[ Robinhood Chain · 4663 · pre-launch ]

Launch a coin.
Get a game.

Trading fees pay an AI to build your coin's game in public. Once a week, two coins battle for an hour.

Play the sample Watch a battle Docs

games shipped — pre-launch

templates on the shelf

0.7%

of every trade to the coin's vault

Sample game · Runner template · level 1 · free

● playable

score 0best 0

Tap to play

space / ↑ / tap to jump

Built from the shelf, not by a coin yet. Every real coin gets its own, themed by its launcher, in about 15 minutes.

[ 01 · the battle ]

Once a week, two coins fight for an hour.

Two agents build side by side on one page. Both logs stream. A meter shows each coin's net inflow. At minute 60 both games go live, and on Monday the losing war chest buys and burns the winner.

Build battle #01 · theme: rooftop chase

SAT · 60:00 → 00:00

$ROOF Rooftop Run

waiting

$SKY Skyline Dash

waiting

Net inflow · ETH bought − ETH sold, this hour

1.84 ETH $ROOF

0.97 ETH $SKY

[ Sample battle — the first real one goes live in Phase 2 ]

MON

Theme vote

Matchstick holders pick this week's theme from five

WED

Matchmaking

two eligible coins within 2× market cap are paired

FRI

Brief lock

both launchers lock the build brief for the theme

SAT

The battle

60 minutes, two agents build side by side, live

SUN

Hold-back

24 h — sells by battle buyers are subtracted

MON

Settlement

the losing war chest buys and burns the winner

eligibility: ≥ 50 holderswar chest ≥ 0.02 ETHone battle per coin every two weekscap 0.5 ETH per battle

[ 02 · how a coin gets a game ]

One sentence in. A playable game in about fifteen minutes.

01
One sentence
A launcher types what the game is. “A runner on city rooftops, dodging drones.”
02
Brief
The AI turns it into a spec: template, theme, three starter levels, which level is holders-first.
03
Launch
The coin goes live on the venue. The MatchVault contract is deployer and fee recipient; the launcher signs from their own wallet.
04
Starter build
About 15 minutes. The agent builds version 1 in an isolated sandbox while the log streams on the coin page.
05
Play
Live at <coin>.playmatchstick.com. Free, instant, no wallet. Level 2 opens for holders first.
06
Fees
Every trade pays 1%. 70% of it lands in the vault and is split on the next run().
07
Iterate
Holders queue features, the agent ships updates from the build budget, and the coin becomes eligible for the weekly battle.

The template shelf · v1

Runner
dodge, jump, collect
lanes + hazards
Arena
survive waves, dash
wave scripts
Tower defense
place, upgrade, hold
paths + waves
Iso builder
gather, build, defend
tile maps

Each template ships with desktop and touch controls, a score screen, a share card and a holder-gate hook. The agent changes theme, rules, levels and art; the engine underneath stays tested. That is what keeps quality high at $10–30 of compute per update.

Gates · all must pass before anything deploys

01Compiles cleanly
02Smoke test: the game boots, the player can move, a level can be finished
03Screenshots at desktop and phone sizes
04Performance budget: frame rate and bundle size
05Reviewer model: the build matches the spec and the content policy

Isolated sandbox · no network · licensed CC0 asset kit · deploy to <coin>.playmatchstick.com

[ 03 · the vault ]

Every fee lands in a contract nobody owns.

Every trade pays 1%. 70% of it lands in the coin's MatchVault, and anyone may call run() to split it 45 / 20 / 20 / 15. The split is fixed at deploy: no owner, no upgrade path, no admin withdrawal. Type a weekly volume and watch the real formula run.

Weekly trading volume on one coin, USD

Trade fee (1%): $500
To the venue (30%): $150
To the coin's vault (70%): $350

One run() of the MatchVault

Build budget 45%$157.5
War chest 20%$70
Matchstick buy and burn 20%$70
Launcher 15%$52.5

≈ 7 shipped updates at $10–30 of compute each. An update every ~$6,500 of volume. War chest: what this coin brings to its next battle.

A model of the contract's fixed split, not a live number. The caller of run() takes 0.5% of the build share as a tip.

