Play three silly games. Climb the leaderboard.

1. Overview

Minicrypt is a browser arcade with three games and a fortnightly prize pool. Players post scores, the best hundred in each game share that game's slice of the pool, and the whole cycle repeats every fourteen days. Rewards are paid in USDC on Solana. A separate token, $MINI, captures fee revenue and is burned rather than distributed.

The design goal is narrow: make a leaderboard worth competing on, where the prize is real and the ranking is trustworthy. Everything else follows from those two requirements.

2. The games

Toilet Run is a three-lane endless runner. The metric is distance in metres. Inputs are four discrete actions: move left, move right, jump, duck.

Longboi is a grid snake on a sixteen-by-sixteen wrapping board with a sixty-second clock, extended three seconds per gold pickup. The metric is points. Inputs are four queued directions, and the simulation already advances on a fixed-step accumulator.

Lily Leap 3D is a vertical climber. The metric is height reached. Inputs include keyboard, pointer and device tilt, the last of which is continuous rather than discrete.

Each game is ranked separately. A player's best run in a game is their entry for that game; runs are not aggregated across games, so being weak at one costs nothing on the others.

3. Seasons

Seasons are fourteen days, beginning 21 September 2026 and running on fixed boundaries thereafter. A fixed boundary is deliberate: everyone knows exactly when the clock resets, and the pool is fixed at the start of the season rather than at the end, so players know what they are competing for on day one.

At close, the board freezes for 48 hours. No new scores are accepted and no payouts move. This window exists so the top of each board can be re-verified and forged runs removed before money is committed. Ranks below a removed run move up.

Boards reset each season. A rolling all-time board would lock out anyone who arrived late; a reset gives every season a genuine start.

4. Score verification

Current state: not implemented. Scores are presently read from the player's own browser and can be forged. This is disclosed in the product and is the single largest gap between the prototype and a launch that handles money.

The planned design does not trust submitted numbers at all. A ranked run submits a server-issued seed, a session token, and the complete input log expressed as tick-indexed events. The server re-runs the same simulation from that seed and derives the score itself. A run is accepted only if the re-simulation reproduces it.

This requires each game's simulation to be separable from its rendering, deterministic, and runnable headless. Two prerequisites follow: all gameplay randomness must come from a seeded generator rather than the browser's, and logic must advance on a fixed timestep rather than on frame time. Longboi already meets the second and needs one call site changed for the first, which is why it is the reference implementation. Toilet Run needs both. Lily Leap additionally needs its tilt input quantised to an integer per tick before it enters the simulation.

A caveat worth recording: the four basic floating-point operations are identical across JavaScript engines, but trigonometric functions are not guaranteed to be. Simulations that depend on them can diverge between a player's browser and the verifying server. Longboi's simulation is pure integer grid state and avoids this entirely; the other two either restrict their simulation maths or verify within a tolerance.

5. Rewards

The reward pool is funded by the 1% rewards share of the 3% trading fee. It is not a fixed figure: it tracks eligible trading volume over the season. One million dollars of eligible volume produces roughly ten thousand dollars of rewards.

The pool divides equally across the three games, with one condition. A game must reach 50 ranked players to draw its full third; below that it pays proportionally and the shortfall rolls into the next season. Without that gate, an empty board would be the highest expected return in the arcade and rational players would abandon the popular games for it.

Within a game the top 20 are paid on a configurable curve that sums to exactly 100%: 19% to first, 13% to second, 10% to third, 8% to fourth, 7% to fifth, 5% each to ranks 6 through 10, 2.4% each to 11 through 15, and 1.2% each to 16 through 20. Sixty winners per season across the arcade.

Only the highest verified score of the season counts. Scores are not cumulative, and no amount of $MINI spent adds points. One payout per verified account per game per season.

5a. Attempts

Every verified player gets five free plays per game per day, fifteen across the arcade, resetting at midnight UTC with no rollover. Beyond that, extra attempts can be bought with $MINI at a target of $0.05 each, capped at ten per game per day.

