How the machine works, stated plainly. The contracts are the authority.
DIVVY is a memecoin on Robinhood Chain with one mechanic: 2% of every trade is routed into a contract — the RewardVault — and settled to holders every four hours, weighted by bag size and by how long a holder has sat still. The team's wallets are excluded from settlement in the vault's constructor.
There is no dashboard to trust and no operator to believe. The vault has no owner, the split has no setter, and the harvest can be triggered by anyone. This page explains each part; the contracts are the authority wherever wording and code could disagree.
Read thisDIVVY is not an investment product and this page promises nothing. The pot is a direct function of real trading volume and can fall to nearly zero. Read 'What it does not do' before anything else.
A buy or sell of $DIVVY pays a 2% creator fee into the launchpad's fee escrow. The rate is set at launch and can never change — the contract enforces this, not policy.
The fee recipient is the RewardVault contract, set at deploy. No private key is in the path; no human ever custodies fee revenue, even briefly.
External, unpermissioned. Pulls the escrow credit into the vault and applies the split in the same transaction. If the team disappears, any holder can call it.
10% to the treasury address, 90% to the open epoch pot. Both constants; neither has a setter.
Every 4 hours: snapshot balances, drop excluded addresses, weight = balance × streak multiplier.
A merkle root of cumulative claimable amounts is published on-chain. The snapshot tool is open source, so any root can be rebuilt from public data and compared.
Cumulative, never expires, one transaction collects everything unclaimed. The claimer pays their own gas.
Burned into the constructor with no setter: every team wallet, the launch-buy wallets, the treasury, the liquidity pool manager, the launchpad locker, the burn address, and the vault itself. Excluded addresses are dropped before weights are computed, so they never dilute anyone and never appear in a merkle leaf.
This is the product's central claim. A promise can be broken; a constructor cannot. The addresses are published before launch so the claim can be checked against the deploy, not after the fact.
Mode is stored on-chain per address, costs one transaction to change, and applies from the next epoch. Nobody is forced into nine thin positions; nobody who wants them is denied.
This attacks the number that kills reward-basket tokens: sells outpacing buys within days of launch. Sitting still has to be worth something, or nobody does it. The exact schedule is fixed at deploy.
The basket is nine seats with fixed weights (15 / 12.5 / 12.5 / 12.5 / 10 / 10 / 10 / 10 / 7.5). Weights live with the seat, not the token, so a contest is only ever about which token deserves the chair.
Send $DIVVY through the SeatBallot naming a seat and a screened replacement. Vote weight equals the amount burned. The burn is final whether the vote wins or loses.
Only candidates passing the on-chain screen are votable: graduated, minimum liquidity, minimum age, minimum holders, not already seated. Free-typed addresses are rejected.
At the rebalance boundary the contract tallies and the leading candidate takes the seat automatically. No review step exists; nobody can veto it, including the team.
The only inbound stream is the 2% fee on $DIVVY trades. There are no subscriptions, no listing fees, no paid board placement, and no treasury trading. When volume is quiet the pot is small, and the site will say so — the decay side of the curve is published next to the live figure.
Launch stance: the curve will not be self-funded. If it has not filled 24 hours after launch, the remainder is funded from treasury and announced in the same hour with the transaction linked.
Snapshots and merkle construction run off-chain, which means the team computes who gets what. That is acceptable for one reason only: the snapshot tool is open source from day one and every input is public chain data, so anyone can rebuild any epoch's root and prove — or disprove — every weight in it. Independent rebuilds are a health metric we publish.
Read thisKnown weaknesses, on the record: we are second to market; the contracts are new code with no patch path; there is exactly one revenue source; and a launchpad failure would break fee routing with it. If any of these is disqualifying for you, do not buy.
Open the RewardVault on the explorer and read the exclusion list — confirm the team addresses published at launch are in it.
Clone the snapshot tool, point it at any closed epoch, and rebuild the merkle root from chain data alone. Compare it with the root stored on-chain.
Call harvest() from any address and watch it run. Permissionless is a property you can test, not a word you have to trust.
If a rebuilt root ever fails to match the published one, the project is over — and should be. That is the deal.