Temur Roar precon upgrade guide

Eshki, Temur's Roar never reads the word Dragon — she checks power as you cast, and the box obliges: 25 of its 34 creature spells draw a card on the way to the stack, and seven of those also burn every opponent. The mana is what lags. Six lands always enter tapped, and turn-three double red — which 19 spells want — shows up in 78.5% of hands. Eleven swaps, about $83, buy speed and make the casting itself the win condition.

Nothing on this page is eyeballed: the shipped Tarkir: Dragonstorm decklist was resolved card by card for land behaviour, pip odds, power counts and the bracket floor, and every figure about the deck after the swaps comes from running the same checks on the post-swap hundred.

What are the best Temur Roar upgrades?

Ranked by how much each one moves this deck. The cuts free the slots.

1. Terror of the Peaks ($28.60) — Eshki pays you when a big creature is cast; this pays you when one enters, which catches everything she misses: Miirym's copies, Lathliss's tokens, every Dragon Ureni cheats into play.
2. Sarkhan's Unsealing ($1.12) — 28 of the upgraded deck's 34 creature spells sit in its power 4–6 window, so nearly every cast now throws 4 damage somewhere useful. Ureni and Dragonlord Atarka hit the second mode and sweep the opposing board for 4.
3. Miirym, Sentinel Wyrm ($6.61) — power 6, so casting it draws a card and burns the table, and every nontoken Dragon after it arrives twice.
4. Goreclaw, Terror of Qal Sisma ($4.92) — takes {2} off 29 of the deck's creature spells and replaces itself on the way in.
5. Surrak Dragonclaw ($2.63) — six power at flash speed, so Eshki's burn can happen on someone else's turn, and nothing you cast afterward can be countered.
6. Scourge of Valkas ($3.62) — a second Dragon Tempest on a 4/4 body that scales with the Dragon count.
7. Nature's Lore ($2.12) and Three Visits ($4.63) — green ramp that fixes red, because Cinder Glade and Sheltered Thicket both carry the Forest type these spells fetch.

The last three slots are lands: Spire Garden ($6.56), Training Center ($17.70) and Rejuvenating Springs ($4.86) enter untapped whenever you have two or more opponents — which is every game of Commander — and replace three lands that cost the deck a turn. Every pick is argued below.

Eshki counts power, not Dragons

The trigger has three rungs. Any creature spell grows Eshki by a +1/+1 counter; power 4 or greater draws a card; power 6 or greater deals her own power — new counter included — to each opponent. A 2/2 for three mana looks modest until you count what feeds her: 25 of the box's 34 creature spells clear the draw line, and 7 clear the burn line. She reads the spell, not the battlefield, so the card is drawn even when the Dragon itself meets a counterspell.

The same wording sets her limits, and the box quietly plays against them in places. Only casting counts. Ureni of the Unwritten pulls Dragons from the top eight straight onto the battlefield — free material, zero triggers. Token copies from Lathliss, Dragon Queen and Nesting Dragon skip her. And she checks power on the stack: Deceptive Frostkite is cast as a 1/1 that only becomes a copy of something large as it enters, so the deck's cheapest Dragon never draws its card. None of that makes those cards bad. It means the deck's engine runs on what leaves your hand, and the upgrade below leans the list toward casts that pay.

Thundermane Dragon deserves a sentence here: it lets you cast power-4-or-greater creatures off the top of your library with haste, and a cast from the library is still a cast. With Eshki out, the top of your deck becomes a second hand that draws cards as you play it.

The draws are in the box; the turns are not

Stacked on the commander, Temur Ascendancy and Elemental Bond draw on big creatures entering, Become the Avalanche draws one per power-4 creature at once, and Zenith Festival's harmonize cost shrinks by the power of a creature you tap — a late Festival off a 6/6 exiles most of a turn's worth of cards for {R}{R}. For a precon, that's an unusually complete engine.

Getting to it is the problem. The average mana value is 4.11 and the curve holds 25 spells at five mana or more, yet six of the 37 lands always enter tapped, and the ramp count sits at 7 against the 8–12 typical for this site's audited decks. Red is the pinch: 19 spells want {R}{R} by turn three and the 28 sources deliver that in 78.5% of hands at the 90%-confidence standard this site's mana tools use. Blue is thin too — 21 sources for a colour with 13 spells — though only two cards in the whole deck ever ask for {U}{U}.

Interaction is five instants: Chaos Warp, Beast Within, Rapid Hybridization, Reality Shift and Spit Flame, with Storm's Wrath and Blasphemous Act as the resets. That suite stays; the additions attack the draw-and-burn plan and the mana instead.