Worked example · per $10,000 of volume · a model, not a live number

Trade fee (1%)$100
To the venue (30%)$30
To the coin's vault (70%)$70
Build budget (45%)$31.50
War chest (20%)$14
Matchstick buy and burn (20%)$14
Launcher (15%)$10.50

Idle build budget — no build paid for 14 days — buys and burns the coin itself. Money either builds or burns.

Live counters · honest zeros

Games shippedpre-launch — the first four are house games in Phase 10
Battles settledthe first two are demo battles in Phase 20
Matchstick burnedhonest zero — nothing to burn yet0 ETH
Tokenit launches after both demo battles settlenot launched

[ 04 · scoring ]

Net inflow, not volume.

Volume can be washed. What stays cannot. The score is ETH bought minus ETH sold during the hour, minus anything the battle's buyers sell within 24 hours. Try the presets.

ETH bought during the 60 minutesETHETH sold during the 60 minutesETHETH sold within 24 h by wallets that bought in the battleETH

score = 3.2 ETH − 0.4 ETH − 0.2 ETH

2.6 ETHcounts — support that stayed

On top of this: one wallet counts for at most 0.5 ETH · buys under 0.002 ETH are ignored · ties go to more unique buying wallets.

Net inflow, not volume

buying and selling the same amount nets to zero, and each round trip costs 1% in fees

24 h hold-back

a last-minute buy that is sold the next morning does not count

Per-wallet cap

one wallet counts for at most 0.5 ETH of score

Dust floor

buys under 0.002 ETH are ignored

Tie-break

the coin with more unique buying wallets

[ 05 · why launchpad coins die ]

Launchpads made coins cheap to create and easy to forget.

Nothing behind the chart

Most coins have nothing behind them after launch day. The chart is the whole product, so it dies with volume.

Tools nobody reopens

The wave of coins that fund an AI builder proved people will pay for a coin that ships something. What ships is gallery pages and trackers with placeholder data.

Budgets that sit idle

In one live comparable, a coin's agent spent about $17 of a $364 budget. The rest sat in a wallet the platform controls.

One seed for every wallet

The same model derives every wallet from one server-side seed. One leak and every coin is exposed.

Holders have no job

No level to unlock, no side to take, no reason to open the page on day eight.

Matchstick's thesis

People do not return for tools. They return for games and for rivalries. So every coin gets a game, built in public by an AI its own fees pay for, and every week two coins meet in a one-hour build battle. The fight is the marketing: two builds side by side, a live net-inflow meter, and a burn transaction at the end that anyone can open.

[ 06 · them vs us ]

Same venue. Different product.

Other launchpads

Matchstick

Launchpad coins with nothing behind the chart

Every coin ships a playable game within minutes

AI app launchpads that build tools nobody opens

Games people play, score and send to friends

Build budgets that sit idle for weeks

Idle budget buys and burns the coin after 14 days

One server key that signs for every wallet

One ownerless vault per coin, no custodial keys

Holders with nothing to do after they buy

Holders vote the next level and back their coin weekly

Launch, then hope for volume

A fixed battle calendar: a reason to return each week

Proof by screenshot

Every burn and payout is a transaction anyone can open

[ 07 · $MATCH ]

Matchstick opens doors. It never buys power.

20% of every coin's vault inflow buys and burns Matchstick, every burn is attested, and holding it opens access: early games, theme votes, a credits line. Inside a game every player plays the same game.

Without $MATCH

— Play every public game
— Watch every battle
— Launch with ETH
— A spectator

With $MATCH

■ Play every new game 24 h before the public
■ Vote the weekly theme and the matchups
■ Pay the starter build in Matchstick: +25% build credit, and the Matchstick is burned
■ A member of the arcade: badges, first plays, a vote

What a token never buys

✕In-game power: no damage boost, no extra lives, no better spawn. Every player plays the same game.
✕Leaderboard position: scores are cosmetic, carry no value and pay nothing.
✕Battle outcomes beyond the coin's own net inflow.

