moment this loop is live, Energy stops being scarce, and Renata (E4) and the Rage Amplifier (E4) are free. The rune curve still binds their Power, one each, and that is the only thing the later turns are really buying. So for the INFINITE class specifically the clock is an over-estimate, in the opposite direction to the under-estimate the companion play found in the same script. Both corrections live in the same class and they push opposite ways — which is why the honest thing is to walk a line rather than sort the table.

The clock never prices an [Equip] cost. costsOfSet reads a card's printed energy and power, and 818.1 makes Equip a separate Activated Ability with its own cost. This line pays two of them — 1 Fury for the Dirk, 1 Order for the Eye — and the clock sees neither. That is the error this project already records as a standing trap ("Always count the [Equip] costs and the Power: two published '18 Energy' figures had counted neither") arriving inside our own instrument. Here it happens not to move the answer, because the binding constraint on T3 is Power on the board rather than Energy; on a line with three or four Equipment it would.

And one thing the entry has right that is worth stating: SFD-171 Renata Glasc, Industrialist buys a turn, not a capability. Recruits made without her are recalled to base exhausted and are readied for free by 315.1.b at your next Awaken, so a line with a spare turn does not need her at all — it just wins on T5 instead of T4. She is four Energy and a rune for exactly one turn of compression, and that is only worth it because the Energy is free by the time you cast her.

5b. The control case: the other Fury/Order Recruit engine, which the clock scores identically

twilight-reveler-eye-facebreaker-recruits is the same identity, the same payoff and the same [payoff: + ready-recruits-grand-plaza] fold, and the clock reads it at T5 vs T6 — the same row as this one. It is also cheaper and simpler: E9 and no Power over four copies of three names (VEN-020 Twilight Reveler ×2, SFD-153 Eye of the Herald, OGN-220 Facebreaker) against E11 + 1 Power over five names here. Its pass costs literally nothing: Reveler A moves in, the Eye makes a Recruit, 383.4.e fires "When I attack, ready another friendly unit" on Reveler B, the combat cleanup recalls A, and B goes next — so a Recruit arrives on every pass the Eye-carrying Reveler makes.

And it cannot do what §4 does, for one reason: it produces token-body-engine and nothing else. This line produces infinite-energy as well, and that is the whole difference.

T1  R=2   Eye of the Herald (E1)
T2  R=4   Reveler A (E3); [Equip] the Eye to A, 1 Order Power -> board 3
T3  R=5   Reveler B (E3); Facebreaker (E2). Loop all turn: Recruits pile up at base EXHAUSTED
T4  R=7   Awaken readies them. Rage Amplifier (E4 + 1 Power) -> board 6. Walk seven in -> Conquer
          Renata (E4) does NOT fit beside it: E4 + E4 = 8 against 7 runes
T5        Beginning Phase: Hold -> 7+ units here -> win

Turn 5, and T5 is its floor — pulling the Conquer back to T3 would need Renata on T3 so the Recruits enter ready, and Reveler B at E3 plus Renata at E4 is seven Energy against five runes.

So two entries the clock cannot separate are a turn apart, and the separator is whether the loop makes Energy. A body engine has to buy its own insurance off the rune curve; an Energy engine buys it out of the loop. That is the §5 finding stated from the other side, and it is also why these two are not one line written twice: Jhin costs two more Energy, one more Power and two more card names, and what he buys with them is the turn.

6. Breaks to — the opponent's one turn, priced

The seven bodies have to survive from your T3 Conquer to your T4 Beginning Phase. That is the whole exposure, and it is where this line is actually decided.

