Risk Assessments of Collateral Exposures on HyperEVM

Hello SummerFi community, BA Labs is posting a comprehensive risk assessments for each collateral being onboarded to SummerFi Fleets on HyperEVM (currently only USDC and USDT LR Fleets are live).

Below find the list of collaterals initially onboarded (at the moment of posting this), for which the assessments are to be posted. For any new collateral addition, the assessments are going to be posted subsequently.

Initial list of collaterals for USDC & USDT LR Fleets on HyperEVM:
[‘USDH’, ‘kHYPE’, ‘UBTC’, ‘UETH’, ‘PT-kHYPE’, ‘WHYPE’, ‘wstHYPE’, ‘beHYPE’, ‘USOL’]

This above collateral exposure includes the following ARKs:

  • Hyperlend (USDC & USDT), Felix Morpho USDC, Felix Morpho USDT0, Gauntlet Morpho USDC, Gauntlet Morpho USDT0.

Assessments are to be posted as one collateral asset per post in this forum thread, which is aimed to be a continuous point of risk transparency for SummerFi’s HyperEVM deployment.

2 Likes

kHYPE Risk Assessment by BA Labs

Part I: Protocol Overview

A.1 Protocol Details

kHYPE and Underlying Collateral

kHYPE is the liquid staking token (LST) issued by the Kinetiq protocol for the Hyperliquid network. The underlying collateral of kHYPE is the native Hyperliquid token HYPE, which users stake via Kinetiq. In exchange for depositing HYPE, users receive kHYPE tokens at a 1:1 rate (initially). Each kHYPE token represents a share of the total staked HYPE pool, analogous to how Lido’s stETH represents staked Ether. The total supply of kHYPE is fully backed by HYPE tokens held and managed by the protocol’s smart contracts and validators. Thus, HYPE is the sole underlying collateral securing kHYPE’s value.

Underlying Yield Sources

The yield for kHYPE holders is generated by staking HYPE on the Hyperliquid network’s validators. Hyperliquid is a high-performance Layer-1 chain (inspired by Cosmos/Tendermint) with a native Proof-of-Stake mechanism. Staked HYPE earns block rewards and transaction fee revenue from validators, increasing the total HYPE backing kHYPE over time. Kinetiq’s design actively maximizes yield by delegating HYPE to only the top-performing validators via an autonomous “StakeHub” scoring system. This means kHYPE yield comes entirely from on-chain staking rewards (inflationary HYPE issuance and validator fees) and potentially minor MEV gains, with no external yield farming. As validators earn rewards, the accrued HYPE is reflected in kHYPE’s exchange rate (HYPE per kHYPE rises). Notably, slashing events are a negative yield source – if a validator misbehaves, a portion of staked HYPE can be slashed, reducing the backing. Even though the slashing mechanism is not live in HyperEVM, Kinetiq has an oracle implementation which has a placeholder for validator slashing and penalties, all the further mentions of slashing is logic related to the slashed value reported by the oracle. Hence Kinetiq’s Oracle and Validator Manager components handle reward and slashing updates to ensure kHYPE supply tracks the actual staked HYPE balance. In summary, validator rewards (minus any slashing and protocol fees) are the sole yield source for kHYPE holders.

Minting (Staking) Flow

To mint kHYPE, a user deposits HYPE into the Kinetiq smart contract on the Hyperliquid network (or a connected HyperEVM environment). Upon deposit, the user’s HYPE is added to the staking pool. The protocol immediately mints an equal amount of kHYPE ERC-20 tokens to the user’s address (assuming a 1:1 exchange rate at inception). Behind the scenes, the StakingManager contract manages this flow: it accepts HYPE deposits and holds them in a buffer, then delegates HYPE to validators in optimal proportions. To maintain liquidity for withdrawals and account for network constraints, Kinetiq employs a HYPE buffer strategy. A portion of deposits may be kept temporarily unbonded (in a buffer) to satisfy immediate withdrawals or to align with the L1’s 8-decimal precision requirements. Any HYPE amount that cannot be cleanly staked due to decimal precision or that exceeds certain thresholds remains in the buffer as idle liquidity. The exchange rate of kHYPE starts at 1 kHYPE = 1 HYPE and gradually increases as staking rewards accrue. Kinetiq’s StakingAccountant contract tracks the total staked HYPE vs. total kHYPE supply to compute this exchange ratio at any time. This ratio mechanism means that when users stake, they mint kHYPE equal to the HYPE deposited divided by the current ratio (initially 1.0). The protocol is designed to allow continuous deposits; if rewards have accumulated, new deposits will mint slightly fewer kHYPE than the HYPE input (since existing kHYPE has appreciated).

Redemption (Unstaking) Flow

To redeem the underlying HYPE, kHYPE holders initiate an unstaking withdrawal. Due to the Hyperliquid network’s staking properties, unstaking is not instantaneous – it involves a queue and potential cooldown period. Users call the queueWithdrawal() function on the StakingManager contract, specifying an amount of kHYPE to redeem. This burns (locks) the kHYPE and places a request in a withdrawal queue. Internally, if the protocol’s HYPE buffer has sufficient liquid HYPE, some withdrawals may be fulfilled quickly from the buffer. Otherwise, the protocol must wait for an unbonding period and/or actively withdraw HYPE from validators to fulfill requests. The design uses a batched operation model: a keeper or automated process calls a function (e.g. processL1Operations) to handle batched deposits and withdrawals, netting them if possible, and triggering cross-chain transfers of HYPE between the EVM contract and the Hyperliquid core (L1). Once the unbonding period is over and HYPE is available, users can call confirmWithdrawal() to receive their HYPE back to their wallet. Initially, Kinetiq implemented a cooldown such that users had to wait a minimum time after staking before withdrawing; however, this was removed after an audit found it could be circumvented by transferring tokens to another address. Now, withdrawal timing is primarily governed by the network’s unbonding period and the queue order. In an emergency, Kinetiq also supports an “Emergency Withdrawal” mechanism – for instance, if validators become unresponsive or the protocol pauses, validator might be allowed to withdraw from the buffer or via a special process, potentially with penalties. Overall, the redemption flow prioritizes fairness (first-in-first-out queue) and safety (only releasing HYPE that is actually unstaked and available), which is critical to prevent bank-run scenarios or insolvent withdrawals.

Fee Structure and Incentives

Kinetiq’s revenue model is primarily an unstaking fee. Whenever a user withdraws (unstakes) HYPE by redeeming kHYPE, a small fee is taken from the amount. The unstake fee rate is set to 10 basis points (0.10%) by default (speculation from details in the Cantina audit, the actual parameters at launch might be different). This fee is collected in kHYPE (a portion of the burned kHYPE is diverted) and sent to the protocol’s treasury address. Notably, this is the only protocol fee currently – there is no fee charged on staking (depositing), and Kinetiq does not take a performance fee cut of staking rewards (unlike some liquid staking providers). Instead, the protocol earns its revenue when users exit. The rationale is to align with long-term stakers (no fee on entry or ongoing rewards) and only tax those who leave, providing a sustainable treasury for maintenance and security costs. Kinetiq’s audits confirm that this 0.1% withdrawal fee was implemented, with fixes ensuring that the fee is not double-charged on the treasury itself and that tiny withdrawals cannot bypass the fee. Aside from fees, incentives for users include the liquidity and utility of kHYPE. Holders can use kHYPE within the Hyperliquid DeFi ecosystem (e.g. as collateral in lending platforms like Felix or for yield farming in Hyperliquid protocols). This extra utility is an incentive to stake via Kinetiq (earning yield + preserving liquidity). Validators in Hyperliquid are indirectly incentivized by Kinetiq’s “meritocratic delegation”: those with better performance get more HYPE stake from Kinetiq, increasing their rewards. This creates a positive incentive for validators to maintain high uptime and performance to attract Kinetiq’s delegation (though Kinetiq does not pay them a fee; they earn standard network rewards). In summary, user incentives are the yield and DeFi utility of kHYPE, while protocol incentives come from the 0.1% exit fee accruing to the treasury (which could be used for further development, insurance, or distributed to Kinetiq token holders if governance is introduced in the future).

Off-Chain Components

While Kinetiq’s core logic is on-chain, several critical components operate off-chain or in a hybrid manner. The foremost is the StakeHub validator scoring system, which continuously evaluates validators on metrics like performance, uptime, latency, integrity, and decentralization. This scoring algorithm likely runs on off-chain infrastructure maintained by Kinetiq (or its partner, Anthias). It produces recommendations or data that is fed on-chain via an Oracle system. Kinetiq has an OracleManager smart contract that ingests off-chain data about validators’ performance and any slashing or reward events. Off-chain “oracle feeders” (these are hosted in Kinetiq’s own server) periodically submit updates to OracleManager, which then triggers on-chain rebalancing or slashing logic. There is a challenge period and validation of these oracle reports to ensure data integrity. These off-chain components are crucial – if they fail or are compromised, the on-chain protocol might not rebalance stakes or could even be fed false performance data. The StakingManager contract’s code suggests that to actually stake HYPE, it calls an L1 contract (HyperCore) with the amount to delegate. This likely uses an internal bridge or message passing system (possibly IBC or a Hyperliquid-specific mechanism). Off-chain relayer processes probably assist in conveying transactions or in listening for completion of unbonding. Additionally, Kinetiq’s front-end and analytics (the Kinetiq website and dashboard) are off-chain components that display validator scores and status in real time. They rely on indexing nodes or the StakeHub backend to provide a user-friendly view of the staking distribution and performance. Finally, administrative keys and governance (if any) reside off-chain: for example, the team’s multisig that can upgrade contracts (since the contracts are upgradeable proxies) and the Secure Staking Alliance’s oversight. These keys and off-chain governance processes are not part of the Solidity code, but represent a significant component of the overall system’s security model. In summary, off-chain components handle data analysis (validator scoring), oracle feeding, cross-chain communication, and governance decisions, all of which complement the on-chain logic to make Kinetiq function end-to-end.

B. System Design

B.1 Monitoring and Observability

Given its role managing staked assets, Kinetiq places heavy emphasis on monitoring both the blockchain and validators. On-chain, the system emits events for key actions (stakes, withdrawals, rebalances) and tracks various metrics via its Accountant and Manager contracts. Off-chain, Kinetiq’s infrastructure continuously monitors validator performance metrics (uptime, block production, latency) and on-chain state (total staked, pending withdrawals, etc.). The OracleManager enforces that updates occur at least every MIN_UPDATE_INTERVAL and rejects stale data beyond MAX_ORACLE_STALENESS to ensure fresh information. If an oracle update is missed or delayed, the protocol can detect it (on-chain) and potentially pause operations, thanks to sanity checks on update timestamps. Kinetiq likely has alerting systems off-chain to flag anomalies: e.g., if a validator’s score drops suddenly or if a slashing event occurs, the team’s systems would trigger an update (or even an emergency rebalance). The protocol also introduced a ValidatorSanityChecker contract (post-audits) that sets bounds on acceptable oracle-reported values (e.g., maximum slashing or reward delta). This provides on-chain observability of data validity and will reject outlier data (which might indicate an oracle bug or attack) – effectively monitoring the oracle input for sanity. In terms of liquidity monitoring, the Kinetiq team considered scenarios with multiple staking pools and emphasized the need to monitor liquidity buffers in each pool. For example, if multiple StakingManager instances exist (say one for retail and one for an institution), the team or the institution must monitor each instance’s available HYPE and rebalance between managers to prevent any shortfall. This implies a off-chain monitoring service that observes if any staking manager is low on liquid HYPE (buffer) relative to pending withdrawals, so that proactive measures (like topping up from another pool or halting new stakes) can be taken. Kinetiq likely also uses typical blockchain observability tools: e.g., a block explorer (HypurScan) is available for Hyperliquid which can track Kinetiq’s contract transactions. The team can monitor contract events and state changes through these tools or custom dashboards. Security monitoring is in place as well – the PauserRegistry allows certain actors to pause the system if something abnormal is detected. For instance, if a critical bug or an exploit attempt is observed (through on-chain monitoring of balances or off-chain detection of an attack pattern), the protocol can be emergency-paused to protect funds. In essence, Kinetiq’s design incorporates multi-layer monitoring: on-chain assertions and events for transparency, and off-chain watchers and alerting for proactive oversight. This observability is crucial to manage technical risks in real time, as discussed further in Part II.

B.2 System Architecture

Kinetiq’s architecture is modular and robust, reflecting an academic approach to system design. At a high level, it consists of multiple smart contracts each with a focused responsibility, orchestrated together, and complemented by off-chain systems. The key on-chain components are:

  • StakingManager: The core contract that handles user interactions (staking and initiating withdrawals) and manages the flow of HYPE to and from the Hyperliquid validators. Each StakingManager is associated with a specific kHYPE token and a set of validators. It keeps an internal “HYPE buffer” (native HYPE held by the contract) and delegates the rest to validators. It also implements the withdrawal queue and processes withdrawals (often in batch). There can be multiple StakingManager instances (e.g., to segregate different tranches of stake or future expansions), but all feed into a common accounting (if tied to the same kHYPE) – in the current version, the team decided to only deploy one public StakingManager per kHYPE to avoid complexity. The StakingManager has privileged roles (MANAGER_ROLE) to allow the protocol or institutions to trigger certain operations like rebalancing or emergency actions.

  • kHYPE Token (ERC20): This is the liquid staking token contract (named KHYPE in code) which represents ownership of the staked HYPE pool. It implements standard ERC-20 functionality plus ERC20Permit (gasless approvals) for user convenience. Minting and burning of kHYPE are restricted to authorized contracts (StakingManager or Accountant), ensuring that supply changes only in response to actual stake/un-stake events. The kHYPE token is non-rebasing; its balance doesn’t automatically increase. Instead, the exchange rate against HYPE changes as rewards accrue, which is tracked via…

  • StakingAccountant: A global bookkeeping contract that tracks total staked HYPE and total kHYPE minted across the system. It computes the exchange rate (HYPE per kHYPE) using these totals. If multiple StakingManagers exist for the same asset, the Accountant aggregates across them, ensuring that the kHYPE token’s value is correctly calculated and that no double-counting of underlying HYPE occurs. It also can facilitate conversions, e.g., given an amount of kHYPE, how much HYPE is withdrawable, and vice versa. Essentially, StakingAccountant ensures the one-to-one peg at genesis evolves into a fluid exchange rate reflecting actual staking yield. It plays a critical role in maintaining solvency: if any discrepancy arises (e.g., due to rounding or slashing), the Accountant’s numbers would reveal it.

  • ValidatorManager: This component encapsulates the logic for managing validator sets and delegations. It doesn’t hold user funds directly but issues instructions on how the StakingManager should allocate stake among validators. Its responsibilities include: maintaining a registry of active validators, assigning each a performance-based score (likely pulled from OracleManager), and deciding when to rebalance stake (move HYPE from one validator to another to optimize performance or reduce risk). The ValidatorManager likely also handles emergency undelegation – if a validator is flagged for slashing or underperformance, it can initiate withdrawal of stake from that validator (this ties into the Oracle’s slashing reports). In architecture terms, think of ValidatorManager as the brain making delegation decisions, while StakingManager is the hands executing HYPE transfers accordingly. ValidatorManager may also implement slashing protection logic, for example capping how much stake is placed on any one validator or reacting to slashing by proportionally distributing the impact.

  • OracleManager: The bridge between off-chain performance data and on-chain action. This contract receives inputs (likely via a transaction) from an authorized oracle source. The data includes metrics like each validator’s uptime score, proposal success rate, perhaps latency, as well as network-wide info on total rewards or any slashing that occurred in the last period. OracleManager aggregates data from possibly multiple sources for redundancy and computes a consolidated view of validator performance. It then passes this data to the ValidatorManager (or directly updates validator scores there). The OracleManager also triggers reward distribution: when it reports that X HYPE rewards were earned by a validator or the pool, the system can then record those rewards. Notably, Kinetiq’s design uses a “pull” model for rewards – rather than continuously compounding rewards on each validator, they may rely on periodic oracle reports to adjust the exchange rate or to claim rewards. The OracleManager also enforces a challenge period – after an oracle report is submitted, there might be a short window where another source or a sanity checker can dispute or verify the data. This guards against incorrect data from a malfunctioning or malicious oracle. In summary, OracleManager is a critical piece that ensures external state (validator performance, earned rewards, slashes) is correctly and securely reflected on-chain.

  • PauserRegistry (and Emergency Stop): This contract provides a centralized way to manage the paused/unpaused status of various protocol components. It maintains roles for who can pause (likely a multi-signature admin or a governance module) and allows flipping a pause flag on contracts like StakingManager, OracleManager, etc. This design allows Kinetiq to halt operations quickly in a crisis – for example, if a bug is found or an oracle misbehaves, pausing would stop new stakes, withdrawals, or oracle updates, “freezing” the system in a safe state. Each contract likely checks with PauserRegistry’s state (e.g., functions have whenNotPaused modifiers) to honor these emergency stops. This layered architecture with a pausable feature is a common design in critical protocols to mitigate damage from unforeseen issues.

These components interact as follows: Users interface with StakingManager/kHYPE for deposits and withdrawals. The StakingManager delegates HYPE to validators as instructed by ValidatorManager, and uses data from OracleManager to adjust to network events. StakingAccountant oversees the tokenomics, ensuring kHYPE supply and HYPE balance stay consistent. Off-chain, the StakeHub scoring engine continuously evaluates validators and sends data through OracleManager to update scores. Off-chain agents (operated by the team or third parties) also call maintenance functions like batch processing of withdrawals, rebalancing triggers, and updating oracle data on a schedule. The upgradeable proxy pattern is used for many of these contracts (the audits note that OracleManager and StakingManager are behind proxies), meaning the implementation logic can be upgraded while the proxy retains state – this adds flexibility for improvements but also places onus on trust in the upgrade keys (discussed later). Role-Based Access Control is enforced (likely using OpenZeppelin’s AccessControl library): for example, only addresses with a MANAGER_ROLE can call certain admin functions (like adding a validator or triggering an emergency withdrawal), and only a PAUSER_ROLE can pause contracts.

Overall, Kinetiq’s architecture is complex but well-segmented. This modular design improves security (each part can be audited in isolation and has limited responsibility) and allows specialist components (e.g., the Oracle system) to be upgraded or tuned without affecting core staking logic. The presence of multiple audits (see below) attests that each module was scrutinized. The architecture also gracefully handles the multi-environment nature of Hyperliquid: bridging between an EVM environment and the Hyperliquid consensus is abstracted into the StakingManager (which likely contains the logic for calling the L1 HYPE contract and receiving funds back), so other parts of the system can remain agnostic of cross-chain details. In summary, Kinetiq’s system architecture is designed to maximize security and performance by dividing duties among dedicated contracts and carefully coordinating them through well-defined interfaces and roles.

