What the protocol actually does, written from the frozen source code — mechanisms, figures, residual powers and accepted limitations. No marketing.
Fixed in the code at deployment. No function can change them — not even for the owner.
| Allocation | Amount | Share | Where the tokens sit |
|---|---|---|---|
| Smart-Vault | 500,000,000 | 50.00% | Held by the $RMHT contract itself, released only through milestones |
| Liquidity (LP) | 433,001,104 | 43.30% | Single-sided Uniswap v3 position, locked forever inside the LiquidityCustodian |
| Founder | 50,000,000 | 5.00% | FounderCustodian, one-year time lock, immutable |
| Community airdrop | 16,998,896 | 1.70% | AirdropCustodian, 18 addresses, individually claimable |
This is a 1-for-1 airdrop: every $MHT-BSC token held at the snapshot converts to 1 $RMHT. It is not a pro-rata share of a fixed envelope — the 16,998,896 total is simply the sum of the real balances of the 18 retained addresses, each truncated down to a whole token. The snapshot was frozen on 29 August 2026, and the table is written into the contract at deployment in a single transaction, with no way to add to it afterwards.
Half the supply is locked inside the contract, and only leaves when market cap crosses a milestone — and stays there, not for the span of one block.
A milestone is not crossed because the price touched a threshold once. It is crossed because market cap stayed above that threshold across several measurements spaced out in time. That is what makes an artificial pump useless here: you would have to hold the price up for thirty minutes of time-weighted average, then twenty more minutes of confirmations — something a flash loan structurally cannot do.
The calculation applies to the remaining vault balance, never to the original 500 million. The amount released therefore shrinks with every milestone. The vault drains asymptotically: there is no exhaustion date, and no single milestone can ever empty it.
Nobody in particular. And that is deliberate.
The contract exposes a public function, pokeMilestone(), that anyone can call, at any time, without permission. It reads the price, compares it to the target, and advances the state machine by one step if the conditions are met. If they are not, it does nothing and costs the caller only a little gas.
There is no third-party keeper, no proprietary bot, no subscription to fund. The protocol depends on no external service to move forward: if a single holder anywhere calls the function three times, ten minutes apart, while the price holds, the milestone fires. This is a deliberate architectural choice, made after what happened to $MHT V2 on BSC, where a paused third-party automation left the mechanism frozen for good.
A milestone cannot be "forgotten". Nor can it be fired early or quietly: the function only evaluates public, verifiable conditions. Whoever calls it receives no privilege, no share, no advantage — they pay the gas, and that is all.
Two sources, never just one, and never an instantaneous price.
Market cap is computed in two stages: the price of $RMHT in ETH, then the price of ETH in dollars. Each stage has its own protection.
The protocol never reads the pool's spot price. It uses a time-weighted average over a full 30 minutes (Uniswap v3 TWAP). Two direct consequences:
pokeMilestone() does nothing. The protocol would rather do nothing than work from partial data.So that this window cannot be flushed by transaction spam, the pool is sized at deployment to store 1,801 observations — the full window, to the second. The contract refuses to be configured against a pool that lacks that capacity.
The dollar conversion comes from the official Chainlink ETH/USD feed on Robinhood Chain. The contract rejects any reading older than 26 hours, refuses a zero or negative price, and verifies that the answering round is not older than the requested one. The 26-hour threshold is not arbitrary: the feed advertises a 24-hour heartbeat, plus two hours of jitter margin.
Manipulating a milestone means defeating all three at once: moving a 30-minute average, holding it across three confirmations spaced 10 minutes apart inside a 2-hour window, and doing so with a fresh ETH/USD price supplied by an oracle you do not control. Any single layer could be worked around. All three together are out of flash-loan territory.
Pro rata to what you hold, with no registration and nothing to do to become eligible.
When a milestone fires, the released tokens are not pushed to every address one by one — that would not scale. The contract increments a global "reward per token" accumulator. Your share follows from your balance at any moment, and you collect it whenever you want by calling claimRewards().
An important consequence: you do not have to do anything to accrue. Your entitlement grows as long as you hold. You lose nothing by waiting, and nothing expires.
Each holder's share is computed against the "eligible supply" — total supply minus everything that must not dilute holders:
| Excluded | Why |
|---|---|
| The vault (tokens still in the contract) | Nobody holds them — counting them would mean paying rewards to ourselves |
| The Uniswap pool | Liquidity is not a holder; including it would dilute everyone in favour of the pool |
| The burn address | Destroyed tokens, with no owner |
Everything else counts, with no exceptions and no allow-list.
Three separate contracts, with different rules, all deployed alongside the token.
| Contract | What it holds | Exit rule |
|---|---|---|
| LiquidityCustodian | The Uniswap v3 position NFT | None. There is no function to withdraw liquidity or transfer the NFT. The position is locked permanently. The only thing that can leave is trading fees, through collect() — callable by anyone, and always paying the configured recipient, never the caller. |
| FounderCustodian | The founder's 50,000,000 | A one-year time lock, set at construction and immutable — no function can bring it forward. The milestone rewards those tokens accrue are, by contrast, claimable at any time. |
| AirdropCustodian | The airdrop's 16,998,896 | Claimable by each beneficiary as soon as either condition is met: one year elapsed, or the 5th milestone reached. Milestone rewards are claimable separately, without waiting for the principal to unlock. |
There is no transaction tax. Revenue comes from Uniswap v3 swap fees, captured by the LiquidityCustodian. The function that collects them is open to everyone — not a weakness, but the guarantee that collection keeps working even if nobody is left to take care of it.
What disappears after launch, and what does not. The part that cannot honestly be summed up as "renounced".