Supply sinks

■Vault burn — 20% of every coin's vault inflow, every run()
■Launch burn — starter builds paid in Matchstick, 100% burned
■B2B burn — 50% of commissions and sponsored theme weeks

Tier

Hold

Access

Visitor

■ every public game and battle
■ launch with ETH

Player

100k · 0.01%

■ every new game 24 h early
■ player badge on score cards

Arcade

1M · 0.1%

■ vote the weekly theme and matchups
■ submit theme ideas to the ballot

Founder

10M · 1%

■ propose house-game concepts
■ credits line in every season recap

Votes are capped at 2% of the ballot per wallet.

Every game's own coin

≥ 0.1% of a coin (1,000,000 of 1B)

■ Play the newest level before everyone else
■ Submit and vote features in that coin's queue
■ Holder badge on that game's score cards

≥ 1% of a coin

■ Name a level or a character in the next update

Holding is a key, never a purchase: it opens content and votes. It never changes a score, a speed or a result.

$MATCH · contract facts

Total supply: 1,000,000,000 Matchstick
Launch: Pons V2 fair launch on Robinhood Chain
On the curve: 100% — every token is bought
Team allocation: 0%
Presale / private: none
Vesting: none — nothing to vest
Graduation: 4.2 ETH raised on the curve, then the venue's pool with its 1% fee hook
Burn cadence: every 10 minutes when at least $5 is pending; each burn is a buy, a burn() call and a published record

Every game's coin

Supply: 1,000,000,000 (venue default)
Launch: Pons V2 fair launch, launcher signs
Creator tax: 0% — the 1% venue fee is the only fee
Fee recipient: that coin's MatchVault (no owner)
Supply burns: battle wins (the other war chest) and idle-budget burns after 14 days

[ 08 · where money enters ]

Games take no money. Trading fees do the work.

01
Trading fees on every coin
1% per trade, 70% to its vault — the only money a game's coin ever receives
02
Starter-build payments
~0.01 ETH at launch, spent on the first build's compute
03
Matchstick's own creator fee
70% of 1% on Matchstick trades, which funds platform operations
04
B2B
commissioned games and sponsored theme weeks, paid in ETH — 50% burns the token, 50% funds operations

Operations vault · how income is spent

Compute ceiling and hosting50%
Security reviews and contract audits20%
Growth: battle production, clips, events20%
Reserve10%

What we will never charge for

—Playing a game
—Watching a battle
—Reading a receipt or a build log
—Withdrawing a launcher share — it is paid to their wallet
—Leaving: games stay up for as long as they have players

[ 09 · architecture ]

No custodial keys. Public by default.

Web app

app · games · battle · docs

API + queue

briefs · builds · battles

Build sandboxes

isolated, no network

Indexer

trades · runs · burns

Managed Postgres

encrypted

Game hosting

static, edge CDN

Robinhood Chain

Pons V2 · MatchVault · BattleRegistry

MatchVault · one per coin · no owner · no upgrade path

run()
sweep + claim accrued fees from the venue escrow, split 45 / 20 / 20 / 15, tip the caller 0.5% of the build share
spendCompute(receiptHash, amount)
only the builder address; per-day cap per coin; every call references a published build receipt
settleBattle(battleId, winnerToken, dataHash)
only the settler, only after the 24 h challenge window; buys the winner and calls burn() (cap 0.5 ETH)
burnIdle()
anyone; if no build was paid for 14 days, the build share buys and burns the coin itself

Security model

■No custodial wallets: users sign with their own wallet
■No private key on any web server
■Builder and settler keys hold only network-fee money; the vault caps what they can draw per day
■Every write endpoint fails closed without a valid signature; rate limits per wallet and per IP
■Generated games cannot touch wallets or the network
■Receipts, burns and battle data are public by default

[ 10 · the ladder ]

Eight-week seasons. Five ranks. One hall.

Coins climb by battle record. The season final is a four-coin bracket on week 8. Player boards are cosmetic: no tokens, no payouts, no value.

Kindling

launched — eligible for matchmaking

Spark

1 battle won — ladder badge

Flame

3 battles won — featured on /games

Blaze

5 battles won — season final seeding

Beacon

season winner — permanent hall entry

Badges

First Strike
shipped v1 in under 10 minutes
Undefeated
won every battle in a season
Comeback
won a battle after losing the last one
Crowd Favourite
most unique buying wallets in a week
Long Burn
30 days with a shipped update each week
Speedrun
a player beat the newest level within an hour of release

[ 11 · anti-gaming ]

Every way to cheat a battle, and what stops it.