1 Like

kHYPE Risk Assessment by BA Labs

Part II: Technical Risk

A. Smart Contract Risk

Developing a liquid staking protocol like Kinetiq entails significant smart contract risk due to the complexity of stake management, cross-chain operations, and reward accounting. The Kinetiq contracts underwent multiple independent security audits to uncover and mitigate these risks (summary in Table 1 below). Each audit identified various vulnerabilities ranging from critical math/accounting bugs to minor optimizations, most of which were subsequently fixed prior to mainnet launch.

Table 1: Audit Findings Summary for Kinetiq (kHYPE)

Audit (Date) High Risk Medium Risk Low Risk Informational/Gas
Spearbit Guild (Mar 2025) 2 findings (2 fixed) 7 findings (5 fixed, 2 acknowledged) 5 findings (3 fixed, 2 acknowledged) 16 Informational (10 fixed, 6 ack.) & 8 Gas (7 fixed)
Pashov Audit Group (Feb 2025) 5 High (all critical issues fixed) 6 Medium (fixed/mitigated) + Several low-severity findings (Full report published)
Zenith Team Audit (Apr 2025) – (details private) (Audit by Code4rena Zenith program; issues resolved)
Code4rena Contest (Apr 2025) 3 High (all fixed) 5 Medium (all fixed) 19 Low/Non-critical (All resolved pre-launch)

(Table 1: Summary of security audit findings for Kinetiq’s smart contracts. “Acknowledged” means the team recognized the issue but opted not to change code, judging it low impact or planning to address later. Zenith’s detailed numbers were not publicly disclosed, but its audit also led to fixes.)

From these audits, the major smart contract risks identified and addressed include:

  • Arithmetic Precision and Accounting Errors: One critical issue uncovered was the risk of latent insolvency due to precision truncation. Hyperliquid’s L1 HYPE uses 8 decimal places, whereas the EVM side uses 18 decimals. In the original design, when users deposited amounts that weren’t an exact multiple of 1e10 (to align 18→8 decimals), the excess tiny fraction of HYPE would be silently lost during the transfer to L1, yet kHYPE was still minted for the full amount. Over multiple operations this could accumulate an unrecoverable discrepancy where total kHYPE claims exceeded actual HYPE staked, potentially causing withdrawal failures (insolvency). This bug, flagged as High severity, was fixed by altering the accounting to use the truncated amount for minting (so kHYPE supply always reflects actual staked HYPE). It highlights the risk of subtle math errors in cross-decimal or cross-chain token handling. Another example was buffer accounting: the Code4rena audit found that some HYPE could get “stuck” in the StakingManager buffer without being either staked or withdrawable (a logic bug in how the buffer was utilized). This would effectively lock user funds in the contract. It was categorized as High (C4 [H-01]) and presumably fixed by adjusting the withdrawal logic to always consider any buffered HYPE for fulfilling requests. These issues underscore that maintaining precise accounting of staked assets vs. token supply is paramount – any mismatch can lead to insolvency or stuck funds. Kinetiq’s extensive testing and auditors’ independent reproduction of scenarios (e.g., tests for partial unstake and restake sequences) helped resolve these flaws.

  • Upgradeability and Initialization Risks: Kinetiq uses upgradeable proxies for core contracts, which introduces risk if constructors or initializers are misused. Spearbit auditors discovered that some state variables were being initialized in the implementation contract rather than via the proxy initializer, meaning those values (e.g., unstakeFeeRate) were never actually set in the proxy’s storage. This would have left critical parameters at default (zero) – for instance, the unstaking fee would always be 0% if not properly initialized. Such an issue could cripple the protocol’s intended fee mechanism and was a symptom of the complexity of proxy patterns. The fix was to move those initializations into the dedicated initialize() function and ensure only constants remain in declarations. This class of risk means that contract upgrades must be handled with extreme care – storage alignment, initial values, and access controls on upgrade functions are all potential pitfalls. If the admin (the Kinetiq team multi-sig) ever mis-executes an upgrade or if an unauthorized upgrade occurs (through key compromise), it could change the logic of Kinetiq entirely. Mitigations include timelocks or multi-sig requirements on upgrades and having monitoring to detect any unauthorized implementation changes. Kinetiq’s audits did not indicate malicious upgrade issues, but they did ensure the upgrade pattern was implemented correctly after initial oversights.

  • Withdrawal Logic and Queue Risks: Because Kinetiq must enforce fairness and avoid abuse in withdrawals, the contracts had complex conditions (like cooldowns, queue ordering, partial withdrawals). One identified risk was the now-removed cooldown mechanism. Originally, after staking, a user had to wait a set time before withdrawing; however, auditors noted a user could transfer their kHYPE to a fresh address and immediately withdraw, bypassing the restriction. This made the cooldown ineffective against a determined user and also the cooldown applied to the entire balance (penalizing even long-held tokens if any new stake was added). Kinetiq chose to eliminate this feature entirely, simplifying the model. Another risk scenario relates to withdrawal ordering under slashing: the Code4rena audit highlighted that if a slashing event occurred, users who had queued withdrawals before the event might get an unfair advantage (withdrawing as if no slash occurred), whereas those after would bear the full loss, leading to an imbalance. This was labeled [H-02] in the contest. Handling slashing fairly is tricky – one fix is to propagate slashing proportionally to all queued withdrawals (perhaps by adjusting the exchange rate retroactively for pending requests), or to enforce that all withdrawals in the queue are processed after applying the slash. Ensuring that no user can “escape” losses that others have to bear is crucial to prevent race conditions around slashing events. Kinetiq’s design now accounts for slashes via the Oracle reports so that the exchange rate (and thus all kHYPE holders) are uniformly affected at the time of slashing. Another related risk is partial withdrawal processing vs. deposits. A Medium finding ([M-04]) noted that if the system processed all pending withdrawals before deposits in a batch cycle, it might leave some deposits unstaked (sitting idle) until next round, or vice versa. While not catastrophic, it could lead to user confusion or slight yield loss. The code has been adjusted to a balanced processing order (likely alternating or netting deposits and withdrawals) to optimize fund use. Overall, withdrawal logic must carefully guard against DoS attacks as well – e.g., one audit finding ([M-05]) pointed out that an attacker could submit a huge number of tiny deposit transactions to clog the queue and cause others’ operations to be delayed. The mitigation is to introduce gas or block limits to the processing loop and possibly rate-limit or batch small deposits. In short, Kinetiq’s withdrawal system after audits is more robust: it removes unnecessary restrictions, plugs bypasses, and is designed to handle extreme scenarios without failing or unfairness.

  • Access Control and Role Risks: The audits also examined who can call what functions. One Spearbit medium finding discovered that a newly introduced ValidatorSanityChecker had setter functions that anyone could call, which would allow an attacker to change tolerance bounds and thereby force the system to reject all oracle reports (a form of griefing attack). This was a straightforward oversight – those functions needed onlyOwner or similar restrictions. It was promptly fixed by restricting access. This exemplifies the risk of privilege escalation or missing access guards in complex systems; a single unprotected function could sabotage the protocol’s operations. The Code4rena low-severity findings also flagged various missing or misused access controls, such as a missing whenNotPaused modifier on the kHYPE mint function (which could allow minting even during a paused state), or the lack of an easy way to remove old validators (which could become a governance issue). While these are minor, if left unchecked they can accumulate tech debt or even open attack vectors. For instance, if mint could be called while paused, an attacker might attempt to exploit a paused state to mint themselves tokens (if they had some role). Kinetiq has resolved these by aligning all functions with the intended access pattern. Nonetheless, one overarching risk remains: the admin roles themselves. The Kinetiq team (or governance) holds keys that can pause the system, upgrade contracts, and potentially even move funds (depending on implementation). These keys must be secure (multi-sig, hardware security, etc.) or the system could be rug-pulled or arbitrarily changed. This is a social/vector: not a flaw in code, but a critical aspect of risk – a malicious insider or external hacker with access to the admin key is an existential threat. Many LST protocols mitigate this by timelocks on upgrades or even plans to progressively decentralize the control. As of launch, Kinetiq is likely using a multi-sig and is a part of the Secure Staking Alliance, indicating a commitment to best practices in key management.

  • Oracle and Validator Selection Risks: The correctness of Kinetiq’s oracle system is vital. If the oracle reports incorrect data (due to error or manipulation), the protocol might delegate stake suboptimally or even lose funds. For example, Pashov’s audit reported an issue (H-01) that _addRebalanceRequest could use an outdated balance when trying to move stake, meaning if a validator’s true balance changed (due to rewards or slashing) since last update, the rebalance logic might fail or withdraw too much. Such timing issues highlight that the oracle updates and rebalancing functions must be tightly integrated – likely Kinetiq ensures that each rebalance uses fresh oracle data, or it recalculates balances on-chain before executing. Another Pashov finding (H-02) noted that in the deactivateValidator function, they initially missed withdrawing the validator’s funds on deactivation, meaning a validator could be marked inactive but its stake left behind earning nothing (funds stuck until manually recovered). This was fixed, but it shows how edge-case handling (like removing a validator) can lead to loss if not done atomically. The validator selection algorithm itself, while off-chain, has implications for contract risk: if it were exploited (e.g., feeding a fake high performance for a malicious validator), Kinetiq might shift a large amount of HYPE to a bad actor’s node which then gets slashed or fails. The contracts do have slashing tolerance checks and would catch some anomalies, but a subtle manipulation might not trigger sanity checks if within expected bounds. This is partly an off-chain risk (covered below), but the contracts are the last line of defense – they must gracefully handle unexpected oracle inputs. The presence of multiple oracle sources and a challenge period mitigates this risk by requiring consensus on data.

In summary, the smart contract risks of kHYPE have been significantly reduced through a rigorous audit process. Many potential failure modes – insolvency, stuck funds, reentrancy (interestingly, none of the audits explicitly reported a classic reentrancy, likely because functions like confirmWithdrawal use Checks-Effects-Interactions properly and only transfer to user at the very end) – were identified and addressed. The current contracts enforce invariant checks such as: total kHYPE * exchange_rate == total HYPE (within rounding error), only authorized can mint/burn, withdrawals cannot exceed available HYPE, oracle updates within bounds, etc. However, residual risks remain: the code is complex (over 1,300 lines of Solidity across 8 contracts), which always leaves room for undiscovered bugs. The use of upgradeable contracts means the security is also only as good as the next implementation – future changes could introduce new vulnerabilities if not carefully audited. The team’s approach of engaging multiple auditors (Spearbit, Pashov, Zenith, Code4rena, etc.) is a strong positive indicator, as it ensures diverse expert perspectives. We can conclude that smart contract risk, while not eliminated, is actively managed and reasonably low given the thorough audits and fixes. The most critical functions – stake, unstake, reward accounting – have been vetted and refactored to prevent known failure modes. Users must still trust that no unknown critical bug remains (a non-zero possibility in any complex DeFi contract), but Kinetiq has an emergency pause as a safety net and presumably has plans for an insurance fund or coverage (the treasury could act as a backstop for minor shortfalls, for instance).

Finally, one should note composability risk: as kHYPE becomes integrated into other DeFi protocols, a bug in Kinetiq could cascade (e.g., if kHYPE price depegs due to a flaw, protocols accepting it as collateral might become under-collateralized). This systemic risk is inherent in DeFi. Kinetiq’s strategy to mitigate that is to be extremely transparent and proactive about security – publishing all audits and even joining alliances to promote best practices. As the protocol hasn’t launched at the time of writing (mainnet launch is pending final security assurances), the contract risk will further be “battle-tested” once real funds are at stake. Continuous monitoring (as described) and possibly on-chain insurance modules could further reduce the impact if something goes away.

B. Off-Chain Component Risk

Off-chain components, while not subject to smart contract bugs, introduce their own risk vectors for kHYPE and the Kinetiq protocol. These include risks from oracle infrastructure, validator selection algorithms, key management, and external dependencies. We outline the major off-chain risks and how the protocol mitigates them:

  1. Oracle Manipulation or Failure: The OracleManager relies on off-chain agents to feed it accurate data on validator performance, rewards, and slashes. This creates a classic oracle problem: if the data source is wrong or compromised, the on-chain decisions will also be wrong. A malicious oracle feeder could, for example, report that a honest validator has very low performance or was slashed when it wasn’t, causing the ValidatorManager to unnecessary reallocate stake or even withdraw from a perfectly good validator, reducing yield. Conversely, it could fail to report a slash, causing the exchange rate to be temporarily overstated (until detected), which might let some users withdraw slightly more HYPE than they should (leaving a deficit for remaining users). The risk is mitigated by several measures. First, Kinetiq likely uses multiple data sources and cross-verification – e.g., the StakeHub scoring might be run by independent parties (or at least by robust servers) and the OracleManager can aggregate data. Second, the design includes a challenge period and sanity checks: if one oracle report is out-of-line, there may be time for a second report or a hardcoded sanity checker (ValidatorSanityChecker) to reject it. For example, the SanityChecker enforces that reported uptime or reward values are within plausible ranges, and anyone can call it to validate the oracle input. However, if all oracles collude or the primary oracle key is hacked, this system could still be fed incorrect data. The impact could be severe – delegations might swing to malicious validators or the protocol might incorrectly slash itself. The Secure Staking Alliance (SSA) and Kinetiq’s partnership with Anthias (risk consulting) indicate that they’ve designed the oracle process with high integrity in mind, possibly having out-of-band monitoring (manual oversight) as well. Nonetheless, oracle risk is not zero. If the Kinetiq oracle infrastructure went offline (due to server failure or attacks like DDoS), the on-chain contracts would stop receiving updates. In the short term this is safe (the contracts will just use last known good data), but over a longer period, not having updates could halt reward distribution and skew validator allocations. The protocol’s MAX_ORACLE_STALENESS might trigger a pause or at least flash a warning on-chain if no update comes after, say, X hours. The Kinetiq team presumably has redundancy (multiple servers, fallback providers) to ensure continuity. In summary, oracle risk is mitigated but remains one of the larger off-chain risks – requiring trust in Kinetiq’s data pipeline. Community decentralization (allowing others to run oracle feeders or verify data) could further reduce this risk over time.

  2. Validator Selection & Delegation Risks: The off-chain algorithm (StakeHub) that scores and selects validators is proprietary to Kinetiq’s approach. If this algorithm has flaws or if an attacker can game it, the stake may be misallocated. For instance, consider an attacker who runs a validator: they might attempt to trick the scoring by momentarily boosting their performance metrics (maybe by architecting tests that the algorithm measures, or by being honest until they receive a large delegation and then turning malicious). If the algorithm overly trusts recent performance or has a narrow set of metrics, a cunning validator could exploit that, win a top score, get Kinetiq’s delegation, and then either underperform (stealing would be difficult without slashing themselves, but they could, say, time a self-slash to harm Kinetiq or cause volatility). This is somewhat analogous to how in some DeFi, if you game a ranking you get an outsized reward. Kinetiq mitigates this by using multiple metrics and historical data – e.g., requiring sustained good performance and including decentralization metrics (not delegating everything to one validator). The risk of centralization is also there: the algorithm might inadvertently delegate too much to a single validator if it’s far superior, creating a single point of failure. Kinetiq’s site explicitly mentions focusing on decentralization metrics, so the algorithm likely has caps (e.g., never delegate more than N% to one). Off-chain, there’s also a governance risk: who updates or fine-tunes this algorithm? If down the line, governance token holders or the team decide to bias the criteria (for example, favor their friends’ validators), it could deviate from being meritocratic and introduce economic centralization or corruption. While speculative, it’s a risk to monitor in any “algorithmic delegation” system. Lastly, stake movement between validators must be orchestrated correctly off-chain. If Kinetiq’s off-chain controller (likely a script that calls rebalance() functions) fails, the system might not move stake when it should, causing suboptimal yields or even liquidity problems. For example, if one validator accumulates a lot of rewards and another doesn’t, users in one StakingManager could theoretically face a shortfall as described in an audit. The mitigation was human/off-chain intervention to add liquidity, but that requires vigilance. So, the risk is that operational lapses off-chain (not running a rebalancing cron, or an institutional partner not following instructions) could lead to temporary imbalances or user impact. Kinetiq’s decision to not mix institutional and public pools (to avoid cross-pool issues) simplifies some of this, but within each pool the off-chain operations must still be executed promptly. We expect Kinetiq to run these processes automatically at defined intervals, with alerts if any transaction fails.

  3. Key Management & Admin Control: As touched on, the admin keys (multisig) controlling upgrades and pause are an off-chain risk. If the multisig’s private keys were compromised (through phishing, insider threat, etc.), an attacker could deploy a malicious upgrade to the Kinetiq contracts. For example, upgrade the StakingManager implementation to one that allows anyone (or just the attacker) to withdraw all staked HYPE to an arbitrary address. This would be catastrophic – effectively a theft of all user funds – and on-chain monitoring might not prevent it if done swiftly. To mitigate this, projects often use multisignature wallets with 4-of-6 or similar setups, require security hardware for approvals, and may even timelock upgrades (so users have time to react if an unexpected upgrade is posted). It’s not publicly known what threshold or timelock Kinetiq uses, but given their collaboration with well-known auditors and alliances, one can assume a high standard (e.g., a 3/5 or 4/7 multisig with core team and possibly third-party signers). Another key management risk is the Oracle private key – if the key that signs oracle data (if any authentication is used) is leaked, an attacker could feed bogus data as mentioned. Also, any infrastructure credentials (servers, cloud accounts) are sensitive – a hacker who gained access could potentially shut down or corrupt the off-chain services. Kinetiq must secure these with best practices (firewalls, 2FA, monitoring access logs). In summary, trust in the team and their operational security is part of the risk model. Users are implicitly trusting Kinetiq not just to write safe contracts but to safeguard the keys that control those contracts.

  4. Dependency on Hyperliquid Network: Being a staking protocol, Kinetiq is inherently tied to Hyperliquid’s health and rules. If something goes wrong at the consensus layer – e.g., a bug in Hyperliquid’s code, a chain halt, extreme volatility in HYPE’s value, or governance changes in Hyperliquid’s economics – Kinetiq and kHYPE are directly affected. For instance, if Hyperliquid governance decided to change the unbonding period or slashing conditions, Kinetiq’s off-chain processes would need to adapt immediately or face mis-synchronization (imagine Kinetiq thinking unbonding takes 1 day, but it was changed to 7 days – users would expect withdrawals sooner than possible). If Hyperliquid suffered a security incident (like BFT consensus failure or major validator collusion), Kinetiq’s staked assets could be at risk of slashing or even loss if the chain had to rollback. These are broad ecosystem risks but worth noting – Kinetiq’s risk cannot be lower than the underlying chain’s risk. Mitigation is mostly out of Kinetiq’s hands; however, being an active participant (they run autonomous validator selection and likely are plugged into Hyperliquid’s community) they can at least have early warning of any issues. Additionally, if Hyperliquid’s tokenomics fail (say the price of HYPE crashes drastically or liquidity dries up), kHYPE could face a depeg situation where it trades at a discount due to fear of being unable to redeem (similar to how stETH traded below ETH during crises). Off-chain sentiment and market factors thus influence kHYPE’s risk. Kinetiq might consider an insurance fund or guarantees to cover slashing events – some LSD protocols self-insure small slashes via their treasuries to maintain user confidence. It’s not stated if Kinetiq will do so, but their treasury from fees could potentially serve that purpose if needed. Kinetiq has been pretty active in conducting various auditing competitions hence there could be a chance that Kinetiq might also consider having a bug bounty program.