What I'd cut

Path of Ancestry is the surprise, because it's an auto-keep in nearly every typal precon. Its scry fires when you cast a creature sharing a type with your commander — and Eshki is a Human Warrior in a deck where 29 of 34 creature spells are Dragons. In the box, three casts in the 99 ever trigger it — two Humans and a changeling — so the land is blank text on 31 of the deck's 34 creature spells. An always-tapped land that scrys almost never loses to any untapped dual, and it loses on this page to a land that makes the same colours a turn sooner.

Taurean Mauler is a 2/2 on the stack, so Eshki's check reads 2 and whiffs; the changeling growth arrives later, on opponents' turns, after the moment that mattered. Dragonmaster Outcast misses twice — power 1 when cast, and its 5/5 tokens enter uncast — and it wants six lands before doing anything at all. Rapacious Dragon spends five mana on a 3/3 that skips the draw; two Treasures don't buy the turn back. Vengeful Ancestor also sits at power 3, and goading one creature a turn with no other goad support mostly just picks a fight the 3/4 body can't finish.

Dragonlord's Servant shaves {1} off Dragon spells. Goreclaw, Terror of Qal Sisma shaves {2} off every power-4 creature spell, draws a card off Eshki as it arrives, and swings for 4. The slot upgrades itself.

Reflections of Littjara copies the creature spell, and a copy is put on the stack without being cast — so the doubled Dragon never triggers Eshki, and the enchantment is five mana that does nothing the turn it lands. It also happens to be the only card in the box that asks you to name a creature type; that requirement leaves with it. Breaching Dragonstorm flips one random nonland card per five-mana cast, and this deck's average flip doesn't justify the rebuy loop. Selvala's Stampede lets the table vote on your six-mana sorcery, and three opponents reliably pick whichever mode hurts you least.

The other two land cuts: Frontier Bivouac taps for everything and always costs a turn, which is the trade the bond lands below refuse to make. Evolving Wilds fetches a basic that enters tapped — a full turn slower than the dual replacing it, in the deck whose stated problem is tapped mana.

What I'd add

The damage package first. Terror of the Peaks converts every creature entry into its power in damage, aimed anywhere. Read that against Eshki's blind spots: Miirym's token copies, Lathliss's 5/5s, anything Ureni cheats in — entries she ignores all pay through the Terror of the Peaks trigger instead, and the two engines together cover every way a creature reaches your battlefield. At $28.60 it's a third of the entire bill and still the first card on the list; nothing else in the deck's price range does this job. Sarkhan's Unsealing does the budget version of the same trick from the cast side: 4 damage per power-4-to-6 creature spell — 28 of the deck's 34 — and a 4-damage sweep of an opponent's board when Ureni of the Unwritten or Dragonlord Atarka comes down. For a deck whose instant-speed interaction is five cards, removal earned by casting Dragons is the cheapest fix available, and at $1.12 it is the best dollar this page spends.

Miirym, Sentinel Wyrm is the deck's missing six-drop: cast it for a card and a table-wide burn, then every later nontoken Dragon arrives twice. The copy is a token, so Eshki ignores it — but Terror of the Peaks, Dragon Tempest and Scourge of Valkas all pay on entry, and Sarkhan, Soul Aflame can become it for a turn. Surrak Dragonclaw puts six power on the stack at flash speed, which turns Eshki's burn into an instant, and his two static lines — your creature spells can't be countered, everything tramples — solve problems this deck otherwise concedes to. Scourge of Valkas is Dragon Tempest again on a body, dealing damage per Dragon you control on every Dragon entry; the {R}{R}{R} cost is real in this mana base, and it's a mid-game play on purpose.

Goreclaw, Terror of Qal Sisma is the ramp piece the count says is missing, worn as a creature: {2} off 29 of the deck's creature spells, a card drawn on arrival, trample handed to the board when it attacks. Nature's Lore and Three Visits are the same spell twice, and both can fetch Cinder Glade or Sheltered Thicket — Mountain-typed Forests — so the green ramp doubles as red fixing. That matters more here than in most decks; the pip section below says why.

The three lands are the Temur bond cycle: Spire Garden, Training Center and Rejuvenating Springs, untapped in any game with two or more opponents. Training Center is the expensive one at $17.70 and stays anyway — it's the only add that feeds both blue, the deck's thinnest colour at 20 sources, and red, its tightest double.

Two cards were priced out rather than argued out. Old Gnawbone ($51.16) turns every point of Dragon combat damage into a Treasure and would fund the whole top end by itself; Ancient Copper Dragon ($120.34) does it faster off a d20. Each costs more than half the entire add list, so they're next purchases, not swaps. Kindred Discovery ($5.87) was argued out instead: naming Dragon makes it a second draw engine, but the deck already draws on the same event three ways, and {U}{U} is the hardest cast in the 99. Garruk's Uprising and Rishkar's Expertise fell to the same crowding — the draw seat is taken.

