the spell that taxes the cheap curve, and why its threshold is exactly where it is

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 first — the whole play is measured against the curve it establishes.

Subject: death-from-below-recurring-kill, and this is the first play about an ENGINE rather than a finisher. That class is 686 of the catalogue's 766 entries and until today none of it had a turn on it at all.

The question an engine play has to ask is different, and narrower. A finisher is measured by the turn it wins. An engine never wins, so the only honest question is what it does with the turns between coming online and the game ending — and the game ends, unopposed, on turn 5 or 6.

1. The card, and a threshold that is not an accident

UNL-186 Death from Below — Spell, Fury/Chaos, E4 + 1 Power — "Kill a unit at a battlefield. Then, if it had 3 Might or less, do this: You may play this from your trash for [rainbow]."

387.1 recognises "Do this:" as a Reflexive Trigger and 388.1 adds it to the chain as its own Pending Item, while 359.3.d has already put the spell in the trash — so the replay is legal and the card is where it needs to be. After the first cast it is one kill per rainbow Power, for as long as the bodies stay small.

Everything turns on "3 Might or less", so that is the thing to measure rather than to assume. Swept over every deckable unit in cards.json, folded by name and type — 466 units, of which 192 (41%) are Might 3 or less. Cut by Energy cost instead, which is the axis the opening actually runs on:

The chain condition is not a Might threshold, it is a COST threshold wearing one. Ninety-eight per cent of the two-drops and none of the six-drops. And the do-nothing Hold curve is built out of exactly the left-hand column: the pool prints two deckable units at Energy 1 or less and everything else starts at Energy 2, which is why that curve conquers on turn 2 at all.

So the bodies that stop the chain are the bodies that cannot be on the board when the race is decided. A four-drop is a turn-4 play at the earliest on a 2-runes-a-turn curve, and by then the free curve has been scoring for two turns.

And the threshold is the FORMAT's, not this card's

That is a claim about one card until you check the others, so I checked the others. Swept over every printing for a printed Might gate — N Might or less, less than N Might — folded by name and type, the pool has thirteen:

Ten of thirteen sit at exactly three, and the single card that reaches five pays ten Energy and two Power for the privilege. This card is not an outlier; it is the middle of the distribution. The line this format draws its removal at is Might 3, and Might 3 is the top of the Energy-3 curve — so the table in §1 is not a fact about UNL-186, it is a fact about what a Riftbound removal spell is for. Eight of the thirteen are Chaos, four Order, and the outlier at five is Body.

2. The identity is forced, and by one word on the card

UNL-186 is a Signature card (signature: true in cards.json, tagged Pyke), and 103.2.d.2 reads "All of the Signature cards must have the Champion tag that corresponds to the Champion Legend of the deck." Of the five Fury/Chaos legend names in the pool — Loose Cannon, Dark Child, Glorious Executioner, Bloodharbor Ripper, Master of Shadows — exactly one carries the Pyke tag: UNL-185 Bloodharbor Ripper. The entry already says so; I checked it rather than assuming, because a Signature card in uses[] makes that a one-legend field and this project has been caught by it before.

3. The mana, which is the real rate limit

The first cast is E4 + 1 Power. Every replay after it is 1 rainbow Power and nothing else, and 135.2.e.5.a makes a rainbow cost payable by Power of any domain.

• 164.2.b — "Recycle this: [Reaction] — Add [C]" — its cost is the recycle, not an exhaust, so a rune already tapped for Energy still pays.
• 161.2.b — a recycled rune goes back to the Rune Deck, so every kill after the first takes a rune off the board.
• 315.3.b — you channel 2 a turn, capped at 12 by 161.2.a.

So the engine has two speeds and they are different numbers. A burst is up to R kills in one turn, emptying the board of runes. A sustained rate is two kills a turn, forever, which is exactly what 315.3.b refills. Two is also the number of battlefields in a Duel (485.4).

4. What it does with the turns, against the curve it is trying to break

The opponent playing the fastest legal opening: two Energy-2 bodies down by turn 2, both walked in, Conquering both battlefields, then 315.2.b.2 Holds both for 2 points a turn.

They score 2 points in five turns instead of 8, and the whole cost after turn 3 is two Power a turn — exactly what the Channel Phase hands back. The engine does not win; it turns a five-turn clock into no clock at all, for one card and a rune a kill.

And that is what an engine play can say that a finisher play cannot. The catalogue records this line as producing repeatable-removal and nothing else, which is true and tells you nothing about whether it is worth a slot. Priced against the curve it is answering, one card denies the entire free-win line of the format at a rate the rune curve pays for.

5. Breaks to

• [Deflect], and the rules make this one explicit. 355.10.b's own worked example is this exact sentence: "'Kill a unit at a battlefield' targets a unit, but not a battlefield, because the units are targets and 'at a battlefield' is a restriction." It targets, so 809.1.c charges "an amount of Power equal to [Deflect Value] more … for each time they choose". 47 printings carry Deflect, and 187.7 gives it to a Bird token by rule — so a garrison of Birds doubles the price of every kill, and the sustained rate falls from two a turn to one. This is the one keyword that beats the engine on its own axis rather than around it.
• A four-drop. Killing a Might-4 body still kills it, but 387.1's condition fails and the spell stays in the trash with no replay — the chain ends and the next kill is a fresh E4 + 1 Power. Per §1 that costs the opponent two turns of curve to set up, which is the trade the play is about.
• Bodies kept at base. The spell reads "a unit at a battlefield" — at a battlefield is the whole clause. A body at base is untouchable — but a body at base is also not Holding anything, which is the point.
• SFD-105 Ruin Runner-style "can't be chosen", which 054.1 makes absolute where Deflect is only a price.
• Your own clock. Fury/Chaos prints no Energy-1 body, so your Hold curve is turn 6, and this line does nothing to shorten it. You are buying time, not points, and something else has to spend the time.

6. Verdict

As a finisher it does not exist; as an answer it is the best rate in the catalogue for the cost. One card, one rune a kill, and the rune comes back at two a turn — against the exact body sizes the format's fastest opening is obliged to use. [2026-10-24, #456: later than that. A recycled rune goes to the bottom of the Rune Deck (416.1, 416.1.b) and 430.1 channels from the top, so the Channel Phase's two a turn (315.3.b) are the runes above it: the rune is channelled again on T7 at the earliest going first and T6 at the earliest going second with only those two a turn (161.2.a, 485.7); each two more kills push it a turn later, and a card that channels more (430.2) brings it back sooner. Each kill is a rune short on every turn until its rune is back. One rune a kill still stands; what the line above overstated is how soon it is back. The wording above is kept as history.]

The general form is worth more than the line, and it is the same trade every play in this series has written has landed on, arriving this time from the removal side rather than the finisher side:

The cheap bodies that make an opening fast are the cheap bodies that removal is priced to kill, and the bodies that are immune cost four Energy or more — which is to say they are not an opening at all. And the pool prints that trade thirteen times, ten of them at the same number: Might 3, which is 98% of the two-drops and 0% of the six-drops. The one card that reaches Might 5 costs ten Energy and two Power.

Every earlier play here found that a finisher buys its speed with small bodies and pays for it in fragility. This is the same sentence read from the other end of the table: the reason small bodies are fragile is that the pool prices its removal exactly there.