  5. External Integrations Risk: Once launched, kHYPE will likely be integrated in various DeFi platforms (Dex pools, lending markets, etc.). While this is mostly a risk for those platforms, it feeds back into Kinetiq: if, say, a lending platform handling kHYPE is hacked, the attacker might dump a huge amount of kHYPE, causing its market price to tank well below the value of underlying HYPE (even if fully backed). This could trigger an arbitrage wave of redemptions that stress-tests Kinetiq’s withdrawal system (a so-called bank run scenario). Off-chain, Kinetiq needs to be prepared for such scenarios by perhaps monitoring secondary markets for kHYPE price deviations and ensuring sufficient HYPE buffer or communication with validators to speed up unbonding if needed. This is more of an economic risk than a technical one, but the line can blur (e.g., if an oracle used kHYPE market price for anything, a manipulated price could confuse the system – though currently, no indication of using external price oracles in protocol).

In conclusion, the off-chain and external risks of kHYPE center on data integrity, operational security, and dependency on the Hyperliquid ecosystem. Kinetiq has addressed many of these by design: the oracle is guarded by sanity checks, the validator selection aims to be manipulation-resistant (using diverse metrics), and the team appears security-conscious (multiple audits, alliances). However, users should recognize that liquid staking is not trustless – unlike a simple self-custody of coins, here one trusts the protocol’s off-chain logic and governance. Kinetiq’s multi-auditor strategy and not launching until thorough testing is a sign of a mature risk mitigation approach. As the protocol goes live, real-world testing will further validate the off-chain processes (e.g., oracle updates arriving regularly, quick response to any validator incidents).

One way Kinetiq could further reduce off-chain risk in the future is to decentralize the oracle updates – for instance, having a network of independent “oracle reporters” or using the blockchain’s own consensus data if available (on some PoS chains, performance stats might be on-chain). Another way is to implement a circuit breaker: if a certain percentage of stake gets slashed or if oracle reports extreme values, automatically pause the protocol (to protect from cascading failures). These are speculative features that could be added. For now, based on available information, Kinetiq’s off-chain risk is managed but not negligible. The protocol’s safety ultimately relies on the competence and honesty of the team operating these off-chain components. Given that Kinetiq is backed by experienced researchers and audited by respected entities, the community has reason for confidence. Nonetheless, continuous diligence is required: any lapse in running the off-chain services or any breach can directly translate to on-chain impact (loss of yield or in worst cases loss of funds). Users and integrators should stay informed via Kinetiq’s communications (Discord, Twitter) for any incidents or updates.

In summary, technical risks for kHYPE have been extensively analyzed and mitigated where possible. Smart contract risks have been reduced through comprehensive audits and a sound architectural design emphasizing modularity and emergency controls. Off-chain risks are acknowledged and addressed via robust oracle mechanisms and operational security, though they inherently require trust. As with any nascent DeFi protocol, ongoing vigilance and possibly additional safeguards (insurance, decentralization of oracles, etc.) will further strengthen kHYPE’s risk profile. If all components function as designed, kHYPE offers a technically sound liquid staking solution on Hyperliquid, marrying high yield with security – but it will be the real-world performance and incident response that ultimately prove the resilience of the design.

1 Like

wstHYPE Risk Assessment by BA Labs

Part I: Protocol Overview

A.1 Protocol Details

A.1.1 Underlying Collateral

The wstHYPE protocol is a liquid staking system for the Hyperliquid network’s native token HYPE. Users deposit HYPE into the protocol’s smart contracts and receive stHYPE (staked HYPE) tokens in return at a 1:1 exchange rate. The deposited HYPE serves as underlying collateral by being staked to Hyperliquid validators via Valantis’s infrastructure. These validators participate in block production on Hyperliquid (also known as HyperCore), earning staking rewards (currently around 2.2% APY base) which accrue to stHYPE holders. In effect, each stHYPE token is fully backed by an equivalent amount of HYPE token that is actively staked on Hyperliquid’s Proof-of-Stake network. At any time, the total supply of stHYPE is intended to be matched by an equal amount of HYPE held in the staking contracts or in the process of being unstaked, ensuring a 1:1 collateralization.

Critically, stHYPE uses Hyperliquid’s native staking interface under the hood. The protocol stakes user-deposited HYPE to a distributed set of professional validators (selected by Valantis) to maximize uptime and performance. By spreading stake across multiple validators and using only high-performance nodes, the protocol mitigates validator downtime or slash events. In addition to standard validator staking, a portion of the collateral may be allocated to specialized staking accounts on Hyperliquid that enable additional yield (discussed below). Regardless of where it is allocated, all underlying HYPE remains within Hyperliquid’s staking system – there is no lending or usage outside of the Hyperliquid chain. This design confines collateral to a single ecosystem, reducing cross-chain risks.

Because the underlying collateral is on Hyperliquid rather than a more familiar chain like Ethereum, users implicitly rely on the security and liveness of the Hyperliquid network. HYPE staked via stHYPE is subject to Hyperliquid’s validator slashing conditions: if a validator misbehaves or the network enforces penalties (e.g. for downtime or double-signing), the staked HYPE can be slashed. The stHYPE protocol attempts to mitigate this by choosing reputable validators and by not over-concentrating stake in any single node. Still, users bear the underlying collateral risk of the Hyperliquid network’s consensus mechanism. In summary, wstHYPE’s collateral model wraps native HYPE stake – giving users a liquid token (stHYPE/wstHYPE) that represents a claim on staked HYPE – with the promise that each unit remains redeemable for 1 HYPE minus any network-level slashing or protocol fees.

A.1.2 Fee and Revenue Sources

The wstHYPE/stHYPE system generates yields from two primary sources: block rewards from staking, and partner protocol incentives. First, all staked HYPE earns the standard Hyperliquid protocol staking reward (~2.2% annually at present) as issued by the network to validators. These rewards automatically accrue to stHYPE holders by increasing the total supply of stHYPE over time (stHYPE is a rebasing token reflecting earned yield). In addition to the base layer rewards, Valantis has structured integrations that provide supplemental yield: specifically, stHYPE allocates portions of its underlying HYPE to external Hyperliquid protocols (via “Specialized Stake Accounts”) in exchange for fixed interest or profit-sharing. Currently, the two notable revenue-generating allocations are a credit line to HyENA (a Hyperliquid perpetual DEX by Ethena Labs) and to USDe (Ethena’s stablecoin quote asset). Together, these allocations of 700k HYPE provide additional HYPE-denominated rewards to stHYPE beyond the base 2.2% staking yield.

Concretely, Valantis extended 500,000 HYPE of the staking pool to HyENA’s exchange and 200,000 HYPE to the USDe stablecoin system in late 2025. In return, Ethena (the operator of HyENA and USDe) agreed to pay fixed yields on those stakes: the HyENA allocation yields an extra 2.0% APY (making the HYPE staked there earn ~4.2% total including base rewards), and the USDe allocation yields an extra 3.0% APY (~5.2% total). These partner-paid yields are distributed to all stHYPE holders, boosting the overall staking APY for the pool. This arrangement effectively monetizes the staked collateral by letting other protocols borrow HYPE stake (to meet Hyperliquid’s 1M HYPE exchange requirement or to collateralize a stablecoin) and pay interest for its use. The revenue sources for stHYPE holders, therefore, include: (a) network staking emissions, and (b) external protocol interest fees.

It’s important to note that Valantis Labs (the protocol operator) takes a protocol fee from the staking rewards as its revenue. The smart contracts implement a fee (in basis points) on yield, similar to how Lido or other liquid staking protocols charge a commission. The code uses a protocolFee parameter to skim a percentage of each reward rebase into a protocolPendingFee pool. This fee percentage is configurable by the protocol’s administrators and was initially set to a value (10% of positive yield) in line with industry standards (exact initial rate not explicitly stated in public docs). For example, if the total staking rewards for a period are 100 HYPE, a protocol fee of 10% would divert 10 HYPE to the Valantis treasury and 90 HYPE to stHYPE holders. All “Annual Protocol Rewards” reported to users are net of fees. This fee provides Valantis with an ongoing revenue stream to fund development, validator operations, and risk management. It also aligns incentives: the more value and yield the protocol generates, the more fee revenue Valantis earns, motivating them to maximize safe yields for stakers. In summary, stHYPE’s fee and revenue model is that users earn net yields from Hyperliquid and partner protocols, while Valantis earns a commission on those yields (but does not charge upfront minting or redemption fees).

A.1.3 Minting and Redemption Flow

Minting (Staking) Flow: Users acquire stHYPE by staking HYPE through the Valantis interface or directly via the smart contract (Overseer). When a user initiates a stake, HYPE tokens are transferred into the Overseer contract (the core contract managing the staking logic). The Overseer then mints an equivalent amount of stHYPE tokens to the user’s address. Under the hood, the new HYPE is immediately delegated across Hyperliquid validators or directed into one of the staking module accounts. The user’s balance of stHYPE represents their share of the staking pool. Because stHYPE is a rebasing token (its balance can increase as rewards accrue), users see their stHYPE count grow over time, or equivalently, the exchange rate HYPE:stHYPE grows. Notably, Valantis employs a dual-token model: users actually have the option to hold a wrapped version wstHYPE which is non-rebasing. However, the system automatically keeps stHYPE and wstHYPE in sync; whenever stHYPE is minted or burned, corresponding wstHYPE shares are created or destroyed so that both tokens represent claims on the same pool. This means users can interact with either interface – stHYPE (rebasing) for simplicity or wstHYPE (fixed-supply) for DeFi integrations – without needing to manually wrap/unwrap. The Overseer contract coordinates this by minting/burning stHYPE and adjusting wstHYPE share balances simultaneously.

Redemption (Unstaking) Flow: To redeem underlying HYPE, a user burns their stHYPE/wstHYPE in exchange for native HYPE. This can be done by calling the burn function on the Overseer or a combined burnAndRedeemIfPossible function. Redemptions are subject to the Hyperliquid unbonding period and the protocol’s liquidity constraints. Hyperliquid imposes a base 7-day unstaking period for validators (plus a 1-day redelegation delay). Therefore, if a user’s HYPE is actively staked, they must wait ~7 days after initiating withdrawal to receive their tokens. The stHYPE design, however, uses multiple strategies to accelerate withdrawalsfor most users:

  • The protocol maintains five standard staking modules (separate HyperCore accounts) to manage unbonding in a staggered fashion. These allow up to 25 unbonding operations to run concurrently (5 per module), effectively increasing throughput of withdrawals. By splitting stake across 5 accounts, stHYPE can process more simultaneous withdrawal requests without hitting single-account limits.
  • The protocol also enforces that any new incoming stake first services pending withdrawals before being staked anew. This means if there is a queue of users waiting for HYPE out, and another user stakes HYPE, the new HYPE will be used to immediately fill withdrawal requests (paying out those users instantly) instead of going to a validator. This mechanism (similar to Lido’s “buffer”) ensures organic inflows provide instant liquidity to exiting users, reducing wait times in normal conditions.
  • If there is insufficient new stake and the withdrawal queue exceeds what the 5 standard modules can handle in the 7-day window, then withdrawals may tap into the non-standard (specialized) stake accounts. These are the portions of stake locked in HyENA and USDe arrangements, which have custom withdrawal conditions (in this case, a maximum 90-day lock). By design, stHYPE will only pull from these long-term allocations as a last resort: normal user exits are handled from the liquid staking modules first, targeting that all redemptions complete in ~7.2 days on average. If an extreme scenario forces using the locked stake, those users might wait up to 90 days, but several failsafes exist (described below in Withdraw Cooldowns).

In practice, a user calling redeem will receive HYPE immediately if the protocol’s liquid reserves are sufficient (including any new stakes coming in). If not, their request is queued and served once the next unbonding operation completes. The Overseer contract tracks all pending burns and completed redemptions, ensuring that each stHYPE burned eventually corresponds to HYPE returned. There is logic to prevent redeeming more than the contract’s available balance at any time. The redemption process is thus a combination of on-chain accounting and off-chain waiting: the user’s stHYPE is burned (reducing the supply), and after the waiting period, the underlying HYPE (which has been unbonded from validators) becomes claimable to the user. Valantis has modeled this process to guarantee that even in worst-case mass withdrawals, the system can orderly unwind all positions given enough time. Notably, if a severe run on the protocol occurs, Valantis can initiate early exits from HyENA/USDe stakes once the circulating stHYPE supply drops below a threshold (1M HYPE). This managed liquidity buffer ensures that as TVL falls, the protocol proactively frees up the locked stake (over 90 days) so that it will be available to satisfy withdrawals by the time the normal liquidity is exhausted. In summary, the mint/burn flow is designed to give a user experience of near-instant liquidity under most conditions, while the actual unbonding happens in the background, managed by the protocol’s multi-account architecture.

A.1.4 Fee Structure and Incentives

The fee structure for users of wstHYPE is straightforward: there are no direct fees to mint or redeem beyond gas costs, but there is a continuous protocol fee on staking rewards. As discussed, Valantis charges a commission (configured via protocolFee in basis points) on the yield earned. This fee is taken before rewards are distributed, effectively reducing the APY that stHYPE holders receive. For example, if Hyperliquid’s base yield is 2.2% and external sources add another 2%, the gross 4.2% would be subject to the protocol fee. If the fee were, say, 10%, the net APY to users would be ~3.78% and ~0.42% (10% of 4.2) would accrue to the protocol. The documentation emphasizes that all reported reward metrics (daily and annual rewards) are net of fees so that users see the actual yield they keep. This aligns with industry practices (Lido, for instance, charges a 10% fee on Ethereum staking rewards). Valantis can adjust the fee via governance/administration (up to a maximum of 100% in the code, though such extremes are unlikely). This adjustable fee means incentives can be realigned over time – for example, reducing the fee to stay competitive if other liquid staking tokens emerge on Hyperliquid, or increasing it modestly to fund new features.

From an incentive alignment perspective, the design ensures that both users and the protocol benefit from maximizing secure yield. Users are incented to stake HYPE because stHYPE offers not just base staking returns but also extra yield from the HyENA and USDe integrations, boosting their ROI. These boosted yields (4.2% and 5.2% on those tranches) make stHYPE significantly more attractive than simply staking HYPE directly (2.2%). This encourages adoption of stHYPE and deepens its liquidity. Meanwhile, Valantis is incented to carefully manage these external allocations: the protocol fee applies to the additional yield as well, so Valantis earns more by keeping the HyENA and other partnerships running smoothly and profitably. It also discourages reckless behavior – if an external allocation carried too much risk of slashing or default, it could ultimately harm Valantis’s revenue (since a slashing event would reduce total staked HYPE and future yield). Thus, Valantis has reason to vet partners and perhaps require safety measures like being a multisig signer (as with HyENA) to protect the stakers’ funds.

The incentive structure also includes mechanisms to maintain liquidity. If stHYPE’s total value locked (TVL) declines (e.g. many withdrawals), the protocol’s average yield to remaining stakers increases because the fixed external yields (on 700k HYPE) become a larger proportional reward for a now smaller pool. This variable APY acts as a market-driven incentive: a high yield when TVL is low will attract new stakers to join (or existing ones to stay), restoring equilibrium and preventing a “bank run” situation. In other words, the yield automatically adjusts in favor of remaining participants if others leave, which is a clever incentive to stabilize the system. Furthermore, since stHYPE tokens are liquid, arbitrageurs and traders can also help maintain the peg (1 stHYPE ≈ 1 HYPE) by trading on secondary markets if it deviates, which indirectly incentivizes keeping confidence in the system. The protocol does not currently have additional token incentives (e.g., liquidity mining) for stHYPE, focusing instead on organic yield incentives. In summary, the fee structure (a performance fee on yields) and the integrated yield sources align Valantis’s interests with users’: both seek a high, sustainable HYPE yield, and the system’s design rewards participants while funding the protocol from the generated yield itself.

A.1.5 Off-chain Components

While the core mint/burn logic of wstHYPE is on-chain, several off-chain components and agreements underpin the system. Chief among these is the credit line extended to the HyENA protocol (and Ethena’s USDe) as mentioned earlier. This arrangement goes beyond a simple smart contract call – it involved a formal partnership between Valantis Labs (Panama) and HyENA, the developer of HyENA HIP-3 markets (the partnership is not directly with the Ethena entity). In essence, Valantis agreed to “lease” 700k HYPE of stakers’ funds to Ethena’s protocols, enabling HyENA to meet Hyperliquid’s requirement of 1M HYPE stake for operating a HIP-3 exchange without having to own all that HYPE outright. In return, HyENA contractually commits to paying the fixed 2% and 3% yields on the respective tranches and to abide by withdrawal constraints (e.g. giving HYPE back within 90 days of notice). This is effectively an off-chain legal agreement (a “Stake Agreement”) complementing the on-chain Stake Marketplace module. Valantis Labs is incorporated in Panama, and likely the agreement with Ethena (HyENA) is governed by applicable contract law – providing a legal recourse if, for instance, HyENA failed to pay the promised yield or return the stake when required.

To provide additional assurance and oversight, Valantis is a signer on HyENA’s multisig (the multi-signature admin wallet that controls the HyENA exchange parameters on Hyperliquid). By holding a key on that multisig, Valantis gains some control or at least veto power in HyENA’s governance, ensuring that the staked HYPE is not misused or suddenly withdrawn contrary to agreement. This off-chain (or cross-protocol) governance role is an important risk mitigator: it ties the success of HyENA to Valantis’s involvement and allows Valantis to monitor HyENA’s solvency and operations directly. In effect, Valantis becomes a stakeholder in HyENA, aligning HyENA’s team to act prudently with the borrowed stake because any drastic action would require Valantis’s signature approval. It’s a form of corporate governance interlock that is not visible on the stHYPE smart contracts, but crucial for managing risk of the credit line.

Aside from the HyENA/Ethena partnership, other off-chain components include the Valantis operations team and multisig that manage upgrades, parameter changes, and validator selection. The protocol’s smart contracts have admin roles (governance roles) assigned likely to a Valantis-controlled multisig wallet. For example, changing the protocolFee or pausing the contract requires an off-chain decision by the team followed by an on-chain multisig transaction. The Valantis multisig thus acts as a critical off-chain component – if compromised, it could potentially change contract logic or withdraw fees. However, being a multisig (typically requiring M-of-N signatures) reduces the risk of a single key failure. The documentation and Terms indicate Valantis’s legal entity is the accountable party for managing these keys. Additionally, Valantis likely runs the validator infrastructure that the HYPE is staked to. Operating validators is an off-chain service (server nodes) that if failed could impact yields or cause slashing. Valantis’s role in running or selecting validators means users trust Valantis off-chain processes to maintain nodes properly.

