the board that cannot be improved, and the three turns of mana you cannot spend on it

Issue #200, 2026-09-14. 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 one body carries four points for the companion case where a turn table refuted its own entry.

Subject: shen-disciple-hand-hammer-exactly-two, calm/order, from the seven-card Vendetta cycle that all reads one board shape — a battlefield where you control exactly two units.

Every other line in this catalogue scales by adding copies. This one cannot. A third friendly body switches off every clause in the cycle at once, so 103.2.b's three copies are unusable and what you stack is clauses on a fixed board. A turn table is the only way to show what that costs, and what it costs is three turns of mana you are not allowed to spend where it would help.

1. Needs

2. The baseline, checked properly this time

The identity's do-nothing Hold curve is T5. VEN-043 Steel Paws is Calm, Energy 1, and is not [Unique], not Signature and not in data/legality.json — so 103.2.b's three copies mean two of them cost 2 Energy and land on turn 1's two runes. Two bodies, two Conquers on T2, eight by T5.

The curve needs two BODIES, not two CARDS, and copies of one card satisfy it. I got that wrong in the companion play and it cost that play its headline; it is checked here before anything is claimed.

Hold one battlefield instead of two and the same curve pays 1 a turn and arrives on T9. Those are the two numbers this line is measured against.

3. The turns, going first — 485.7 gives the extra rune to the player going second

Against an opponent who takes and garrisons the other battlefield. R = runes after that turn's Channel Phase; a rune pays 1 Energy by exhausting (164.2.a) and can also be recycled for 1 Power of its domain (164.2.b, whose cost is the recycle), the recycled rune leaving for the Rune Deck (161.2.b) and returning at 2 a turn (315.3.b). Idle Energy is written down.

       R    spend                                          idle    score
T1     2    Disciple of Shen (E2)                           E0       0
T2     4    move to battlefield A -> Conquer                E0      +1 = 1
            Hand Hammer (E2), enters ready at base
T3     6    HOLD A: Score +1                                E0      +1 = 2
            Shen (E6 + 2 Order Power)          (R -> 4)
T4     6    HOLD A: Score +1                                E6      +1 = 3
            move Shen to A  -> the board is now COMPLETE
            [Equip] Hand Hammer to Shen (1 Calm) (R -> 5)
T5     7    HOLD A: Score +1, Shen +1                       E7      +2 = 5
T6     9    HOLD A: Score +1, Shen +1                       E9      +2 = 7
T7    11    HOLD A: Score +1, Shen +1                       E11     +2 = 9   WIN

T7. Two turns slower than doing nothing unopposed, two turns faster than the contested curve — and from T4 the deck idles six, seven, nine and eleven Energy while its rate never moves.

4. The board is complete on T4 and cannot be improved after it

This is the sentence an entry cannot say. From T4 the battlefield is finished: Shen and the Disciple, and every clause in the cycle reading "exactly one other unit you control" is satisfied. Any third body you play there switches off all three at once — Shen's point, the Hand Hammer's +2, and the Disciple's own [Shield 3].

So the idle Energy is real and it is not laziness. 355.2.a lets you play a unit to "the controller's Base or a battlefield the controller controls", so the mana CAN buy bodies — just never at A. On the turn you draw your third body, the correct play is to hold it or send it somewhere else, and that is a genuinely unusual instruction to give a deck with nine spare Energy.

What the mana should buy, and it follows from the exactness rather than from taste:

• the second battlefield. A spare body at B is another Hold, +1 a turn, and it is the only way this deck's rate ever increases. It is also the only use of the idle Energy that changes the turn count.
• a replacement body IN HAND. The failure below is losing one of the two, and the repair is a body you can play immediately to restore the count — which 355.2.a permits directly at A.
• not a bigger garrison at A. Ever.

5. Both walls are one card away, and they are opposite walls

Too many switches it off: a third friendly body at A.

Too few switches it off just as hard. The Disciple is printed Might 1, and 143.2.a kills on "nonzero damage marked on it equalling or exceeding its Might" — so OGN-133 Flurry of Blades (Body, E1, [Reaction], "Deal 1 to all units at battlefields") removes it for one Energy at Reaction speed on either player's turn. Shen is then alone at A, his clause finds no other unit, and the rate drops from 2 a turn back to 1 — which is the do-nothing curve. One Energy undoes three cards.

And the Disciple's own [Shield 3] does not save it: 814.1.c makes [Shield] short for "While I am a defender, I have +X [M]", and 814.1.d makes a defender a unit that "has gained the Defender designation during Combat". Outside a combat it is Might 1 and nothing else.

The window is exactly two and both edges are a single card. That is the price of stacking clauses instead of copies, and no entry field can carry it.

6. The trap inside the cycle

VEN-129 Sacred Protector (Order, E4 + 1 Power, M6) reads "I don't deal combat damage unless I'm at a battlefield with exactly one other unit you control." It is in the cycle and it must not be in this deck's battlefield: its exactness is a penalty it escapes, not a bonus it earns, so adding it as a third body at A switches off Shen, the Hammer and the Disciple in order to switch itself on.

VEN-119 Keeper of Law is the softer version — its reward is a cost reduction paid once as you play it, so it is a discount on joining the shape rather than a reason to hold it.

Read the SIGN of an exactness before adding the card. Five of the seven are paid for holding the shape; two are paid for escaping it.

7. What the instruments cannot see, and here it is everything

Neither of this project's two clocks prices this line at all. scripts/adversarial-check.mjs:194 sets FINISHER = new Set(["INFINITE","BURST","CHAIN","ALT_WIN"]) and line 1182 skips everything else, so the turn clock and the --stalled classifier are both blind to it — and to 688 of the catalogue's entries, which is the ENGINE class.

--engines exists and reports a histogram rather than a row: 688 engines, median T5, mean T4.74, with 154 of them deployable only after the do-nothing curve has already won in their identity. There is no number for this entry to check a hand walk against, which is why §3 is a hand walk and why the format exists.

And what a deployment clock would say is the wrong thing anyway: this line's deployment finishes on T4 and it wins on T7. The three turns in between are the shape of the deck, not a gap in it.

8. Verdict

It is at its best on the one board where the free curve is slowest, which is the right shape, and it is two turns worse than doing nothing on the board where doing nothing was already enough. Two points a turn from two bodies at one battlefield, against a curve that needs both battlefields to pay the same rate.

But its ceiling is fixed on turn four and it has no second gear. Everything this catalogue normally does to raise a rate — a second copy, a multiplier, a bigger garrison — is forbidden here by the very clause that pays. sacred-protector-disciple-of-shen-pair says of the cycle's defensive half that "nothing here scores"; the honest verdict on the scoring half is that something here scores, exactly twice a turn, forever, and nothing you draw afterwards can make it three.

9. Not verified

I did not walk the opponent's turn in detail. §5 prices the cheapest answer and names it, but I have not checked whether a calm/order deck can profitably hold up a [Reaction] answer to OGN-133 through its own turn, which is what would decide how often the wall at "too few" is actually reached. Read §5 as the cost of the removal LANDING, not as a claim about how likely it is.