The attack

The defence

Wash trading to inflate a score

net inflow — buying and selling the same amount nets to zero

Buy at minute 59, sell on Sunday

24 h hold-back subtracts those sells before settlement

One whale decides the fight

one wallet counts for at most 0.5 ETH

A thousand dust wallets

buys under 0.002 ETH are ignored

A big coin farms a tiny one

matchmaking only within 2× market cap

Battle every week for burns

one battle per coin every two weeks

A launcher votes for their own fight

launchers cannot vote in a ballot that includes their coin

Cheating a game's high score

scores carry no value — there is nothing to win but a mark on a board

[ 12 · roadmap ]

Product first. Token after two demo battles.

Phase 0

done

Foundations

— Concept locked: coins that ship games + weekly battles
— Name, handles and the pixel M
— Economy draft: 45 / 20 / 20 / 15 vault split
— The blueprint

Phase 1

● now

MVP

— MatchVault + BattleRegistry, simulated against the live venue factory
— Builder v1: brief → template → sandbox → gates → deploy
— Four templates: runner, arena, tower defense, iso builder
— Web app, indexer, four house games

Phase 2

First battles

— Two demo battles between house coins, fully public
— Matchstick fair launch after both demo battles settle
— Launch queue opens, curated: one new coin per week
— Docs: contracts, receipts, scoring, content policy

Phase 3

later

Open arcade

— Open launches with the content check in the brief
— Season 1 ladder and the week-8 final bracket
— Theme voting, sponsored theme weeks, commissioned games
— Templates 5–8: puzzle, racing, co-op, platformer

Phase 4

later

On-chain referee

— Own pool hook: net inflow recorded on-chain per window
— Permissionless settlement from on-chain data
— Community templates after security review
— Cross-coin tournaments, 2 vs 2, embeddable widget

[ 13 · key decisions ]

What we chose, and why.

Games, not tools

people return for games and rivalries, not for trackers

One ownerless vault per coin

no custodial seed; the split cannot be changed by anyone

Split 45/20/20/15

build budgets idle elsewhere; part becomes battle burns

Loser's war chest burns the winner (≤ 0.5 ETH)

a real consequence makes the battle worth watching

Net inflow scoring

wash trading scores zero

24 h hold-back

flip-and-dump does not count

Starter build at launch (~0.01 ETH)

a game within ~15 minutes, never an empty coin page

Curated queue first

quality and cadence before open launches

Matchstick after 2 demo battles

product first, token second

Scores carry no value

nothing to cheat for and nothing to cash out

Idle budget burn (14 d)

money either builds or burns

[ 14 · what we watch ]

And what we ignore on purpose.

Time to first playablethe launch hook
Weekly players per gameis it a game or a chart
Returning players (D7)retention
Battle page viewersthe weekly ritual
Net inflow per battlesupport that stays
Build budget spent vs idlemoney doing work
Matchstick burned (ETH)the platform loop

Ignored

✕ Raw trading volume — it can be washed
✕ Launches per day — cadence beats count
✕ Follower counts

[ 15 · the honest take ]

Where this is strong, and where it can fail.

Strengths

■The first launchpad on Robinhood Chain where every coin ships a game and coins meet in a weekly build battle
■Every battle advertises both games; more coins mean better matchups and more viewers
■Every build log, gate result and player session makes the next brief and template better
■A coin's game, levels, boards and ladder history live here — leaving means leaving its players
■A tested template shelf plus a licensed asset kit turns $10–30 of compute into a real game

Risks we accept

✕Venue dependency: the launchpad owner can reroute a coin's future fee recipient after a 3-day timelock. Contained by monitoring alerts and our own hook in Phase 4.
✕Settler trust in the MVP. Contained by public data, a 24 h challenge window and a per-battle cap.
✕Game quality at small budgets. Contained by templates, gates and a reviewer model before every deploy.
✕Low-effort launches. Contained by the starter-build cost, the content check and a curated queue first.
✕Market risk: coins can lose all value. Every page says so above the fold.

Launch a coin. Get a game.

Play the sample Read the docs

Launch a coin.Get a game.