Finally, data and transparency around stHYPE’s operations rely on off-chain services. While on-chain data can show total supply and some parameters, Valantis provides documentation and possibly a web dashboard (previously via stakedhype.fi) to break down how funds are allocated (e.g. how much in HyENA vs validators) and performance metrics. Maintaining this transparency is an off-chain responsibility – e.g., publishing updates if the allocation changes or if new stake agreements are entered. The legal Terms of Service and any insurance or backstop commitments (if they exist) are also off-chain. To date, no insurer or slashing cover is mentioned, but if Valantis were to socialize losses from slashing, that too would be an off-chain commitment. In summary, the off-chain components include legal agreements (Valantis–Ethena), governance roles (Valantis multisig control, HyENA multisig participation), validator operations, and informational outlets (dashboards/docs). These components are essential to wstHYPE’s functioning but also introduce trust dependencies beyond the code itself.

B. System Design

B.1 Monitoring and Observability

A critical aspect of any staking protocol is transparency into its operations. In the case of wstHYPE/stHYPE, the monitoring and observability of the underlying HYPE and its usage is only partially exposed to the public. On one hand, Valantis provides documentation on how staked HYPE is allocated – for example, stating that 5 accounts are used for normal staking and 2 specialized accounts (HyENA and USDe) hold 700k HYPE. The “Stake Accounts” section of Valantis’s docs outlines the design and even gives the fixed yields and withdrawal locks for those external allocations. This indicates a degree of openness about the strategy. Additionally, on-chain, one can in principle track the stHYPE total supply and the balances of the known contract addresses (Overseer, staking modules, etc.). Hyperliquid’s blockchain would show the stake accounts and any slashing events, so technically a determined analyst could monitor the state. Valantis has encouraged understanding these mechanisms and even linked to Hyperliquid’s official docs on slashing risks, which suggests they expect users to be aware of these details.

However, from an end-user perspective, the data is relatively opaque. There is currently no real-time public dashboard (as of late 2025) that shows exactly how much HYPE is staked in each validator or exactly which addresses correspond to the HyENA and USDe allocations. Unlike some Ethereum staking providers that have transparency dashboards (showing validator lists, performance, etc.), Valantis’s approach has been to summarize this in documentation. For example, the “Transparency and Risks” page provides a high-level overview of how the system works and points to the Stake Accounts breakdown, but it does not give a live feed of metrics. The old stHYPE site had an “Analytics” page describing metrics like total supply and net APR, implying a UI was available to see daily rewards, but since the acquisition, this information may only be accessible through community tools or block explorers. There have been community posts and tweets indicating that one can “track real-time yields, validator performance, uptime, commission rates, and much more” for stHYPE, but these likely refer to Hyperliquid’s explorer or API rather than a dedicated Valantis interface.

In practical terms, a stHYPE holder must trust Valantis’s reporting for details like how 700k HYPE is allocated or when exactly the 90-day lock countdown started for HyENA. Important information – such as whether HyENA has honored its profit share payments, or how close the protocol is to initiating a forced withdrawal from HyENA – is not immediately transparent on-chain to a casual user. Valantis might share such updates via social media or announcements. This relative opacity is a risk: users rely on Valantis to be truthful and timely in disclosing any issues (for instance, if HyENA were struggling to pay yields or if a portion of stake got slashed). The observability of validator performance is another angle – since stakers cannot choose or see all validators, they depend on Valantis to monitor validators off-chain. It’s mentioned that only top-performing validators are used, but if a validator’s performance degrades, users only find out if yields drop or if Valantis announces a change.

On the positive side, the architecture uses on-chain enforcement for critical pieces of data. The smart contracts themselves keep track of how many HYPE are pending withdrawal, how many are allocated to each module, etc. For example, the contract will not allow more stHYPE to be redeemed than there is HYPE available (it checks balances vs. burns). This means some integrity is programmatically ensured. Moreover, the specialized stake accounts operate under smart contract-defined withdrawal locks (via the Hyperliquid Stake Marketplace module), so one can observe on-chain that those HYPE cannot be withdrawn by Ethena until conditions are met. These aspects are transparent to those who inspect the chain or the code. In summary, public monitoring is limited – there is no simple dashboard showing, for instance, “X HYPE staked in validators A, B, C; Y HYPE in HyENA (day 30 of 90 lock)” – but the framework has some built-in transparency through code. Enhancing observability (e.g. publishing regular reports or on-chain signals of HyENA’s status) would improve user confidence. As it stands, stHYPE is more opaque than an on-chain-only protocol; much depends on Valantis’s off-chain disclosures and users’ ability to parse on-chain data from Hyperliquid’s network.

B.2 System Architecture

The wstHYPE/stHYPE system architecture consists of multiple smart contracts and modules working together, following a modular design with upgradability and role-based access control. At a high level, the core components are:

  • stHYPE Token Contract: a rebasing ERC-20 token representing staked HYPE. This contract likely uses an upgradeable proxy (address 0xfFaa4… on mainnet) and implements the logic for rebases (supply adjustments when rewards come in) and transfers. It works akin to an ERC-4626 vault share token, with some custom logic for the dual interface.
  • wstHYPE Token Contract: a wrapper around stHYPE that represents a fixed share of the pool. wstHYPE is implemented such that it reads the share balance from the stHYPE contract and can admin-transfer stHYPE balances. In practice, this means wstHYPE doesn’t maintain its own accounting of underlying except by referencing stHYPE’s state; when users transfer or trade wstHYPE, behind the scenes it adjusts their stHYPE balance accordingly. The wstHYPE contract (proxy at 0x94e8396…) ensures seamless integration with DeFi by providing a non-rebasing asset whose value increases over time relative to stHYPE.
  • Overseer Contract: the brain of the protocol, which orchestrates staking, unstaking, and reward distribution. The Overseer holds roles to mint and burn stHYPE, initiate stake to validators, manage the stake accounts, and handle user interactions like burnAndRedeemIfPossible. It likely also holds critical configuration like the protocolFee, references to addresses of stake modules, and pause functions. In the contract addresses, OverseerV1 is listed (proxy at 0xB96f07…), indicating an upgradeable Overseer.
  • HyperCore Staking Module Contracts: these are a set of contracts or contract-instances that interface directly with Hyperliquid’s validator staking. The audits reference HyperCoreStakingModule.sol and multiple accounts, which suggests that Valantis deployed several instances of a StakingModule contract, each controlling a Hyperliquid stake account (likely via a key or a dedicated account on Hyperliquid). These modules abstract the lower-level staking actions (delegate, undelegate) and feed data back to the Overseer. By having multiple instances (5 for standard operations), the system can parallelize unstakes.
  • External Stake Management Module: this is a specialized contract for managing the allocations to HyENA and USDe. The November 2025 Pashov audit specifically looked at StakingModuleExternalManagement.sol, indicating that a new module was introduced to handle external (non-validator) stake uses. This module likely implements the logic for the 90-day withdrawal locks, coordinating with Hyperliquid’s HIP-3 Stake Marketplace. It might escrow the 700k HYPE and enforce conditions: e.g., it will not allow withdrawal until a certain time or trigger (like stHYPE supply drop) occurs. The audit findings suggest it had functions like depositCap and amountDeposited to limit and track how much external stake is allocated.
  • Role-Based Access Control & Multisig: All contracts use OpenZeppelin-style access control (roles such as DEFAULT_ADMIN_ROLE, REBASER_ROLE, etc.). The admin roles are almost certainly held by a multisig controlled by Valantis. For example, the ability to call changeProtocolFee() or to pause operations lies with an address having DEFAULT_ADMIN_ROLE. Similarly, executing rebase() periodically might be restricted to a REBASER_ROLEthat Valantis’s backend or a keeper address holds. The use of a multisig for admin reduces single-point risk, and upgrades to the proxies would require multisig authorization.

The data flow in this architecture is as follows: when users stake, the Overseer updates internal records and calls the appropriate StakingModule to stake HYPE on Hyperliquid. The stHYPE total supply is increased via mint. The wstHYPE contract, which monitors stHYPE’s share supply, can be queried for an equivalent wstHYPE amount (though users can hold either token, the system keeps them 1:1 in value). As rewards accumulate on Hyperliquid (validators earning HYPE), those rewards are not automatically pushed on-chain to Ethereum; instead, a rebase event is triggered (either periodically or when someone calls it) to sync the increased balances. The rebase calculation pulls the new total HYPE across all modules (using HyperCoreUtils to sum balances perhaps) and compares with the previous supply. It then mints the difference as new stHYPE to the Overseer, which either distributes to users proportionally or keeps aside the protocol fee portion. The stHYPE contract’s balances increase (this is how a user’s wallet sees more stHYPE). For wstHYPE, since it’s just a wrapper, its conversion rate to stHYPE increases accordingly (1 wstHYPE now corresponds to slightly more HYPE than before the rebase). Throughout this, any transfers of stHYPE trigger corresponding updates in wstHYPE due to the dual-entry points design. Notably, this dual-token design required careful handling to avoid exploits (the docs mention a known vulnerability pattern if not handled), which the team mitigated by providing a setSelfDisableTransfer function to disable one token’s transfers if needed.

In terms of upgradeability and safety, the architecture is proxy-based with modular contracts, which means new versions of Overseer or modules can be deployed and the proxies pointed to them. This allows Valantis to patch issues (as they did following audits) but also introduces upgrade risk – a malicious or compromised admin could deploy a bad implementation. The Guardian audit in Nov 2025 gave a Very High Confidence rating to the code security, finding no critical flaws, which speaks to the soundness of the architecture as reviewed. The system architecture also integrates Hyperliquid’s CoreWriter/HIP-3 modules on-chain: essentially, Valantis’s contracts interact with Hyperliquid’s chain (HyperEVM) through special calls. For example, the Stake Marketplace functionality likely involves Valantis calling a Hyperliquid smart contract to negotiate stake agreements (those “Withdrawal Lock” contracts in HSM). This cross-chain/cross-contract interaction is a novel architectural element – effectively, stHYPE acts as an on-chain (Ethereum or HyperEVM) coordinator for activities happening in HyperCore.

To summarize, wstHYPE’s architecture is a combination of:

  • Token layer: stHYPE and wstHYPE tokens for user-facing balances.
  • Control layer: Overseer and modules (staking modules, external module) that handle logic and integrate with Hyperliquid.
  • Governance layer: Admin roles and multisig for upgrades and parameter tweaks.
  • External interfaces: connections to Hyperliquid’s validator system and stake marketplace (through which HyENA and others interact).

This design emphasizes modularity (different stake modules for different purposes), safety (upgradeable but audited contracts, limited roles, fallback disable functions), and flexibility (capable of adding new specialized modules for future integrations). It sacrifices some simplicity – as evidenced by the complexity found in audits (e.g., handling rounding in share conversions, unbounded loop concerns, etc.) – but that complexity is needed to achieve high performance and yield. From a risk standpoint, each module and contract is an attack surface, but the layered design confines potential failures (for instance, an issue in the External Management module ideally shouldn’t affect the base staking modules, and vice versa). Overall, the system architecture reflects a modern DeFi protocol that extends a base liquid staking model with additional yield strategies, requiring tight coordination between on-chain logic and off-chain agreements.

1 Like

wstHYPE Risk Assessment by BA Labs

Part II: Technical Risk

A. Smart Contract Risk

Smart contract risk for wstHYPE can be assessed by examining both the complexity of its design (dual token, rebasing, multi-module architecture) and the findings from security audits. The contracts have undergone multiple audits in 2025 by independent firms, which helps illuminate potential vulnerabilities. Overall, while no critical bugs were found in final code, the audits did surface several medium and low-severity issues that were addressed. The presence of upgradeable proxies and admin roles also introduces custodial risk (keys controlling upgrades), though this is more of an off-chain risk (discussed later). Below, we summarize the security findings from four audit reports – Three Sigma (Feb 2025), Pashov Audit Group (Oct 2025 and Nov 2025), and Guardian (Nov 2025) – which collectively provide a comprehensive view of the smart contract risks identified and how they were resolved. These cover issues ranging from mathematical precision and accounting bugs to gas optimizations and best practices. Each issue is listed with an identifier, title, category, severity, and status (resolution).

Audit Findings – Three Sigma (Feb 2025)

This audit was performed when the protocol was originally developed by Thunderhead Labs (before Valantis’s acquisition). Three Sigma used Immunefi’s taxonomy and found a few low-severity bugs and some informational suggestions. The issues and their status at that time were:

ID Title Category Severity Status
3S–Thunderhead Labs–L01 Inflation attack via minting to zero address in stHYPE contract Bug Low Addressed in commit #536af6e
3S–Thunderhead Labs–L02 Rounding in _transfer favors caller on burning (underestimates burned shares) Bug Low Acknowledged (not immediately fixed)
3S–Thunderhead Labs–L03 Unbounded loop in getBurnIds/getBurns can cause view function failure Bug Low Addressed in commit #3b9ee81
3S–Thunderhead Labs–N01 Missing event emission in pause functions Suggestion Informational Addressed in commit #24de282
3S–Thunderhead Labs–N02 Inconsistency between maxRedeemable and _redeemable re: protocol fees Suggestion Informational Addressed in commit #c0494ec
3S–Thunderhead Labs–N03 Lack of input validation in setter functions Suggestion Informational Addressed in commit #b7a8607

Source: Three Sigma Audit Report, Feb 2025. The Three Sigma audit identified subtle issues. For instance, L01 was a scenario where minting stHYPE to the zero address could inflate the price per share (an attacker could exploit this to steal value); this was fixed by disallowing zero-address mints. L02, about rounding, pointed out that when converting HYPE to shares in _transfer, always rounding down benefitted users burning tokens (they’d lose slightly fewer shares) – a minor fairness bug acknowledged by the team. L03 noted that as the list of burns grows, iterating through them could become too gas-costly; the team addressed this by introducing mappings to track burns per address, avoiding unbounded loops. The N-series were “notes” (informational): adding event emissions for pause actions (for transparency), aligning internal calculations for consistency, and adding input validations on admin-setter functions – all of which were implemented in subsequent updates. By the time Valantis took over, these issues were largely resolved, improving the contract’s safety.

Audit Findings – Pashov (Oct 2025)

Pashov Audit Group reviewed the Valantis stHYPE contracts in October 2025 (just after Valantis’s acquisition of the project) focusing on core functionality (Overseer, HyperCore modules). They reported 9 issues (4 medium, 5 low severity):

ID Title Category Severity Status
M-01 Incorrect management of redeemable funds (redeem logic flaw) Medium Resolved
M-02 Deposits might fail if staking module account is not activated Medium Resolved
M-03 Outdated precompile data might cause miscalculation of total supply Medium Resolved
M-04 rebase() can be blocked by large supply increase (uint overflow risk) Medium Resolved
L-01 APR can be overestimated by up to 2 basis points Low Resolved
L-02 Deposit blockage by liability increase may leave dust (edge-case) Low Acknowledged (won’t fix)
L-03 If aprChange == 0, rebase() will always revert Low Resolved
L-04 protocolFee change without prior rebase() leads to incorrect fees Low Resolved
L-05 APR calculated before fees in rebase() inflates displayed value Low Resolved

Source: Pashov Audit Report, Oct 2025. These findings were more focused on financial logic and edge cases. M-01highlighted a scenario where the contract’s tracking of funds available for redemption could be wrong if earlier redeems weren’t accounted for, possibly allowing an insolvent state (this was fixed by adjusting the _redeemable() function to consider actual contract balance). M-02 noted that if a staking module (account) hadn’t been initialized, a deposit might revert – the team resolved this by ensuring modules are activated or by handling inactive modules gracefully. M-03pointed out that using outdated on-chain data (precompile data, perhaps for Hyperliquid state) could cause total supply miscalc; updating to the latest data each rebase fixed this. M-04 found that a very large increase in supply (perhaps if many rewards accrued at once) could hit a check that reverts rebase (like an overflow guard); adjustments were made to avoid false triggers. Among low-severity issues, L-03 and L-05 both dealt with APR calculation nuances – if certain values were zero, functions reverted or APR was overstated; these were patched for accuracy. L-04 pointed at a governance function changeProtocolFee: without calling rebase first, changing the fee could apply the new rate retroactively for the ongoing epoch, slightly mis-accounting fees. The recommendation (which was implemented) was to force a rebase() (settle current rewards) before allowing the fee change, ensuring clean accounting. L-02 about “dust” was acknowledged as a benign edge case (a tiny amount of HYPE might remain locked due to rounding or increased liability – not critical to fix). All other issues were fixed in code updates, indicating the protocol was quite responsive to audit feedback.

Audit Findings – Pashov (Nov 2025)

A follow-up audit in November 2025 by Pashov focused on the new External Management module and changes around it. This audit found 4 issues, all low severity:

ID Title Category Severity Status
L-01 StakingModuleExternalManagement does not expose depositCap and amountDeposited (no getters) Low Resolved
L-02 ExternalManagement module uses the same storage slot as HyperCore StakingModule (storage collision risk) Low Resolved
L-03 rebase does not have protection for sudden balance drops (e.g., if external stake is slashed or removed) Low Acknowledged
L-04 amountDeposited is never decremented, blocking new deposits (logic bug on withdrawal from external) Low Acknowledged

Source: Pashov Audit Report, Nov 2025. These issues were relatively minor and localized to the new module handling HyENA/USDe stakes. L-01 pointed out that certain state variables (depositCap – max external stake allowed, and amountDeposited – current external stake) were not exposed via public getters, meaning other contracts or interfaces couldn’t easily query them. The team resolved this by adding functions or making those variables public, improving transparency and integration. L-02 discovered that the ExternalManagement module’s storage overlapped with the base StakingModule’s storage when used in the same proxy – a potential memory collision. This was a technical deployment issue (perhaps they reused a proxy slot); it was fixed likely by separating the modules properly or adjusting storage layout. L-03 noted that if a large amount of staked HYPE suddenly “disappeared” (for example, if an external partner like HyENA slashed or removed stake unexpectedly), the rebase() function might not handle it well (no specific check for a negative APR scenario beyond the existing threshold logic). The auditors marked this as acknowledged – likely because Hyperliquid slashing was not yet enabled or because the scenario is covered by other failsafes (the slashThreshold parameter). L-04 was an accounting bug: the amountDeposited variable (tracking total HYPE given to external modules) was only ever incremented on deposit, and never decremented on withdrawal, meaning if the protocol tried to withdraw and reallocate, it might erroneously think it’s at the cap and refuse new deposits. The team acknowledged this but may have deemed it an edge case that could be managed administratively (or they planned a fix later). None of these were critical; two were fixed and two acknowledged. This indicates that by Nov 2025, after these fixes, the contract suite had very few lingering issues, all of low severity.