After deployment, the configuration is locked (lockConfig()) and ownership is then renounced on all four contracts. From that point the pool is frozen forever, allocations can no longer be changed, no token can be frozen or blacklisted, and there is no pause function anywhere.
One role deliberately remains, and it deserves to be stated plainly rather than discovered in the code.
This role is entirely separate from ownership and survives renouncement. It can do exactly one thing: replace the addresses of the two Chainlink oracles — the ETH/USD feed and the sequencer uptime feed. It cannot touch the pool, the allocations, the vault, any balance, or any threshold or constant in the protocol. Every other administrative function is out of its reach, and that is enforced by test.
Why it exists: on $MHT V2, on BSC, an oracle dependency failed after contract ownership had been renounced. There was then no way left to repair anything, and roughly 76% of the supply ended up frozen for good. feedGovernor is the answer to that specific failure — the minimum ability to repair a broken oracle, and nothing more.
It is held by a hardware wallet, separate from the deployment key, can be transferred in two steps to a multisig, and can be permanently destroyed by its holder through renounceFeedGovernor() on the day the feeds are considered settled. From that day, the protocol has no residual power of any kind.
| After launch | Status |
|---|---|
| Change the pool, the milestones, the supply, any threshold | Impossible — constants, or locked configuration |
| Add or modify an airdrop allocation | Impossible — one-shot, then ownership renounced |
| Freeze, blacklist or seize tokens | Impossible — the function does not exist |
| Pause the protocol | Impossible — the function does not exist |
| Withdraw liquidity | Impossible — the function does not exist in the custodian |
| Replace a Chainlink oracle address | Possible, by feedGovernor only |
What is not perfect, and why this trade-off was chosen over the alternative.
The code knows how to verify that an L2 sequencer is running before accepting a price, and that path is tested. But Chainlink publishes no uptime feed for Robinhood Chain, and has announced it is no longer deploying them to new networks. Making the check mandatory would have blocked every price read forever, freezing half the supply. The check is therefore disabled at deployment, and feedGovernor can switch it on without redeployment the day a feed exists. In the meantime, the Chainlink freshness check, the 30-minute average and the three confirmations all remain active.
One of the 18 snapshot addresses is a custodial wallet whose holder does not control the keys. Its 150,000 tokens — 0.88% of the airdrop — were credited anyway, so that the announced rule applies with no exception in either direction. The mechanical consequence, stated here rather than hidden: that share stays counted as active and will permanently take 0.88% of the milestone rewards otherwise going to the other 17 beneficiaries.
Because each milestone releases 1.5% of what remains, the series converges towards zero without reaching it. That is a property of the model, not a defect: it guarantees there is always something left to distribute, but it also means a residual fraction stays in the contract indefinitely.
The TWAP needs 30 minutes of observations before it can be used. The final configuration lock therefore cannot run in the same breath as the deployment: there is a short period during which the token is tradable while the configuration is not yet frozen. That window ran from the deployment to the configuration lock, both on 4 September 2026; what happened inside it, and what the launch cost down to the wei, is written up on the transparency page.
Nothing above requires taking our word for it.
Every constant quoted on this page is a public variable of the contract, readable from any explorer or from the command line. The most useful ones:
vaultBalance — what is left in the vault, livemilestonesReached / nextMilestoneUSD — how many milestones have fired, and the next targetmilestoneStatus() — the current state of the machine: armed or not, confirmations gathered, measured market capgetMarketCap() — market cap as the contract computes it, not as an aggregator estimates itisTwapReady() — whether the pool already covers the 30-minute windowowner() and feedGovernor() — to check the governance state for yourselfpendingRewardsOf(your_address) — what you can claim right now322 automated tests, including invariants run across 128,000 random calls with no violation. Eight independent analysis tools run against the code: Slither, Solhint, Mythril, Wake, Aderyn, Surya, SolidityScan and the Foundry suite. Line coverage between 98.4% and 100% depending on the contract. SCSVS/SCSTG checklist: 227 items reviewed, no Critical, High or Medium item left open.
Automated scanning additionally raises two Critical and one High. All three are analysed as false positives, and the reasoning is written into the source rather than kept private:
collect() — the fees it moves can only reach the configured recipient, never a caller-chosen address;require(slot0().unlocked) and covered by a dedicated test.A scan re-run by anyone will show those three. They are listed here so the numbers match what an independent check returns.
Deployed on Robinhood Chain (chain 4663) on 4 September 2026. The four contracts we wrote are verified on Blockscout — readable source, exact match against the deployed bytecode — and ownership has been renounced on all four.
| Contract | Address | State |
|---|---|---|
| $RMHT token and vault | 0xbD4487dad62d23e0677E6a94c99cB0AE45328bA4 | Verified · owner() = address(0) · configuration locked |
| RMHTAirdropCustodian the airdrop's 16,998,896 | 0xdEc68344a41Cc6104ed5ff2C88F2D16a794688C1 | Verified · owner() = address(0) |
| RMHTFounderCustodian the founder's 50,000,000 | 0x4C944c27160A1532b9De3a69d0Ec64164a6f03d8 | Verified · owner() = address(0) |
| RMHTLiquidityCustodian holder of the LP position | 0x50336e2a1396364895702a5222faC47bD0d38407 | Verified · owner() = address(0) · position tokenId 1004464 |
| Uniswap v3 pool RMHT / WETH | 0xabe3B1fF5Fc6a9638D0c46f2379B20f7e6173bEe | Deployed by the Uniswap factory, not by us |
The only role that survives the renounce is feedGovernor, held by 0xC90e404A9f2ea55cbD2C81cf0054FAe4A721889a; what it can and cannot do is set out in section 07. The scanner findings, the market-cap gap against aggregators and what the launch cost are answered point by point on the transparency page.