A separate and more important limit applies: ranked runs are capped at 60 per game per season for everyone, paid or free. This exists because best-of scoring rewards volume independently of skill — a player with four times the attempts has a materially higher expected maximum. Without a shared ceiling, paid attempts would buy rank in substance while the documentation claimed otherwise. The cap makes the claim true. Paid attempts change how fast you can use your allowance, never how large it is.

Pricing is set per hour from an oracle or TWAP rather than per transaction, and clamped between $0.03 and $0.10. A stale, unavailable or implausible price pauses paid attempts; it never falls back to a guess.

6. Token

$MINI has a maximum supply of one billion at nine decimals, and no mint function. Mint authority is revoked at deployment and can be verified by anyone on-chain. Supply moves in one direction only.

A 3% ecosystem fee on trading splits three ways: 1.5% buys $MINI and burns all of it, 1% funds USDC player rewards, and 0.5% goes to treasury. Rewards are never paid in newly minted or newly allocated $MINI.

Where the fee is collected matters. A transfer tax on Solana requires the Token-2022 fee extension, which taxes every transfer including wallet-to-wallet sends, has uneven AMM and aggregator support, and is declined by most centralised exchanges. Collecting the same 3% from liquidity-pool activity instead gives identical economics, scopes the fee to actual trading as intended, and imposes no listing ceiling. This is the recommended architecture.

The second burn is independent of trading. Paid attempts are bought with $MINI and split 70% burned, 20% treasury, 10% ecosystem. This sink scales with how much the games are played rather than how much the token is traded, which makes it the only supply reduction tied to genuine usage.

Rewards are denominated in USDC deliberately. Paying winners in the same token the buyback is bidding up means they sell into that bid every fortnight and the two flows substantially cancel. Separating them makes the burn unambiguous and gives players something they can hold.

Allocation of the one billion: 350M public and liquidity, 200M ecosystem on a scheduled release, 150M treasury under multisig, 100M team on a 12-month cliff followed by 24-month linear vesting, 100M marketing and partnerships vested, 50M founder holding locked for 12 months, 50M development and contingency. Sixty-five percent therefore begins outside public hands. Every allocation has its own published wallet and release schedule, published at launch rather than afterwards.

Circulating supply, locked supply and burned supply are reported separately and never conflated. An allocation that exists but is locked is not circulating.

A known structural weakness remains: the reward pool scales with trading volume rather than with player count, so prizes shrink during quiet trading precisely when retention matters most. The gameplay burn partly offsets this on the supply side but does not fund rewards. Candidate mitigations are a treasury buffer smoothing payouts across seasons, or routing a share of paid-attempt revenue into the reward pool. Unresolved.

7. Roadmap

Phase 1 — verification. Split Longboi into simulation and rendering modules, seed its randomness, and stand up the server-side replay pipeline against it. Establish the submission contract.

Phase 2 — backend. Seed issuance, session tokens, wallet signature authentication, season tables, and the freeze-and-review workflow. Run complete seasons with no money attached and observe what breaks.

Phase 3 — port. Bring Toilet Run and Lily Leap onto the same contract.

Phase 4 — payouts. Token deployment, claim contract with per-season Merkle roots, and the first live season.

Phases 1 to 3 constitute a working arcade with trustworthy leaderboards and no token. That is deliberate: if the token timeline slips, there is still a product.

8. Risks and disclaimers

Nothing on this site is an offer, a solicitation, or investment advice. $MINI is not an investment product and confers no ownership, revenue share, or governance rights. Prize figures shown anywhere in the product are projections against a placeholder pool, not commitments.

Paying value based on competitive performance may be regulated differently across jurisdictions, and may engage rules on digital payment tokens, gambling, or prize competitions depending on how it is structured and where a player is located. Access may be restricted in some jurisdictions. This document is not legal advice and the project has not yet completed a regulatory review.

Smart contracts carry risk of loss through bugs or exploitation. Token prices can go to zero. Play with money you can afford to lose entirely, and treat any prize as a bonus rather than income.