Audit Findings – Guardian (Nov 2025)

Guardian’s audit in mid-November 2025 was one of the final security reviews of the staking modules. Guardian’s team found no high or critical vulnerabilities, assigning the code a “Very High Confidence” rating. They did report one high-severity issue (which was promptly fixed) and a handful of informational issues:

ID Title Category Severity Status
H-01 Reverting burn recipient bricks future redeems (DoS via burnTo address) DoS High Resolved
I-01 SMEM (StakingModuleExternalManagement) temporarily underreports balance Warning Informational Resolved
I-02 SMEM does not emit any event (on certain actions) Best Practices Informational Resolved
I-03 No plan for aligned quote assets (documentation clarity issue) Documentation Informational Acknowledged
I-04 Withdrawal amounts must be in Wei (documentation issue) Documentation Informational Resolved
I-05 Incorrect stHYPE supply adjustment (validation logic edge case) Validation Informational Acknowledged

Source: Guardian Audit Report, Nov 2025. H-01 was the most serious issue Guardian found: it appears to relate to a Denial-of-Service vector where if a burn (unstake) was directed to a contract address that reverts on receive, it could “brick” future redeems. In other words, a malicious or accidental use of a burn-to address that can’t accept HYPE would block the queue. The fix was to handle such cases or restrict burn recipients, which was implemented, eliminating the DoS risk. The informational issues were minor: I-01 noted that the External Management module could temporarily show a lower balance than actual (perhaps due to an async update issue), but it didn’t affect security and was fixed for accuracy. I-02 mirrored Three Sigma’s N01 note – some actions lacked event emissions, which Guardian recommended adding for transparency (and the team did so) for better monitoring. I-03 and I-04 were about documentation – ensuring there is a clear plan/description for quote asset integrations and clarifying units (Wei) for withdrawal functions; one was acknowledged (likely to be documented later) and one corrected in docs. I-05 flagged a potential mis-calculation in supply adjustment under rare conditions (perhaps a scenario of slashing or manual adjustment) – it was acknowledged as a known issue that did not materially affect security, left for future improvement.

In aggregate, the audit history suggests the smart contract risk is reasonably low post-remediation. The contracts benefited from multiple audit iterations, and by late 2025 the known issues are informational or very low impact. The dual-token mechanism and complex staking logic introduced risk of accounting bugs (as seen in Pashov’s findings and Three Sigma’s inflation bug), but these were caught early. Areas of remaining caution include:

  • Upgrade/Proxy Risk: Since the system is upgradeable, a bad upgrade or compromised admin could introduce vulnerabilities. This is not a flaw in the deployed code per se, but a risk in any proxy-based design.
  • Complexity: Functions like rebase() perform non-trivial calculations with edge-case considerations (APR threshold, fee pending, etc.). While audited, complexity can hide unknown issues. For instance, the interplay of maxRedeemable vs _redeemable was flagged and resolved to avoid confusion, showing that even non-critical inconsistencies can arise.
  • Integration Risks: The contract relies on external calls (to Hyperliquid’s precompiles or modules). If Hyperliquid’s interface changes or returns unexpected data, that could affect stHYPE. For example, Pashov’s note about outdated precompile data (M-03) hints at this risk – it was resolved by updating usage, but it’s a reminder that changes in Hyperliquid might necessitate contract updates.

Overall, after addressing the audit findings, the smart contracts are judged to be secure and robust. The Guardian audit’s top confidence ranking supports this. The remaining risks now shift toward how these contracts are used and controlled, which leads into off-chain and protocol-level risks.

B. Off-chain Component Risk

Off-chain components pose significant risks to the wstHYPE protocol, as they involve elements that cannot be fully controlled or enforced by smart contracts. These include the stake agreements with external protocols (HyENA/USDe), the governance and key management of Valantis’s contracts, and the broader Hyperliquid network conditions like slashing or consensus failures. We discuss each in turn:

  1. HyENA Credit Line and Partner Risk: The partnership with Ethena’s HyENA DEX introduces counterparty risk. While the arrangement is governed by a legal contract and enforced on-chain via the Stake Marketplace module (with a 90-day withdrawal lock), there is still risk that Ethena could default or run into trouble. For instance, if HyENA’s exchange incurred heavy losses or was hacked, Ethena might be unable to honor the promised 2% yield or even to return the full 500k HYPE when due. In such a scenario, the recourse is partly off-chain (legal action through Valantis Panama) and partly on-chain (Valantis invoking the withdraw lock contract to yank the stake after 90 days). During that period, stHYPE holders are exposed – their withdrawals could be delayed beyond the usual 7 days, and if Ethena outright failed, there is a question of whether the 500k HYPE backing HyENA could suffer a slash or loss. According to Hyperliquid’s rules, if a HIP-3 exchange misbehaves (e.g., becomes insolvent), the staked HYPE could be slashed as a penalty. This is the “HIP-3 slashing risk” noted in documentation: stHYPE holders collectively bear that risk by providing stake to HyENA. The magnitude of this risk is hard to quantify but non-zero – essentially, stHYPE has turned into a quasi lender to HyENA, so a failure of HyENA could translate into a loss of some portion of the 500k HYPE (if slashed by the network to cover losses) or at least loss of expected yield. Valantis being on HyENA’s multisig mitigates some of the operational risk (they would know early if something is wrong and could potentially help steer the response), but it does not eliminate financial risk if HyENA fails. In a worst case, stHYPE could suffer a hit to its collateral (making it slightly undercollateralized until slashed portions are earned back or covered by someone). Users should recognize that by chasing the higher yield from HyENA, they implicitly accept the credit risk of Ethena/HyENA.

  2. Governance, Admin Keys, and Upgrades: The off-chain management of keys is a classic risk for upgradeable protocols. Valantis’s multisig holds the keys to the kingdom: it can upgrade contracts, change fees, pause the protocol, and more. If that multisig is compromised (through hacking, insider collusion, or legal coercion), an attacker could potentially drain funds or permanently lock user withdrawals. Even without malicious intent, there’s governance risk – for example, Valantis could unilaterally raise the protocol fee to an exorbitant level, diminishing user returns, or alter withdrawal rules. While such actions would likely destroy the protocol’s reputation, they are technically possible. It’s not known if Valantis has any timelock or DAO mechanism for changes; given it’s a centralized entity, changes could be fast. This places a lot of trust in the team’s stewardship. The legal jurisdiction (Panama) might offer some protection in terms of accountability, but it also may make it difficult for users elsewhere to pursue any remedies. This category of risk was not directly highlighted in audits (auditors assume admin keys are honest), but it’s a lurking hazard: an off-chain failure of key security or governance process could bypass all on-chain safety. Mitigations include requiring multiple signatures (which they do), and possibly using a timelock for upgrades (unclear if implemented).

  3. Validator and Network Risks: Since the underlying collateral is staked on Hyperliquid, stHYPE is subject to network-level risks such as slashing and prolonged network downtime. If Hyperliquid experiences a catastrophic event (security breach, chain halt), stHYPE holders could be unable to redeem simply because the network can’t process unbondings. Validator misbehavior could lead to slashing of staked HYPE – the documentation downplays this (“highly unlikely” given selected validators), but it remains a risk. Especially as stHYPE grows, it becomes a large target: if an attacker managed to compromise Valantis’s validators or the delegation keys, they could cause a slash (though Hyperliquid’s slashing parameters were reportedly initially set to zero for some conditions, this may change). Additionally, the permissionless quote asset slashing risk is relevant to the USDe allocation: if something goes wrong with USDe (for example, it depegs severely or its mechanism causes loss), the Hyperliquid network could slash the stake used to support it. These forms of slashing are unique to Hyperliquid’s design (to ensure staked backing for quote assets and exchanges remains reliable). They effectively externalize risk from those protocols back to the stake providers (stHYPE). Users may not fully realize that, unlike simply staking HYPE where only validator performance matters, with stHYPE a portion of their assets could be slashed due to external protocol failures. This is an off-chain risk in that it’s a policy of the Hyperliquid network and behavior of other protocols, not something the stHYPE smart contracts can prevent.

  4. Transparency and Information Risk: As noted, the relative opacity means users might not promptly see early warning signs. For example, if Ethena/HyENA is in trouble, Valantis as a multisig signer might know first. There could be a delay or incentive for Valantis to quietly solve it rather than immediately inform holders. This information asymmetry is an off-chain risk – users rely on Valantis’s integrity and communication. If Valantis delayed action (say HyENA was failing but they waited weeks before pulling the stake), that could worsen losses. On the flip side, if Valantis ever needed to use emergency powers (like pausing stHYPE because of a network issue), users would have to trust that this action is in their best interest and not an arbitrary decision. Essentially, any operational failure – be it human error, miscommunication, or delay in reacting to an incident – can impact user funds even if the smart contracts function correctly.

  5. Regulatory/Legal Risk: As a Panamanian entity dealing with potentially large user deposits, Valantis could face legal issues that spill over into the protocol. For instance, an injunction or asset freeze against Valantis could complicate their ability to operate the multisig or validators. Similarly, if Hyperliquid or Ethena faced regulatory crackdowns, stHYPE might indirectly suffer (if, say, US persons were barred from using it, affecting liquidity or if Ethena’s operations were halted by authorities, risking the USDe arrangement). While speculative, these legal off-chain risks are part of the landscape for any CeDeFi (centralized-DeFi hybrid) service.

In conclusion, off-chain component risk centers on trust and external dependency: trust in partners (Ethena/HyENA) to uphold agreements and not introduce losses, trust in Valantis to manage keys and validators responsibly, and dependency on Hyperliquid’s rules and stability. The technical design has tried to reduce some off-chain risk (e.g., using smart contract enforced withdraw locks instead of pure gentleman’s agreements, and by Valantis embedding itself into partner governance). Nonetheless, stHYPE cannot escape the reality that it is tying the fortunes of its users to the performance and honesty of multiple actors beyond just code. Users should weigh these risks – for example, the higher APY is essentially a reward for bearing Ethena’s credit risk and Hyperliquid’s expanded slashing conditions. From a risk assessment standpoint, these off-chain factors likely dominate the residual risk profile of wstHYPE, given that the smart contracts themselves have been hardened. Ensuring ongoing transparency (perhaps real-time metrics or proof-of-reserves style reporting for the stake allocations) and having contingency plans (insurance fund or slashing cover) would be prudent steps to mitigate these off-chain risks as the protocol matures.

1 Like

USDH Risk Assessment by BA Labs

Part I: Protocol Overview

A. Protocol Breakdown

A.1 Protocol Details

USDH is a fiat-backed stablecoin introduced as a native dollar-pegged asset on the Hyperliquid blockchain in late 2025. Unlike decentralized stablecoins, USDH is fully collateralized by traditional financial assets (fiat currency and U.S. Treasuries) rather than crypto. The Hyperliquid community selected Native Markets – a new team formed by Hyperliquid builders – to issue and manage USDH after a competitive governance vote. This vote, concluded on Sept. 15, 2025, saw Native Markets prevail over established issuers (like Paxos and Frax) despite the newcomer’s shorter track record. The decision reflects a preference for ecosystem alignment and technical integration over incumbent reputation: Native Markets promised tighter integration with Hyperliquid’s platform and a fast rollout, aligning USDH closely with Hyperliquid’s needs.

As a flagship stablecoin of the Hyperliquid ecosystem, USDH is designed to reduce reliance on external stablecoins (e.g. USDC) and capture the yield that was previously leaking out of the ecosystem. By launching USDH, Native Markets aims to internalize the benefits of Hyperliquid’s large stablecoin balances – for instance, redirecting interest income (seigniorage) to the community rather than to third-party issuers. USDH operates natively on Hyperliquid’s EVM-compatible chain (HyperEVM), making it a core on-chain asset for trading, margin, and settlements. It is pegged 1:1 to the US dollar and intended to maintain that peg through full reserve backing and redeemability. The rollout of USDH began with a limited-cap testing phase on HyperEVM to ensure stability before wider circulation.

From a governance and issuance perspective, USDH represents a novel model where the stablecoin is effectively provided as a service by an external deployer (Native Markets) under the host chain’s oversight. Hyperliquid has now set standards for “Hyperliquid-aligned” stablecoins as an announcement in their discord server.. Native Markets’ winning proposal defined USDH as a fully reserved, fiat-collateralized stablecoin integrated with Stripe’s Bridge platform for custody and compliance. It also committed to share USDH’s revenue with Hyperliquid (details in section A.1.2) to align incentives. In summary, USDH is a centrally issued stablecoin on a decentralized exchange’s L1, bringing the familiar architecture of USDC/USDP-style stablecoins into a new ecosystem but with a revenue-sharing twist. Below, we detail its collateral backing, economic model, minting/redemption mechanisms, and system architecture, before analyzing the technical risks.

A.1.1 Underlying Collateral

USDH is fully backed by cash and U.S. Treasury assets held both on- and off-chain by regulated custodians. According to the proposal, every USDH token corresponds to 1 USD in reserves, primarily consisting of USD fiat deposits, compliant stablecoins, and short-duration U.S. Treasury bills (or equivalents). In practice, Native Markets leverages partnerships with major financial institutions to custody and invest these reserves:

  • Cash Holdings: A portion of USDH’s backing is kept in cash or cash-equivalents to facilitate redemptions. These cash reserves are initially managed via Lead Bank custodial accounts.
  • U.S. Treasuries: The majority of reserves are intended to be invested in short-term U.S. Treasuries or Treasury-backed funds to generate yield. Short-dated T-bills (e.g. 1–3 month maturities) are favored for their stability and liquidity. Notably, Native Markets partnered with Superstate – a financial platform that tokenizes T-bill investments – to manage on-chain Treasury reserves. Superstate’s product (USTB, a tokenized Treasury fund) holds real T-bills in a regulated fund structure. By using Superstate via the Bridge platform, part of USDH’s reserves can be represented on-chain as tokens backed by Treasuries. This on-chain reserve component provides additional transparency and potentially allows on-chain verification of some assets.
  • Custodial Distribution: Initially, off-chain reserves are held by BlackRock (likely in a custodial trust or fund), while on-chain reserves are managed through Superstate. Over time, the proposal mentions expanding custodians to include Fidelity, BNY Mellon, and other major institutions. This implies a diversified custody approach: multiple trusted asset managers and banks holding parts of the collateral, reducing single-point failure risk. However, all these are traditional finance entities, meaning the reserves ultimately sit under conventional legal agreements (bank accounts, trust funds, etc.) rather than on the blockchain.

Importantly, each USDH is backed by high-quality, liquid assets – there is no fractional reserve or algorithmic element. The reserve composition (cash + T-bills) mirrors the model of regulated stablecoins like USDC and PYUSD, aiming for low credit risk and minimal volatility. U.S. Treasuries are considered one of the safest assets, and by limiting to short durations, interest rate risk is low. The cash portion covers immediate liquidity needs and any gap between token redemptions and selling of T-bills.

One design choice to highlight is the hybrid on-chain/off-chain reserve model. Part of the reserves may be tokenized on-chain (Treasuries via Superstate), potentially allowing programmatic audits or transparency for that portion. Meanwhile, the cash portion remains off-chain (banked). This split introduces a complexity: the peg stability relies on both on-chain token integrity and off-chain asset guarantees. Native Markets will need to ensure that the total circulating USDH never exceeds the sum of off-chain cash plus on-chain T-bill tokens held as reserves. In Section B.1 we discuss how observable this reserve is to the public. Nonetheless, from a collateral standpoint, USDH’s backing assets are conventional and conservative: dollars in bank and Treasuries in custody, which historically pose minimal default risk. The primary residual risks are operational (e.g. a custodian’s failure or legal freeze) rather than market volatility.

A.1.2 Fee and Revenue Sources

Unlike decentralized stablecoins that may charge stability fees or give interest to holders, USDH’s model is closer to fiat-backed stablecoins with no direct fees for end users to mint or redeem at par (based on the substack proposal, where user side fees have been criticized). Instead, the financial model centers on capturing the yield from the invested reserves and distributing it between the issuer and the ecosystem. The key revenue source is the interest income earned on USDH’s collateral (primarily from those U.S. T-bills yielding interest). At current Treasury rates (~3% APR for 3-month T-bills in late 2025), a substantial annual yield is generated on every dollar of USDH in circulation. This yield is analogous to the revenue that Circle earns on USDC reserves, but USDH’s innovation is to share it with Hyperliquid stakeholders rather than keeping it all.

According to Native Markets’ proposal, 100% of net reserve yield will be reinvested into the Hyperliquid ecosystem rather than retained as issuer profit. Practically, they pledge to split the reserve earnings 50/50 between two buckets:

  • Hyperliquid Assistance Fund (HYPE Buybacks): Half of all interest accrued on USDH reserves will go to Hyperliquid’s Assistance Fund for purchasing and burning HYPE (Hyperliquid’s native token) or similar value-accretive activities. This effectively recycles revenue to Hyperliquid’s token holders, boosting the HYPE token’s value. It aligns with Hyperliquid’s goal of “economic sovereignty”, ensuring the protocol benefits from the stablecoin’s adoption. For example, if USDH grows to a $5B supply fully backed by Treasuries at 4%, roughly $200M/year in interest is generated; under this scheme, about $100M would be used to buy back HYPE, supporting its price.
  • Ecosystem Growth Initiatives: The other 50% of yield is allocated to fund the growth of USDH and the Hyperliquid ecosystem. This broad category includes user incentives, partnerships, developer grants, and integrations. In practice, Native Markets might use these funds to incentivize USDH liquidity (e.g., provide rewards for market makers or liquidity providers who make a deep USDH market), to subsidize on/off-ramp fees for users, or to integrate USDH into more DeFi applications on HyperEVM. It can also cover operational costs of maintaining USDH and developing new features. Essentially, this portion acts as a war chest to drive adoption and utility of USDH (and by extension Hyperliquid’s ecosystem).

Notably, Native Markets is not taking a cut of the reserve interest for itself initially. This is in contrast to some competitors’ proposals (Paxos, etc.) which wanted to retain a small percentage as profit. Native Markets’ zero-profit stance (all yield to community) was likely a strategic move to win the bid, but it raises questions of long-term sustainability. The team has no other revenue-generating product, so foregoing any direct income means they rely on the ecosystem grant (that 50% for growth) to cover their expenses. In other words, their “profit” must come indirectly from that same pool used to grow USDH or from the value of their HYPE holdings, unless other revenue-generating product lines are introduced. This could be revisited in the future if the model is unsustainable, but for now the fee structure is intentionally generous to Hyperliquid.

Beyond interest yield, are there any fees to users? As of launch, there are no protocol-imposed transaction fees on USDH. Transfers of USDH on HyperEVM incur only normal network gas costs (like any ERC-20). Minting and redemption (discussed in A.1.3) appear to be offered at par (1 USDH for $1 collateral), with Native Markets not charging a spread or fee on these conversions. This is important for maintaining a 1:1 peg without friction. Some stablecoin issuers (e.g., Tether historically) imposed small redemption fees or minimums; USDH is positioning itself to be user-friendly with free conversions. Indeed, Stripe’s Bridge platform, which facilitates USDH issuance, advertises free conversions between stablecoins and fiat for developers in its network. This suggests that if a user swaps USDC → USDH or vice versa via Bridge, they would not pay a conversion fee (aside from any third-party on-chain swap fees).

One indirect cost might be on fiat on/off-ramp services: if a user wires money in to mint USDH or withdraws to a bank, banking fees (wire fees) could apply, though those are external costs. Native Markets may choose to cover or subsidize such fees using the ecosystem fund to encourage adoption (this would fall under “user incentives”).