• OGN-133 Flurry of Blades — Body, E1, [Reaction] — "Deal 1 to all units at battlefields." One Energy answers seven Might-1 Recruits, because 143.2.a kills on marked damage at or above Might and 370.1.a.2 makes them die simultaneously. This is why §4 spends E4 + 1 Power on the Rage Amplifier before passing the turn, and it is the single most important line of the play.
• The recorded fix is Mind and is ILLEGAL here. ready-recruits-grand-plaza names UNL-077 Soul Shepherd ("Your token units have +1 Might") and correctly says he "narrows an all-Order line to Mind/Order". It does not say what a Fury/Order engine does, and Mind plus Fury plus Order is three domains, which 103.1.b forbids outright. Swept for a static, non-"this turn" +N Might grant to your units with domains inside {fury, order} or colourless, folded by name and type, the identity's own answers are three and a half: - VEN-018 Rage Amplifier — Fury, gear, E4 + 1 Power: "Your units have +1 Might. If I'm [Empowered], they have +2 Might instead." No location clause, no body slot, and 359.2.d enters it ready at base. This is the one, and it is already in the catalogue on two other Plaza entries (desert-call-vi-sand-soldier-plaza, ferrous-forerunner-karthus-mech-plaza) — nobody had connected it to the two Fury/Order Recruit engines. - OGS-013 Garen, Commander and OGN-243 Darius, Executioner — Order, E6 + 1 Power, M5 and M6: "Other friendly units have +1 Might here." The here is load-bearing and a first draft of this play missed it — they must stand at the Plaza themselves, which costs a body slot and also fills one of the seven. - OGN-294 Trifarian War Camp and UNL-T03 Brush are battlefields and are therefore mutually exclusive with the Plaza: 485.4.a has each player provide three battlefields of which only one is used, and 103.4.c forbids duplicate names.
• Above Might 2 the answers get expensive, and 22 of the 23 Plaza entries name none of them. Swept with /deal \d+ to (all|each) units?|kill all units/i over text and effect, folded by name+type, the pool prints seven mass answers and only four reach a Might-2 body at a battlefield you control: OGS-018 Tibbers (Fury/Chaos, E8 + 2 Power, deal 3 on play), OGN-190 Kog'Maw, Caustic (Chaos, E3 + 1 Power, deal 4 — but its Deathknell is scoped to "my battlefield", so it has to be walked into the Plaza first, which contests it), OGN-123 Unchecked Power (Mind, E7 + 2 Power, deal 12) and UNL-180 The Ruination (Order, E9 + 3 Power, kill all units). UNL-212 Frozen Fortress is a battlefield and only reaches units at itself. Nineteen of the 23 Plaza entries named OGN-133 when this was written on 2026-09-13 and exactly one named anything larger — the family had been measured against a single one-Energy card. It is 20 of 23 within the day, because the finding was applied to the catalogue the same day; the three that still do not are dragonstorm-confront-grand-plaza, leblanc-bashful-bloom-trevor-plaza and desert-call-vi-sand-soldier-plaza. The count is perishable and the point is not: one Energy was the only threat the family had ever been priced against.
• VEN-133 Glowstone is the cheapest hard answer in the pool and it misses by a phase. Order, E2: "Disempower this, exhaust: Choose a player. They gain control of this and recall it… At the end of your turn, kill this and deal 5 to all units you control." Five damage kills a Might-2 or Might-4 Recruit board outright, and 381 makes them hand it over on their turn — so it fires at the end of your next turn, and 315.2.b.2 has already won the game in that turn's Beginning Phase. Against this line it is a turn too slow; against a Plaza line that needs a second turn it is lethal for two Energy, and it is in no entry.
• Contesting the Plaza instead of sweeping it. They move a body in; 190.3.a.1 applies Contested and 323.9 stages a Combat. Seven Might-2 Recruits are 14 summed Might (465.2.c) and 466.5 leaves Control with the player who already had it, so this is the worst of their options unless they can win the combat outright.
• Killing Fiora, which the entry already names. She has no protection where Jhin has [Deflect], and every pass opens a Showdown in which the opponent holds priority. Killing her does not undo the Recruits already at the Plaza — it stops the engine, not the payoff.
• No enemy garrison at all. 323.9 stages no Combat, so Jhin walks into an empty battlefield, takes it by 344.2 with no Attacker designation, and the loop never starts. That is the credential from §1 read from the other side, and it is the one board on which this deck has no plan.

7. Verdict, in three cases

Fighting for the Plaza, which is the third case worked

The swarm at base is Might 2 under the Rage Amplifier and unbounded in size. Walk K of them into the Plaza the opponent is holding with a garrison of summed Might D:

• 465.2.c: "Starting with the Attacker, each player assigns an amount of damage equal to their summed Might among the other's Units." You assign 2K; they assign D.
• 465.2.c.4 caps each assignment at minimum lethal and 465.2.c.3 forces it onto one unit in full before moving on, so their D kills floor(D/2) of your bodies and no more.
• The binding constraint is that you must WIPE the garrison, not merely out-damage it. 466.3.a gives the combat to the player who is "the only Player that has units remaining at this battlefield", and if even one defender lives, 466.1.a.2 recalls every attacker you have — no Control, no Conquer, no Plaza. With Might-2 bodies that is 2K ≥ the garrison's total lethal requirement, which unbounded K clears against anything.
• Win it and 466.5 establishes Control, 466.5.d makes it a Conquer, and the survivors — K − floor(D/2) — are standing on the Plaza for next turn's Hold. Overshoot K and seven is not the constraint either.

So this is the one case in which the loop and the payoff want OPPOSITE combat results, and that is the thing to get right at the table. Every pass of the engine deliberately LOSES its combat — it needs defenders to remain so 466.1.a.2 sends Jhin home to go again. The Plaza attack must WIN its combat, because a surviving defender recalls the swarm. They are two different battlefields on the same turn and the order does not matter, but running the engine into the Plaza would be a mistake: it would kill the garrison you are farming.

SFD-177 Azir, Sovereign is not needed for any of this and the third case does not want him. His clause moves token units "to this battlefield" — the one he is attacking — and 420.3.a puts the exhaust on the Standard Move alone, so what he uniquely does is move exhausted tokens battlefield to battlefield with no [Ganking]. This swarm is neither: 466.1.a.2 already put it at your base and 315.1.b already readied it, so 144.4.a walks it in for free. He is a four-Energy body here.

The finisher earns its slot in the middle case and the middle case is the normal one. That is more than the companion play's Renata engine could say, and the difference is one property: this line's payoff is gated on a battlefield you control, but its engine is gated on a battlefield they control, so a stall feeds it rather than starving it.