The cards that look like cuts and aren't

Blasphemous Act is a nine-mana sorcery in a deck that wants its casts to be creatures, and it stays anyway: with three opponents' boards in play it usually costs one or two mana, and it is the deck's only true reset button next to Storm's Wrath. You cast it from behind, where losing your own two Dragons is the good trade.

Nogi, Draco-Zealot and Sarkhan, Soul Aflame are the off-type legends a Dragon purist eyes first, and both stay. Each discounts every one of the deck's thirty Dragon spells by {1} — a discount Eshki converts into earlier draws — and each stops being off-type on demand: Nogi becomes a 5/5 flying Dragon when three Dragons attack, and Sarkhan, Soul Aflame copies whichever Dragon just entered, Miirym's 6/6 tokens included.

Mosswort Bridge always enters tapped and keeps its slot because its condition is trivial here — 30 of 34 creature spells have power 4 or greater, so total power 10 is a low bar — and the hideaway exile can hand back Blasphemous Act for {G}. Sheltered Thicket stays tapped too: it's one of the two lands Nature's Lore, Three Visits and Farseek can fetch, and it cycles when the 37th land is one too many. Temple of Abandon and Temple of Mystery survive where Path of Ancestry didn't because their scry is unconditional, and each feeds a colour the checks call short.

Same curve, sharper casts

The swaps don't slim the deck down — average mana value actually edges from 4.11 to 4.15 — they make the weight pay. Creature spells that draw off Eshki go from 25 of 34 to 30 of 34; casts that burn the table go from 7 to 9; and with Sarkhan's Unsealing out, 30 of 34 creature casts deal damage somewhere. Always-tapped lands fall from six to four, with the three bond lands untapped in any multiplayer game. The counted ramp stays at 7 before and after, but the seven now includes two spells that fetch red-producing lands, and Goreclaw's discount never shows up in a ramp count at all.

One number this page refuses to oversell: double red. The turn-three {R}{R} check reads 78.5% stock and 76.4% after the swaps — the cut lands made red from the bench (tapped, or a turn late through Evolving Wilds), so the source tally dips by one even as the mana gets faster in practice. No eleven-card swap fixes a 20-spell double-red demand; a rebuilt mana base could, and would roughly double the bill. The playable answer is sequencing: lead with the red duals, point Farseek and both Forest fetches at Cinder Glade or Sheltered Thicket, and treat the {U}{U} casts — there are exactly two — as late-game luxuries.

The bracket floor is 1 on both sides of the swap, with zero Game Changers in either list. The box's singles run $81.12 and the upgraded hundred $157.67, both August 2026 market prices for default printings.

The upgraded decklist

The full 100 after the eleven swaps. Hover any name for the card and its current price.

Load into the deck builder Copy decklist

“Load into the deck builder” opens this list in your account, pinned to the printings the box shipped, ready to edit or save. “Copy decklist” gives you plain text for Moxfield, Archidekt or Arena — no account needed.

FAQ

What does Temur Roar do well without any changes?

The draw engine ships working: 25 of the box's 34 creature spells have power 4 or greater, so each one draws a card off Eshki, Temur's Roar as it's cast — and seven hit power 6 and burn every opponent too. Temur Ascendancy and Dragon Tempest hand the Dragons haste, and Thundermane Dragon casts big creatures straight off the top of the library. The weak spots are speed and mana: six lands always enter tapped, and turn-three double red arrives in 78.5% of hands against 19 spells that want it.

How much do the eleven Temur Roar adds cost?

About $83 at August 2026 market prices for default printings. Terror of the Peaks ($28.60) and Training Center ($17.70) are more than half of that; the other nine adds are $6.61 or less each. The box's own singles total about $81, so the upgrade roughly matches the deck's price without replacing its identity.

Does Eshki trigger when a Dragon enters without being cast?

No. Eshki, Temur's Roar triggers only on creature spells you cast, and she checks the spell's power on the stack. Dragons that Ureni of the Unwritten puts onto the battlefield, token copies from Miirym, Sentinel Wyrm or Lathliss, Dragon Queen, and spell copies from Reflections of Littjara are never cast, so none of them draw you a card. Deceptive Frostkite is cast as a 1/1, so even though it enters as a copy of something large, it misses the draw as well.

Could Ureni of the Unwritten lead this deck instead?

Ureni is the box's designated second commander — a seven-mana 7/7 Spirit Dragon in the same green-blue-red identity that puts a Dragon from your top eight cards onto the battlefield whenever it enters or attacks. It works as a top-end commander, but the 99 loses its card engine: without Eshki checking every cast, the deck keeps only Temur Ascendancy, Elemental Bond and a handful of one-shot draw spells.

Does upgrading Temur Roar change its bracket?

No — the bracket floor is 1 before and after the swaps, and neither list contains a Game Changer. The upgrade adds consistency and damage rather than tutors or fast mana, so the deck still belongs in the same precon-level pods it shipped for, just with fewer dead turns.