In summary, USDH’s revenue model is driven by reserve yields, not by taxing users. All yield flows to the ecosystem (split between HYPE token support and growth initiatives), making USDH a kind of public utility for Hyperliquid. This model prioritizes adoption and alignment over immediate profit. It will be crucial that the reserve is large (i.e., USDH widely used) to generate meaningful revenue – a classic bootstrap problem. That is likely why a portion of yield is earmarked for incentives: to jump-start usage until network effects take hold. We will later analyze the risk that Native Markets’ no-profit approach could affect its ability to sustain operations or respond to emergencies (see Part II). But from a fee perspective, USDH is very attractive to users (no mint/redeem fees, no negative carry), and from a network perspective, it creates a new revenue stream for Hyperliquid out of what was previously “leaked” to external issuers.

A.1.3 Minting and Redemption Flow

USDH’s mint and redemption process combines on-chain token transactions with off-chain fiat transfers and an API-driven orchestration, primarily handled by Stripe’s Bridge platform. In essence, Bridge acts as the gateway for converting external assets (fiat or other stablecoins) into USDH and vice-versa. Below we outline the typical flows:

  • Minting USDH (Fiat to USDH): A user or institution that wishes to mint new USDH deposits U.S. dollars through one of Bridge’s supported payment rails. In practice, this involves initiating a bank wire or ACH transfer to a custodial account held by Bridge on behalf of Native Markets. Because Bridge is a regulated Money Services Business (MSB), it can hold fiat reserves and verify counterparties under strict KYC/KYB standards. Once the depositor is KYB-verified and the USD funds are received, Bridge’s backend triggers on-chain issuance. The authorized Bridge key calls the USDH smart contract’s mint() function to create the equivalent amount of USDH tokens and deliver them to the user’s address on HyperEVM. This process is automated via Bridge’s API, allowing direct integration into external applications or exchanges. In the future, gateway APIs such as Stripe Crypto or MoonPay are expected to streamline this further — enabling flows where, for example, a user on an exchange like Hyperliquid can click “Deposit USD for USDH,” and behind the scenes, Bridge’s API handles the bank withdrawal and on-chain minting. Minting occurs at a 1:1 rate (1 USDH per $1 deposited), assuming no issuance fees, with the entire process designed for regulated, programmatic fiat onboarding.
  • Redemption (USDH to Fiat): Conversely, when redeeming, the user sends USDH tokens to the issuer (likely to a specific redemption address or smart contract controlled by Bridge/Native). The Bridge platform will burn those tokens (removing them from circulation) and simultaneously initiate a fiat payout of the equivalent USD amount to the user’s bank account. This could be via wire, ACH, or other supported payout methods. The redemption is also API-driven; for instance, a user might request a withdrawal on a web interface, which prompts Bridge to provide instructions to send USDH to a certain address. After the on-chain burn is detected, Bridge releases fiat funds to the user off-chain. The design ensures that total USDH in circulation decreases upon redemption, keeping the supply backed 1:1 by remaining reserves.
  • Stablecoin Conversions: In addition to direct fiat ↔ USDH ramps, the system supports stablecoin inflows as a secondary minting path. Since Hyperliquid initially held substantial USDC liquidity, a key early flow is USDC → USDH conversion. Native Markets confirmed that this route is live today via Arbitrum USDC, with other networks (Ethereum, Base) and USDT on Tron to follow. Practically, users deposit USDC through Bridge; the backend either holds or retires the USDC and then mints USDH on HyperEVM to the user’s address. This provides near-instant conversion without touching banking rails. In parallel, a USDH/USDC spot pair on Hyperliquid enables free-market swaps, ensuring tight peg arbitrage—traders can buy or sell USDH against USDC whenever it drifts from $1. As more networks are added, these flows will deepen liquidity and make USDH a cross-network stablecoin hub bridging fiat and crypto liquidity into Hyperliquid.
  • Bridge Orchestration Network: A notable feature is that Bridge’s stablecoin orchestration network can handle multi-currency and multi-chain operations. This means USDH issuance isn’t necessarily confined to just accepting USD from U.S. bank accounts. For example, a user in Europe could send EUR to Bridge, Bridge might convert it to USD on the back-end, and then mint USDH. Or a user with USDC on Ethereum could have Bridge burn that USDC (via Circle’s protocol) and mint USDH on HyperEVM. All such flows are abstracted behind the API. The goal is a seamless experience where users can obtain USDH from whatever source is convenient, without dealing with the low-level steps. This interoperability is a strength of using Stripe’s Bridge, as it already connects to various payment methods and other stablecoins.
  • KYC Requirements: Because fiat transfers are involved, users minting or redeeming directly will face KYC/AML checks (likely through Stripe/Bridge’s compliance system). Institutional minters must provide verification. This means permissioned access at the on/off-ramp level, though on-chain transfers of USDH remain permissionless between any non-blacklisted addresses. In other words, holding and trading USDH on Hyperliquid is open to all, but converting to/from real USD likely requires approved accounts to prevent money laundering. This is similar to USDC’s model. Bridge’s integration provides Stripe’s robust KYC pipeline, which is a benefit for compliance but also a point of centralization.

In terms of user experience, initially USDH minting may be dominated by institutional players or market makers rather than small users wiring funds. These institutions can bring liquidity by swapping large amounts of USDC for USDH or by depositing fiat to mint and then distributing USDH via trading. Indeed, Native Markets signaled that “select institutions” will provide on-chain liquidity at launch, likely meaning they arranged for partners to mint or swap a base supply of USDH to seed the ecosystem..

Finally, redemption guarantees are crucial for confidence: Native Markets must honor one USDH for $1 at any time (for authorized participants). They will likely mirror the policies of other stablecoins – e.g., no redemption limits except perhaps a minimum amount for wire transfers, and prompt (same-day or T+1) settlement. The backing of Stripe’s infrastructure suggests redemptions can be processed efficiently (Stripe has banking integrations globally). The presence of multiple custodians (BlackRock for Treasuries, etc.) means that to fulfill large redemptions, assets might need to be sold (e.g., liquidating T-bills). This introduces a timing aspect: if a huge volume of USDH is redeemed at once, Native Markets would have to rapidly liquidate Treasuries or draw on credit lines since Treasuries settle trade in 1 day. We will consider this liquidity risk later, but presumably a buffer of cash is maintained to handle routine redemptions quickly.

In summary, minting and redemption of USDH rely on a centralized pipeline (Bridge) that links traditional money to Hyperliquid’s blockchain. The system emphasizes automation (via API) and seamless swaps, so ecosystem participants can easily move into USDH. This design prioritizes convenience and speed (e.g., instant stablecoin swaps) while maintaining full collateralization off-chain. However, it also means trusting the issuer’s infrastructure for these flows – if Bridge or the issuer halts operations, new minting/redemption could pause (an aspect we discuss under risk). The on-chain smart contracts (covered in Part II.A) enforce the supply accounting (mint/burn), but the decision to mint or burn and the custody of assets is handled by off-chain agreements and software.

A.1.4 Fee Structure and Incentives

USDH’s fee structure is minimal by design, to encourage adoption and usage within Hyperliquid. As noted, users are not charged explicit fees for converting to or from USDH at par value. This zero-fee approach extends to trading: Hyperliquid’s exchange treats USDH like a base currency for spot and perpetual markets, subject only to the normal trading fees of the platform (no extra surcharge for using USDH). By removing financial friction, USDH can compete toe-to-toe with incumbent stablecoins (USDC, USDT) which also generally allow free creation/redemption for institutional clients.

The absence of fees, however, does not mean there are no incentives – in fact, the incentive design is central to USDH’s strategy, aligning stakeholders to promote the stablecoin’s growth:

  • Validator and HYPE Holder Incentives: The 50% reserve yield allocated to HYPE buybacks (see A.1.2) is essentially an incentive for Hyperliquid’s validators and token holders to prefer USDH. By driving up the HYPE token’s value (or distributing yield via the Assistance Fund), the network’s stakeholders economically benefit from USDH adoption. This creates a strong alignment: validators who supported Native Markets in the vote will see returns only if USDH succeeds and captures a large TVL. In effect, Hyperliquid’s security providers are “paid” in part by the stablecoin’s yield, motivating them to integrate and use USDH wherever possible. This incentive may also cover staking rewards indirectly; e.g., if some yield is channeled to increase validator rewards or subsidize lower fees on USDH trades. The key point is that the Hyperliquid community is financially rewarded for helping USDH thrive (a contrast to USDC where Hyperliquid got no kickback).
  • User Incentives (Yield Redistribution): While USDH holders themselves do not earn interest directly (1 USDH stays 1 USDH, with no rebasing or dividend), Native Markets suggested that yield could be used to boost user yields in DeFi apps on HyperEVM. For example, a lending protocol on HyperEVM that uses USDH might offer a higher savings rate because Native Markets could supplement it using the reserve earnings. Or a liquidity pool pairing USDH could have extra rewards funded by the ecosystem allocation. These kinds of “yield farming” incentives effectively pass some of the stablecoin’s revenue to end-users to encourage uptake. It’s a way to mimic interest on holding USDH without complicating the token itself – instead of accruing interest within USDH (like some interest-bearing stablecoins attempt), they distribute it through ecosystem programs. This approach can attract liquidity: users might prefer USDH over USDC if participating in Hyperliquid DeFi gives them better rewards thanks to those subsidies.
  • Developer/Integration Incentives: Another facet is incentivizing other projects to integrate USDH. Bridge’s model, for instance, shares the majority of yield with developers who use its stablecoin USDB. Native Markets can apply a similar idea: if a partner protocol or exchange agrees to use USDH as a base currency or on-ramp, they could receive a portion of the yield or grants from the ecosystem fund. This encourages third parties (wallets, dApps) to add USDH support. The proposal explicitly mentions funding “partnerships with Builder Code interfaces, HIP-3 markets, HyperEVM apps” from the 50% growth allocation. Essentially, subsidies are available to integrate USDH into new markets and products (HIP-3 refers to a proposal that lets anyone launch new markets on Hyperliquid). By lowering integration costs or even paying projects to prefer USDH, Native Markets accelerates its network effect.
  • Fee Structure for Stability: Interestingly, not charging fees can itself be seen as an incentive: it eliminates arbitrage frictions. If, say, USDH drifted below $1, arbitrageurs can buy it cheap and redeem 1:1 for $1 with no fee, pocketing profit, which helps restore the peg. If there were hefty redemption fees, arbitrage would be less attractive and the peg could stray more. So the no-fee policy supports tighter peg stability by encouraging arbitrage and convertibility in both directions.

One potential incentive that USDH does not currently offer is an interest-bearing version of the token to holders (e.g., distributing yield directly to wallets). This could be provided by third-party platforms that tokenize lending market deposits or automated strategy vault positions.

To summarize, USDH’s fee structure is deliberately low or zero across the board, shifting the focus to incentive alignment rather than revenue extraction. Native Markets is effectively subsidizing the early adoption of USDH using the promise of future yield. The incentives are structured such that every key player (validators, users, developers) has something to gain if USDH succeeds: validators see their HYPE investment gain value, DeFi users get better yields or rewards, and developers/protocols receive funding support. This “growth hacking” with financial incentives is crucial because USDH, as a newcomer, must entice users away from entrenched stablecoins like USDC that they already trust. The true test will be whether these incentives can overcome the inherent switching costs and trust barriers; we will discuss those risks (e.g. trust and transparency) later. But from a design viewpoint, USDH is an example of a modern stablecoin that treats its yield as a public good for the ecosystem, essentially sharing the seigniorage to bootstrap a competitive moat.

A.1.5 Off-Chain Components

USDH’s system relies on several critical off-chain components and services, which handle functions that cannot be done purely on-chain given the fiat-backed nature of the asset. These off-chain elements introduce points of centralization and trust, but they are also what allow USDH to interface with the traditional financial system. The major off-chain components are:

• Stripe’s Bridge Platform:
Bridge (acquired by Stripe in 2025) is the core infrastructure for USDH issuance, redemption, and compliance. It provides banking integrations and APIs to move money between bank accounts and stablecoins, handles KYC/AML, and monitors transactions. When dollars come in, Bridge instructs on-chain contracts to mint USDH; on redemption, it confirms the burn and triggers fiat payouts. This makes Bridge a critical off-chain dependency: if it goes down, minting/redemptions halt. Bridge also maintains identity mappings for compliance and has administrative powers (e.g., freezing/blacklisting), executed on-chain but decided off-chain.

• Custodial Bank Accounts and Custodians:
USDH reserves are held in off-chain custodial accounts, initially via BlackRock’s setup and partner banks (e.g., Lead), with others like BNY Mellon or Fidelity mentioned as future custodians. A portion likely sits in BlackRock-managed money market funds (T-bills), plus cash for liquidity. These reserves, custody arrangements, and fund movements are entirely off-chain and governed by legal agreements. Verification of backing relies on off-chain audits or attestations by third parties, which users must trust.

• Superstate (On-Chain Treasuries, Off-Chain Assets):
Superstate’s USTB is on-chain, but the T-bills are held off-chain in a trust. Superstate handles asset purchase, custody, and reconciliation, and publishes attestations or filings. The 1 USTB = $1 claim is therefore an off-chain assurance, and any legal or financial issue at Superstate would affect that portion of USDH’s reserves.

• Compliance and Legal Layer:
USDH depends on licensed off-chain entities. Bridge/Stripe holds the regulatory permissions (MSB/MTLs, non-US entities, etc.), which Native Markets leverages. Regulatory orders could halt issuance or freeze reserves. All governing relationships—issuer–Stripe, custodians, user terms—are off-chain contracts that ultimately define redemption rights and restrictions.

• Administration & Control:
USDH smart contracts almost certainly include admin keys (pause, freeze, authorized minting). These are controlled off-chain by organizations or multisigs. Any emergency action (pauses, upgrades) requires off-chain coordination. Unlike algorithmic systems, USDH doesn’t rely on price oracles for collateralization.

• Monitoring and Alerts:
Operational monitoring is fully off-chain: systems track mint/burns, large transfers, reserve balances, and compliance signals, with alerts for discrepancies or suspicious activity. This off-chain layer is essential for day-to-day risk management and incident response.

In summary, while USDH transacts on a blockchain, its full trust model extends into the off-chain realm of banks, institutions, and centralized services:

  • You trust that Bridge securely holds your dollars and will honor redemptions, and that it won’t be hacked or go bankrupt.
  • You trust that the custodial banks/funds actually have the dollars and T-bills they claim (with proper segregation such that if, say, Bridge had financial issues, those reserves are not claimable by Stripe’s creditors – this is typically managed via trust accounts).
  • You trust that the issuer’s legal/compliance framework will not be suddenly shut down by regulators, or if it is, there’s a process for holders to redeem in an orderly fashion.
  • You also rely on off-chain governance decisions: for example, if a bug is found, the team will fix it promptly (they have the keys to do so), or if someone is sanctioned, they might freeze their USDH per legal requirements.

This off-chain dependence is a classic risk for fiat stablecoins – essentially USDH inherits the counterparty risk of the traditional financial system. Native Markets has tried to mitigate this by using reputable partners (Stripe, BlackRock, etc.), but it cannot eliminate it. The trade-off is clear: without these off-chain components, USDH couldn’t exist as a fiat-backed stablecoin; but with them, USDH cannot be trustless or entirely transparent in real-time. We will delve more into the implications of this opaqueness and trust requirement in the Monitoring (B.1) and Risk (Part II) sections.

[TO BE CONTINUED BELOW]

1 Like

B. System Design

B.1 Monitoring and Observability

One critical aspect of a stablecoin’s robustness is the transparency of its reserves and the ability for the community to monitor its health. Given USDH’s partially opaque off-chain backing, monitoring and observability are areas of focus and potential weakness. Here we consider what information is (or isn’t) available to the public and how the system’s design affects transparency:

Reserve Transparency: As of now, USDH does not have a real-time public dashboard of its reserve assets. Users must rely on periodic disclosures and the reputations of the custodians. Native Markets has indicated intentions to comply with regulatory standards (e.g., GENIUS Act in the US) which would entail regular audits, but until formal audits are released, the backing is effectively an article of trust. This contrasts with some competitor proposals during the bidding: for instance, Sky (formerly MakerDAO) offered real-time dashboards for their reserves. In USDH’s case, some reserves are held via Superstate’s on-chain tokens, which provides partial transparency – anyone can inspect the USDH smart contract or relevant custody contract on Hyperliquid to see how many USTB (or similar) tokens it holds, if that design is used. However, understanding that requires insight into the architecture (and currently, detailed documentation of USDH’s reserve contract is not public). The off-chain cash portion is entirely opaque to on-chain observers; one must trust financial statements. We expect Native Markets/Bridge to publish monthly attestation reports (e.g., by a firm like Deloitte or a licensed accountant) confirming the total reserves equal or exceed total USDH supply. These would be analogous to USDC’s monthly attestations. Additionally, because Bridge is regulated, it will likely have to provide reports to regulators. While those mechanisms provide some oversight, they are not on-chain, real-time proofs.

Public Supply Tracking: On the positive side, the circulating supply of USDH is fully observable on-chain. Hyperliquid’s blockchain (HyperEVM) records every mint and burn event. Community members can monitor the USDH token contract for changes in total supply. This at least allows tracking if new USDH is being issued and potentially correlating it with announcements (e.g., large mints might coincide with big deposits by partners). The Hyperliquid explorer or third-party block explorers can show the top holders of USDH, flows between addresses, etc. This is a standard transparency feature for any ERC-20 token. It doesn’t directly prove collateral, but it sets an upper bound on needed collateral. As long as independent audits confirm collateral ≥ supply, and the on-chain supply is known, one can infer no under-collateralization.

On-Chain vs Off-Chain Discrepancies: Because part of the reserve may be held in on-chain instruments (like Superstate’s tokenized T-bills), the Bridge platform could potentially provide on-chain attestations by linking those tokens. For example, if an address (say, a Bridge-controlled custody address on Hyperliquid or Ethereum) holds X units of a token that represents T-bills, that’s visible. If the rest (cash) is small or kept mostly for liquidity, one might even programmatically prove 80-90% of reserves. However, at launch it’s unclear how much, if any, of the reserve was moved on-chain vs kept as cash. Without a dedicated dashboard or proof-of-reserve system, external observers currently face opacity regarding where the backing assets sit. This is a conscious trade-off to use TradFi infrastructure – it doesn’t update on a public ledger.

Opaqueness and Trust: The lack of public real-time data on reserves means USDH holders must trust the issuer and its partners, which is a risk factor. This opaqueness was highlighted in community discussions: some viewed relying on Stripe/Bridge as adding a “black box” in the middle of a supposedly decentralized ecosystem. Stripe is a private company, and while highly reputable, it doesn’t routinely publish crypto reserve info to the public. The community will likely demand regular transparency reports from Native Markets – perhaps quarterly full audits and monthly attestations at minimum – to build confidence. In stablecoin history, opacity has correlated with mistrust (e.g., Tether faced criticism for years when it provided limited transparency). Native Markets, being new, has “to prove trust” as The Defiant noted. Until a track record of accurate reporting and perhaps third-party verification is established, USDH may trade or be treated with caution by some users (for instance, if USDH had a depeg scare, lack of transparency could exacerbate panic since users can’t instantly verify reserves).

Monitoring Underlyings: The question of dashboards: Some stablecoins provide a live breakdown of their holdings (e.g., Circle sometimes shares the composition of USDC reserves by asset type). Native Markets has not launched a public dashboard yet. However, given that Bridge’s own stablecoin (USDB) is fully backed by BlackRock’s fund, there might be data from BlackRock about the fund’s total assets. If USDH reserves are co-mingled in, say, the same BlackRock liquidity fund used by Stripe’s stablecoin accounts, one could indirectly track that fund’s total AUM (which BlackRock publishes daily for some funds). But mapping it specifically to USDH might not be straightforward.

Audits and Regulatory Filings: Under the hood, Bridge’s regulated status means authorities could request reports. In the EU, for example, under MiCA (Markets in Crypto-Assets regulation), issuers must publish reserve information and have certain audit requirements. In the US, if GENIUS Act compliance is targeted, that would involve stringent oversight (the GENIUS Act, as referenced, calls for frequent disclosures and potentially real-time reporting to regulators). While these mechanisms are not immediately visible to end-users, they provide some assurance that a governing body is watching. Of course, regulators act after the fact, not in real-time, so they might catch issues only periodically.

In conclusion, observability of USDH’s backing is relatively low for the general public right now, leaning on traditional assurances over on-chain proof. This opaque nature is a risk – it requires ongoing trust in Native Markets and Stripe’s integrity and competence. It stands in some contrast to the fully on-chain ethos of DeFi. However, the team is operating within the constraints of today’s fiat system, where complete on-chain transparency of dollars in a bank is not yet feasible. We expect improvements in transparency over time (e.g., publishing addresses holding any on-chain reserves, releasing attestation reports regularly, maybe even zk-proof based attestations in the future). For now, users and the community should monitor the available signals closely: total USDH supply changes, any official reports, and even secondary market behavior (if USDH consistently trades at $1 or if it develops a risk premium in price, that’s an indirect indicator of trust). Opaqueness can cause doubt for stakeholders, so Native Markets will need to proactively address this by increasing clarity.

From a risk perspective, the limited observability means issues could potentially go unnoticed longer than in a fully transparent system. For example, if reserves fell short (due to a loss or freeze), the public might not know immediately – a situation that has caused crises for other stablecoins historically. We will revisit this in Part II when discussing off-chain risks. For now, the recommendation is that improving public monitoring tools should be a priority for USDH’s team to bolster confidence.

B.2 System Architecture

Architecture Overview: USDH’s design combines off-chain fiat custody with on-chain token issuance and cross-chain integration into Hyperliquid’s trading network. The key components are: (1) the Bridge backend (off-chain) which handles user KYC, fiat payments, and reserve management; (2) the USDH stablecoin contract on Hyperliquid’s EVM chain (HyperEVM), which is an ERC-20 token (using Bridge’s StablecoinTemplateV3) that implements mint/burn logic under Bridge’s control; (3) a CoreRouter smart contract (deployed on HyperEVM) that acts as a custody bridge between HyperEVM and Hyperliquid’s trading chain (HyperCore); and (4) the HyperCore network itself, where USDH is used for trading (e.g. as a base asset for spot and perp markets). Native Labs explicitly chose to issue USDH natively on HyperEVM and then bridge it into HyperCore, tightly integrating with Hyperliquid’s two-layer architecture. All on-chain contracts (the USDH token and CoreRouter) have been audited and open-sourced, while off-chain processes are handled by Bridge’s regulated platform.

Minting (Fiat On-Ramp to USDH): To issue new USDH, a user first interacts with Bridge off-chain. The user creates an account with Bridge and passes KYC/AML checks (Bridge is a licensed money transmitter in the US/EU). The user then sends fiat (e.g. an ACH or wire in USD) to Bridge’s bank account. Once the fiat is received, Bridge’s backend triggers an on-chain mint: Bridge (as the authorized issuer) calls the USDH token contract on HyperEVM to mint the equivalent amount of USDH to the user’s blockchain address. The stablecoin contract is permissioned – it has a MINTER_ROLE that Bridge’s issuer key holds – ensuring that only Bridge can mint or burn tokens. (The contract also includes compliance features like pausable transfers and an address blacklist.) This means USDH has a centralized mint authority: every token originates from Bridge’s action, backed by an off-chain fiat deposit. After minting, the user receives USDH on HyperEVM (the token’s total supply increases accordingly) at the address they have provided. At this point, the user can proceed to use it within Hyperliquid. Notably, Native Markets built a mechanism to atomically mint and bridge USDH in one step: their proposal introduced the CoreRouter contract to allow immediate issuance directly into HyperCore. In practice at launch, minting involves HyperEVM issuance followed by a separate deposit step, but the infrastructure is designed to optimize this flow.

Deposit to HyperCore (Trading Custody): Once a user has USDH on HyperEVM, they will often move it into Hyperliquid’s trading environment (HyperCore) to use it as margin or to trade. The design goal was to make this deposit atomic and trust-minimized. USDH uses the same technology that runs on HyperBFT to bridge it between HyperCore and HyperEVM.

Withdrawal and Redemption (USDH Off-Ramp): If a user in HyperCore wants to withdraw back to the HyperEVM chain (for example, to redeem USDH for cash), the flow essentially reverses. The user initiates a withdraw on HyperCore, reducing their HyperCore USDH balance and signaling the CoreRouter contract to release funds. The CoreRouter then unlocks the corresponding USDH amount on HyperEVM and transfers the tokens back to the user’s HyperEVM address. At this point, the user once again holds USDH on the HyperEVM chain. To redeem for actual dollars, the user goes through Bridge’s off-chain process: they submit a redemption request (via Bridge’s app/API), and then send their USDH tokens back to Bridge (typically by transferring them to a designated burn address or calling a Bridge-provided burn function). Bridge verifies the request and the user’s KYC status, then burns the USDH tokens on-chain (reducing the total supply) and sends out an equivalent amount of fiat USD from its reserves to the user’s bank account (or supported stablecoins to their provided address). This completes the redemption. The mint/burn logic is designed such that USDH in circulation always matches the off-chain reserves. However, the process relies on Bridge’s custody and compliance controls – for example, Bridge could pause minting or block certain addresses if required (it inherits these controls via the StablecoinTemplate contract). It’s worth noting that Bridge holds all reserve assets (cash at banks and Treasuries via custodians) off-chain; there is no on-chain collateral pool to audit. Thus, redeemability hinges on Bridge managing reserves conservatively. The system does not have programmatic on-chain enforcement of 1:1 backing, but rather legal/regulatory safeguards. This is a centralized model similar to USDC/USDP, and it introduces some trust requirements (the Oak research report flagged that reliance on a TradFi entity like Bridge/Stripe is a centralization risk).

Security and Technical Considerations: USDH’s architecture deliberately balances on-chain and off-chain elements. On-chain, transparency and security come from the audited code: the USDH token contract and CoreRouter are open-source and were audited (e.g. Zenith and Zellic audits) with no major issues found. The HyperEVM contract state (total supply, balances, blacklisted addresses, etc.) is public. Because Bridge’s issuance platform uses a standard template, USDH inherits features like pause and blocklist for compliance, and a maxSupply parameter that can be set if needed. The CoreRouter logic is crucial – it must ensure that tokens cannot be double-issued between HyperEVM and HyperCore. The use of an atomic cross-chain message (validated by Hyperliquid’s validators) means the deposit and credit happen in a single logical operation, or not at all, preventing race conditions. This design avoids a scenario where USDH is minted on HyperEVM but fails to credit on HyperCore (or vice versa). Still, the mechanism relies on the fact that HyperCore and HyperEVM are tightly coupled (the same validator set), rather than a standard trustless bridge – this was a deliberate choice to make transfers fast and secure within Hyperliquid’s ecosystem. Off-chain, Bridge’s role introduces a “black box” element: users must trust that Bridge actually holds the corresponding fiat for every USDH and will honor redemptions. There is currently no on-chain proof-of-reserve for the cash held in bank or the T-bill accounts (aside from the small on-chain portion). This is mitigated by regulatory compliance (Bridge is Stripe-owned and regulated) and future transparency commitments, but it is less transparent than crypto-collateralized stablecoins. Circulating supply verification is straightforward on-chain (e.g. ~15 million USDH were minted in the first 24 hours), but verifying the fiat reserves backing that supply is done via third-party trust (auditors, etc.). In summary, USDH’s system architecture is a hybrid: transparent and deterministic on-chain for the token and Hyperliquid integration, but centrally administered off-chain in terms of issuance and backing

To visualize the architecture, imagine the flow of a new deposit:


Each arrow in this flow crosses trust boundaries between off-chain and on-chain, which is why each interface (Bridge API to contract, CoreRouter bridging, etc.) has been carefully designed and audited.

In summary, USDH’s architecture is a mix of blockchain contracts (providing the token ledger and intra-Hyperliquid mobility) and off-chain systems (providing fiat connectivity, compliance, and reserve management). It integrates tightly with Hyperliquid’s dual-layer design, ensuring that USDH truly functions as a native currency within the ecosystem. The architecture prioritizes performance (Hyperliquid’s core can handle high throughput with USDH) and compliance (Bridge handles the heavy KYC work). However, it inherits centralized elements that are single points of failure if not properly managed. The smart contract components (token and router) create a robust on-chain framework for USDH, but they ultimately depend on the off-chain components to be effective and trustworthy.

One can think of USDH’s architecture as two concentric systems:

  • The inner on-chain system (Hyperliquid network) which is transparent to validators and users on-chain.
  • The outer off-chain system (issuers, banks, monitors) which is opaque but critical.

This dual architecture will be a recurring theme in risk analysis: on-chain failures vs off-chain failures. Both need to work in harmony to maintain the stablecoin’s integrity. Native Markets has put significant engineering into the on-chain side (e.g., testing on testnet, auditing) and significant partnership effort on the off-chain side (Stripe, Lead, BlackRock).

1 Like

USDH Risk Assessment by BA Labs

Part II: Technical Risk

A. Smart Contract Risk

The on-chain component of USDH relies directly on Bridge’s StablecoinTemplateV3 contract, which governs issuance, burning, and transfer restrictions. As USDH’s foundation, its correctness and security are paramount for both peg stability and systemic trust in Hyperliquid markets.

Audit Coverage

The Zenith Security Assessment (July 2025) targeted the stablecoin-template repository (commit hash 433b5473fa7207d88c2e11d5d501452b4bb2d807)

Bridge (Stablecoin Template) - …

. This scope included StablecoinTemplateV3.sol, the template contract used to deploy fiat-backed stablecoins like USDH. The review lasted three days (July 15–17, 2025), with the final report published on July 29, 2025.

Zenith applied its standard risk classification framework (Critical, High, Medium, Low, Informational). Notably, no critical, high, medium, or low-severity issues were identified. Only a single informational issue was reported and subsequently resolved.

Findings Table

ID Description Severity Impact Status Resolution
I-1 Incorrect comment in StablecoinTemplateV3.sol (line 179 referred to _beforeTokenTransfer() after logic moved to _update() override). Informational None (documentation only) Resolved Fixed in PR-15, verified by Zenith

Bridge (Stablecoin Template) - …|

Observations

  1. Security Posture
    The absence of substantive issues suggests that the contract design inherits from a mature and battle-tested template, with no exploitable flaws in minting, burning, or transfer controls identified.
  2. Limitations of Audit
    The audit was narrow in scope (three days, a single repository, one commit hash). While effective for surfacing obvious flaws, this timeframe limits deep formal verification, edge-case fuzzing, or economic attack surface exploration.
  3. Transparency and Reproducibility
    The publication of the exact repository and commit hash is a positive step, enabling reproducibility. However, Bridge’s audit cadence and coverage of ancillary components (governance, admin key management, off-chain bridge integrations) remain opaque. As USDH depends on these off-chain hooks for minting/redemption, unverified pathways remain a systemic risk.
  4. Residual Risks
  • Centralization of Control: Even with a clean audit, the mint/burn logic can be misused if admin keys or multisig governance is compromised.
  • Integration Risks: The StablecoinTemplateV3 audit does not cover USDH’s downstream usage within CoreRouter, nor its interaction with HyperCore exchange credits.
  • Code Drift: The audit certifies only a specific commit; any upgrades or patches must be re-audited.

Conclusion

The Bridge audit establishes a baseline of confidence: no exploitable contract bugs were detected in StablecoinTemplateV3 at the time of review. However, systemic smart contract risk for USDH remains non-zero due to governance centralization, scope limitations, and integration with off-chain mint/burn flows. We recommend:

  • Publishing future audits with broader scope (including upgradeability and off-chain components).
  • Formal verification of invariants (total supply conservation, transfer blocklist enforcement).
  • Independent re-audits for CoreRouter integration, as USDH is systemically important to Hyperliquid.

B. Off-Chain Component Risk

Off-chain risks encompass everything that can go wrong outside the blockchain smart contracts – and for USDH, this is a crucial area. Since USDH bridges the crypto and traditional finance worlds, it inherits all the vulnerabilities of the off-chain systems: counterparty risk, regulatory risk, operational failures, etc. We analyze these risks point by point:

  1. Custodial Risk (Reserve Security): All USDH tokens represent claims on assets held by third-party custodians (banks and investment funds). This setup carries the risk of custodian failure or loss. For instance, if a bank holding USDH cash reserves were to fail or freeze the account (perhaps due to legal action or insolvency), those funds might become unreachable. Even if eventually recovered, redemptions could be suspended in the interim. Similarly, though U.S. Treasury bills are extremely safe in terms of credit risk, they are held via an intermediary (BlackRock’s fund or a broker). There’s a remote risk of technical issues or fraud at that intermediary. Segregation of assets is critical – the USDH reserves are held in trust or custodial accounts separate from any operating funds of Native Markets or Bridge. If not, they could be subject to claims in bankruptcy. We trust that, being licensed, Bridge set it up properly (e.g., like Paxos uses trust companies to hold reserves off balance sheet). Nonetheless, the users have no direct claim on the underlying assets except via the issuer. This is standard for centralized stablecoins but is a risk: you rely on the issuer to honor redemption; if the issuer disappeared, users might face legal battles to get funds from the custodian.

  2. Regulatory and Legal Risk: USDH is launching at a time of evolving stablecoin regulation. Bridge (Stripe) has MSB and money transmitter licenses, but stablecoins in the US may soon require special stablecoin issuer charters (the draft GENIUS Act is mentioned). Native Markets aims to be “GENIUS-compliant” via Bridge’s framework, but there’s uncertainty here: if US regulators decide that only banks or trust companies can issue stablecoins, how will USDH be classified? It could be forced to obtain additional licensing. Regulatory risk also includes enforcement actions: e.g., the NYDFS could theoretically say “Stripe’s stablecoin activities need a New York BitLicense or trust charter” and until that’s obtained, operations cease. Or if authorities suspected USDH was being used for illicit finance, they could compel the issuer to freeze certain funds or even freeze all activity. We saw something analogous when Paxos was ordered to stop issuing BUSD by NYDFS – that kind of sudden order could hit USDH if an issue arose. A particular risk is that Native Markets is a new entity without its own regulatory standing; they rely on Stripe’s licenses. If Stripe/Bridge ever had a compliance lapse or penalties, USDH could be indirectly impacted (for instance, Stripe might decide to off-board high-risk clients, etc.). Additionally, since Stripe is a US company, USDH is subject to US sanctions law – addresses or people on e.g. OFAC lists will be frozen. This introduces the risk of fund seizure: though rare, regulators could instruct the issuer to outright redeem or block certain tokens (as happened with some USDC linked to sanctioned groups). Such actions could dent confidence broadly.

  3. Operational Failure (Bridge Platform): Bridge as an API service could experience downtime or technical failure. If Bridge’s systems went offline due to an outage or cyberattack, minting and redemption would be impossible until service is restored. During that time, USDH would effectively become a closed system – still transferable on-chain, but no one could get new ones or cash out. Short outages might not be dire, but a prolonged one could break arbitrage with $1 and cause market price fluctuations (if people fear they can’t redeem for real USD, they might sell USDH at a discount). There’s also the risk of bugs in Bridge’s off-chain software – e.g., an accounting bug that mis-records a deposit or triggers an incorrect mint. If such a bug happened, it could result in an unbacked USDH being minted (if the system mistakenly thought money arrived) or conversely a user not getting USDH when they should. Bridge likely has reconciliation processes to catch mismatches (comparing bank receipts vs token mints daily). Still, operational risk can’t be zero – even big institutions sometimes double-charge or have outages.

  4. Security Breach (Cybersecurity): The off-chain infrastructure (Bridge’s servers, databases, private keys) could be targeted by hackers. A major concern is the compromise of the issuer’s private keys as mentioned in smart contract risk. If an attacker got hold of the minting key, they could create tokens. If they got hold of custodial account credentials, they might attempt to withdraw reserves (though banks have strong checks, it’s not impossible via fraudulent instructions). Bridge’s integration with many systems presents a broad attack surface. We have to consider insider threats too: an employee at Stripe with high-level access could, in theory, subvert controls. Stripe is known for good security, but as USDH grows, it becomes an attractive target. A specific scenario: a hacker could try to spoof API calls to the blockchain – perhaps forging a mint transaction if they found a vulnerability in how Bridge signs them. The mitigations likely in place include HSM-based key storage, multi-party approval for large operations, rate-limits, and monitoring (so a rogue large mint triggers alarms). While these reduce probability, the impact of such an event is so high that it remains a key risk to highlight.

  5. Economic and Peg Risk from Off-chain Events: If something goes wrong off-chain, it can manifest as a peg instability. For example, if Superstate’s on-chain Treasury fund (USTB) stops functioning or has a delay – say the fund can’t mint/redeem tokens for a time due to an SEC issue or technical bug – the portion of USDH backed by that might be temporarily illiquid. If that coincided with large redemption demands, there could be a shortfall in immediately available cash, causing redemption queues or limits. That scenario might cause USDH to trade below $1 until resolved (similar to how USDT or USDC have dipped when people worry about reserves). An expert cited in The Defiant noted that a hybrid reserve model could see peg volatility in early days if anything goes awry off-chain. For instance, if one of the reserve managers lags (imagine Superstate’s system is slow to process a sell order while lots of USDH selling is happening on-chain), arbitrage could weaken temporarily.

  6. Redemption / Liquidity Risk: Off-chain, they must maintain enough liquid assets to meet redemption. If USDH supply grows huge and a large portion tries to redeem at once (scenario: maybe a panic or a big arbitrage if USDC or others have issues), can Native handle it? They’d have to liquidate T-bills swiftly. Normally selling T-bills is easy, but if markets are closed (weekends) or in a severe crisis (e.g., debt ceiling standoff affecting T-bill liquidity), there might be short delays. If redemption is not instant or gets queued, arbitrageurs might start selling USDH at slight discounts, leading to peg strain. So far, we don’t know if Native Markets has arranged any credit lines or liquidity facilities or insurance fund for emergencies – some issuers do (Circle has lines of credit to ensure liquidity). Without that, they rely purely on asset liquidity.

  7. Lack of Insurance or Lender of Last Resort: Unlike a bank, stablecoin reserves aren’t insured by FDIC or similar (at best, some portion in bank accounts might be insured up to $250k, which is negligible relative to supply). If there’s a shortfall in reserves due to any reason (hack, custodial loss), there’s no automatic insurance covering users. Users are essentially unsecured creditors of the issuer for redemption. This risk is inherent in all stablecoins, but worth noting – trust is the only backstop. Hyperliquid’s Assistance Fund might serve as a quasi-backer (for instance, could Hyperliquid governance decide to use other treasury funds to fill a gap to protect their ecosystem’s reputation?), but that’s speculative and not guaranteed.

Many of these risks boil down to one thing: trust in the issuer and its partners to remain solvent, honest, and competent. This is why experts emphasize that stablecoins depend on trust, and Native Markets has yet to build a long track record. In contrast, Paxos or Circle have weathered minor crises and have institutional processes. Hyperliquid’s community essentially is betting that the benefits of alignment outweigh the increased operational risk of a newer issuer.

We have already seen a mild effect of these trust factors: upon launch, USDH held its peg well, but observers are keenly watching for any sign of trouble. If Bridge were to “freeze” or Superstate “lags” under high load, it “could cause problems” – meaning arbitrage could break down or users might panic. An example scenario: a sudden Hyperliquid market event leads to huge USDH demand for withdrawal, Bridge’s bank wires can’t go out until next business day – USDH might trade at $0.99 until those wires hit, just because of timing.

Mitigations in place include:

  • Using only very secure, short-duration assets (minimizes market and credit risk).
  • Keeping a healthy cash buffer for redemptions (hopefully Native is doing this).
  • Employing robust compliance to avoid regulatory breaches that could cause a shutdown.
  • Running 24/7 operations to handle issues (Stripe’s infrastructure presumably is 24/7, unlike a small startup).
  • The community and Native could establish a dashboard or proof-of-reserves to improve transparency, which would help quell rumors quickly if something seems off.

In summary, off-chain risks are significant and are the primary source of existential threat to USDH. A smart contract bug is less likely and more easily contained (since it can be fixed or paused), whereas a collapse in trust off-chain (due to lost reserves or inability to redeem) would directly translate to a loss of peg and utility. The decision to rely on centralized off-chain components means USDH inherits all the vulnerabilities of traditional finance (regulatory whims, banking system stability, human errors) on top of the need to maintain blockchain security.

Yet, these trade-offs were consciously made to achieve scale and integration – indeed, Hyperliquid’s bet is that a “permissioned” stablecoin can safely serve a largely permissionless trading platform. The coming months/years will test how well Native Markets can manage these off-chain risks. If they do it well, USDH could become a model of a new wave of exchange-native stablecoins capturing value for their ecosystems. If they falter, it will reinforce the importance of experienced operations and perhaps drive Hyperliquid to reconsider more established issuers or different designs.

The prudent approach for users is diversification: treat USDH as one of several stablecoins and watch for transparency reports. For the issuers, a recommendation is to over-collateralize or have emergency plans (even an informal backstop from Hyperliquid Foundation if needed) to bolster confidence. Remember that confidence is everything for a stablecoin – once lost, it’s hard to regain. Native Markets will need to continually demonstrate that every USDH is indeed as good as a dollar, under all conditions, through both technical rigor and transparent operations.

1 Like

beHYPE Risk Assessment by BA Labs

Part I: Protocol Overview

A.1 Protocol Details

A.1.1 Underlying Collateral

beHYPE is a liquid staking token fully backed by HYPE, the native token of the Hyperliquid blockchain. When users stake HYPE through Hyperbeat’s contracts, their HYPE is delegated to Hyperbeat’s validator on Hyperliquid (via the HyperCore staking engine). The staked HYPE remains in the protocol’s control on Hyperliquid (as the underlying collateral), while the user receives an equivalent amount of beHYPE tokens. This ensures that every beHYPE is always redeemable for a share of the underlying HYPE in the staking system. In essence, beHYPE represents a claim on staked HYPE, including any accumulated staking rewards. Hyperbeat’s validator operations secure the network and generate the staking rewards that increase beHYPE’s backing over time. Notably, Hyperliquid’s staking has no automatic slashing currently, which simplifies collateral management (though slashing could be introduced in the future).

A.1.2 Fee and Revenue Sources

The Hyperbeat protocol employs a few revenue mechanisms built into beHYPE’s design. First, an instant withdrawal fee (a percentage of the withdrawn amount) is charged for users who exit immediately without waiting for the unstaking period. This fee is set in basis points by the Protocol Guardian and represents revenue to the protocol’s treasury or liquidity providers (currently 30 bips is deducted while withdrawing which can be checked here by looking at instantWithdrawalFeeInBips). Second, Hyperbeat’s validator may earn a validator commission on staking rewards – Hyperliquid validators can charge up to 1% commission to delegators. This commission (if set by Hyperbeat’s validator) would skim a small portion of the staking yield as protocol revenue. Aside from these, there are no recurring management fees on the vault; all standard staking rewards accrue to beHYPE holders. Hyperbeat has also been incentivizing early users with “Hearts” points (over 10 million points distributed) as a one-time reward program. These Hearts do not dilute beHYPE’s collateral but serve as loyalty rewards potentially convertible to future governance tokens, indirectly promoting adoption of beHYPE. Overall, the primary monetary flows are the instant withdrawal fee and any validator commission, making the protocol’s revenue directly tied to user activity and staking yields.

A.1.3 Minting and Redemption Flow

Minting (Staking): To mint beHYPE, a user supplies HYPE tokens to the staking contract (called StakingCore). The StakingCore contract invokes the Hyperliquid CoreWriter system to delegate the HYPE to Hyperbeat’s validator on HyperCore, and then mints beHYPE to the user at the current exchange rate. The exchange ratio starts at 1:1 (1 beHYPE per 1 HYPE) and grows as staking rewards accrue. When the user calls the stake() function on StakingCore, the contract transfers the HYPE into the protocol’s control and immediately mints beHYPE to the user’s address. This entire process is permissionless for users and happens in one transaction.

Redemption (Withdrawals): To redeem HYPE, a user calls the withdraw() function on the WithdrawManagercontract and transfers their beHYPE tokens to it. Withdrawals can be Instant or Standard. For an instant withdrawal, the contract will immediately return HYPE to the user (burning the beHYPE) if the amount is within the available instant liquidity and rate limits. The user pays the instant withdrawal fee in this case (deducted from the returned HYPE). Under the hood, the WithdrawManager taps into a pool of readily available HYPE (or uses capacity from recent deposits) to fulfill the request. For a standard withdrawal, the request is recorded in a queue rather than paid out immediately. The beHYPE tokens are locked in the WithdrawManager, and the exact amount of HYPE owed is fixed at the time of the request using the current exchange ratio. The user then waits for the Hyperliquid unstaking period (typically 7 days) to elapse.

Finalization: After the waiting period (usually 3–7 days for standard withdrawals), an off-chain service or script signals that unstaked HYPE is available. A Protocol Admin must then call finalizeWithdrawals() on WithdrawManager. This function will burn the user’s beHYPE and transfer the corresponding HYPE out to the user’s wallet, completing the redemption. Finalization processes the queued withdrawals in batch, calling StakingCore to send HYPE for each entry and marking them finalized. If an instant withdrawal was done, finalization is not needed for that user (their transaction was already completed); however, the protocol internally accounts for that HYPE by adjusting its liquidity pool or capacity. The diagram below summarizes the stake and withdraw flows:

Figure 1: Overview of beHYPE minting and redemption flows. Users stake HYPE to mint beHYPE (flow 1), or burn beHYPE to withdraw HYPE via either instant withdrawal or queued standard withdrawal (flow 2). Off-chain steps (admin actions) are required to finalize standard withdrawals.

A.1.4 Fee Structure and Incentives

Hyperbeat has designed beHYPE’s economics to align incentives for both long-term stakers and liquidity-seeking users. The fee structure is simple: only users who utilize instant withdrawals pay a fee, whereas standard withdrawals are free (apart from waiting time). The instant withdrawal fee is configured as instantWithdrawalFeeInBps (basis points) by the Protocol Guardian. For example, if this fee were set to 50 bps, an immediate withdrawal of 100 HYPE worth of beHYPE would incur a 0.5 HYPE fee, leaving 99.5 HYPE returned to the user. This fee effectively goes to the protocol treasury or to replenish the instant withdrawal liquidity. By charging users who skip the queue, the system incentivizes patience – users willing to wait help maintain protocol stability, while impatient users compensate the system via fees.

The incentives for beHYPE holders include continuous staking rewards and enhanced token utility. As long as a user holds beHYPE, their stake is earning HYPE rewards automatically compounding (reflected as an increasing beHYPE-to-HYPE exchange rate). Unlike “vanilla” direct staking which locks HYPE for 7 days on unstake, beHYPE holders retain liquidity, able to trade or use beHYPE in DeFi immediately. To bootstrap adoption, Hyperbeat ran a pre-deposit campaign (the preHYPE vault) where early stakers earned Hearts rewards – 10 million Hearts (loyalty points) were allocated to early participants. These points (over 20% of the total Hearts supply) were distributed to give early users a stake in the ecosystem’s future token, compensating them for the initial bootstrap risk. The Hearts rewards and the promise of eventual cross-chain utility (beHYPE is planned to roll out on Ethereum mainnet for broader integrations) aim to drive demand for staking HYPE via Hyperbeat. Additionally, Hyperbeat’s validator operations mean beHYPE stakers indirectly contribute to network security and may benefit from any validator commission rebates or promotions Hyperbeat offers. Overall, the fee and incentive structure is designed such that: long-term holders benefit from compounding yields and potential airdrops, while short-term liquidity seekers pay a modest fee that supports the protocol.

A.1.5 Off-chain Components

While the core of beHYPE operates on-chain, a few critical components rely on off-chain processes and actors. Notably, withdrawal finalization is not autonomous – it requires the Protocol Admin (a trusted multisig in collaboration with ether.fi) to trigger the release of HYPE after the unstaking period. This implies an off-chain service or cron job monitors pending withdrawals and the Hyperliquid 7-day unstaking queue, then calls withdrawFromHyperCore and finalizeWithdrawals when funds are available. The off-chain role here introduces a liveness and trust assumption: users depend on the admin to promptly finalize withdrawals once possible. Hyperbeat mitigates this by using a robust multisig and by time-limiting withdrawals(e.g. aiming to finalize within 3–7 days as stated).

Another off-chain component is the exchange rate updater. Only Protocol Admin can call updateExchangeRatio() on StakingCore, and this call needs to be made periodically (e.g. daily or weekly) to record new staking rewards and increase the beHYPE/HYPE ratio. In practice, an off-chain agent (script or keeper) must trigger this function at appropriate intervals. The design includes safety checks (like acceptableAPR guard rails) to prevent malicious or accidental large jumps. The admin’s off-chain process should schedule updates such that the guard isn’t overly restrictive (e.g. avoid calling too soon after a previous call).

Finally, administrative keys represent an off-chain risk. The Protocol Admin and Guardian roles are presumably held by multisig wallets controlled by Hyperbeat’s team or trusted community members. These roles can change system parameters (fee rates, rate limits), perform emergency pauses, and even upgrade the contracts (there is an upgrade hook controlled by an authorized upgrader). Thus, the security of beHYPE is partly dependent on the operational security of these off-chain keys. Hyperbeat’s approach to this is to use multisig governance (minimizing single-point risk) and to keep most logic on-chain (the admin actions are transparent and limited in scope). In summary, off-chain components in beHYPE are kept to a minimum, but human coordination is required for reward updates and withdrawal finality, and trust is placed in the protocol’s admins to act in users’ best interests.

B. System Design

B.1 Monitoring and Observability

Uploading: Screenshot 2026-01-27 at 14.18.36.png…
Figure 2: Public monitoring dashboard for beHYPE

Hyperbeat has recently introduced a public monitoring dashboard for beHYPE, which significantly improves transparency and observability of the protocol. The dashboard exposes both system-level aggregates and validator-level details, giving users direct insight into the state of the liquid staking vault and the underlying validator set.

System-level metrics are prominently displayed, including:

  • Total Value Locked (TVL): Currently shown in USD terms, reflecting the dollar-denominated value of all HYPE staked through beHYPE.
  • HYPE Staked: Expressed both as a total token amount (e.g., 2.52M HYPE) and translated into TVL.
  • Annual Percentage Yield (APY): The realized staking reward rate for beHYPE holders, updated dynamically as Hyperliquid validator rewards accrue.
  • Number of Holders: The count of distinct addresses holding beHYPE, which serves as a proxy for user adoption and distribution.

Validator-level monitoring adds another layer of observability. The dashboard lists each validator in the active set, alongside:

  • The share of beHYPE stake delegated to them.
  • Their total validator stake (including delegations from outside beHYPE).
  • Uptime metrics (currently showing 100% across the set, indicating high operational reliability).
  • The validator’s current APR contribution.
  • Status indicators (ACTIVE/INACTIVE), allowing users to confirm live participation in consensus.

From a user’s perspective, the interface also provides staking flow transparency: when a user inputs a staking amount, the dashboard immediately displays the expected beHYPE received, the exchange rate (currently very close to 1:1), and network costs (gas fees in HYPE). For withdrawals, the dashboard provides separate tabs for Unstake and Instant Redeem, making clear the trade-offs between waiting for the Hyperliquid unstaking cycle versus paying the instant withdrawal fee.

Taken together, this dashboard reduces the opaqueness previously associated with liquid staking systems. Instead of relying solely on block explorers, users can now monitor both global protocol health and validator-level performance in a single interface. This observability layer enhances user trust by surfacing key risks (e.g., validator downtime, APR variability, liquidity constraints) in real time, and by aligning with best practices in transparent proof-of-stake protocols.

B.2 System Architecture

The beHYPE system is composed of a set of smart contracts with clearly defined responsibilities, interacting with the Hyperliquid blockchain’s dual execution layers. The core contracts are StakingCore, WithdrawManager, beHYPE (ERC-20), and RoleRegistry, supported by the Hyperliquid CoreWriter system contract and the HyperCore staking subsystem. The figure below (based on the audit’s communication diagram) illustrates these components and their interactions:

  • User – Holds HYPE and beHYPE. Initiates stake and withdrawal transactions.
  • StakingCore – The primary contract where users stake HYPE. It handles the logic for minting beHYPE and interacting with HyperCore. Upon stake(), StakingCore uses CoreWriter to delegate HYPE to the validator, then calls BeHYPE.mint() to issue tokens to the user. StakingCore also holds administrative functions like updateExchangeRatio() and bridging functions (depositToHyperCore, withdrawFromHyperCore, etc.). It keeps track of the exchange ratio and total protocol balance (summing EVM balance, HyperCore staking, and HyperCore spot liquidity).
  • beHYPE Token – An ERC-20 token contract representing the liquid staking token. It has standard ERC-20 features (transfer, permit) and can only be minted by StakingCore and burned by WithdrawManager under the proper conditions. It tracks balances of users but does not contain complex logic beyond mint/burn authorization.
  • WithdrawManager – Manages the withdrawal process. It receives beHYPE from users who want to redeem and either executes an instant withdrawal or queues the request. For instant withdrawals, it calculates the fee and dispenses HYPE immediately (drawing from available liquidity). For standard withdrawals, it records the request (user, amount, and fixed HYPE owed) in a queue. Later, when finalizeWithdrawals() is called by the admin, WithdrawManager will iterate through the queue and instruct StakingCore to send out the HYPE to each user, then burn the corresponding beHYPE. It enforces global rate limits: a capacity (maximum instant HYPE that can be withdrawn at once) and a refill rate that replenishes the instant withdrawal allowance over time. These ensure that the staking pool isn’t drained too quickly.
  • HyperCore & CoreWriter – HyperCore is the Hyperliquid consensus layer handling actual HYPE delegation and reward accrual. CoreWriter (at a fixed address 0x333…3333) is an internal contract that relays actions from HyperEVM to HyperCore. In this architecture, StakingCore uses CoreWriter to perform actions like delegate tokens to the validator, withdraw HYPE from HyperCore (after unstaking), and move funds between the staking account and spot account on HyperCore. These cross-component calls are authenticated by HyperCore (only the Hyperbeat contracts can manage the Hyperbeat validator’s stake). HyperCore keeps track of all staked HYPE and pending withdrawals, and it automatically handles reward distribution (auto-compounding into the validator stake) each epoch.
  • RoleRegistry and Admin Keys – All privileged operations are gated by roles. The Protocol Admin (a 3-of-6 multisig) can call critical functions: updateExchangeRatio, finalizeWithdrawals, and the HyperCore deposit/withdraw functions. The Protocol Guardian (a separate role, possibly also multisig) can adjust “guard rails” like the acceptable APR threshold for exchange rate changes and the instant withdrawal fee percentage. RoleRegistry also can pause the StakingCore or WithdrawManager in emergencies. These roles ensure that even if one contract’s function (e.g. withdraw) is externally callable, any sensitive action (like moving funds) requires the caller to hold the appropriate role.

Overall, the architecture follows a modular design: StakingCore for inbound staking and maintaining state, WithdrawManager for outbound flows and user-initiated exits, and the beHYPE token as the interface for DeFi composability. They interact with HyperCore to actually stake and unstake HYPE. Off-chain actors (admins) bridge the gap between these modules when time-based events occur (reward updates, withdrawal finality). This design was depicted in the audit’s communication diagram, illustrating how a user’s stake flows through StakingCore to HyperCore and back through WithdrawManager on withdrawal. The use of roles and a timelock/guardian pattern is intended to provide security: e.g., an admin must explicitly finalize withdrawals, preventing automatic drains. The potential drawback is added complexity and reliance on admin actions (as discussed in Part II). Nonetheless, the system architecture prioritizes security (through controlled roles) and flexibility (upgradeable contracts and parameter tuning) while leveraging Hyperliquid’s fast finality and dual-engine design to enable a permissionless yet controlled liquid staking protocol.

1 Like

beHYPE Risk Assessment by BA Labs

Part II: Technical Risk

A. Smart Contract Risk

The beHYPE protocol has undergone multiple security audits in which reviewers analyzed its Solidity code and cross-chain interactions. The findings from two independent audits – one by Nethermind Security and one by Certora – indicate that the contracts are generally well-designed but did include some vulnerabilities and design concerns. Table 1summarizes key issues identified in these audits, along with their severity and resolution status:

Table 1: Audit Findings for Hyperbeat’s beHYPE Contracts

Issue (Source) Severity Status Description
Malicious user can block withdrawal finalization (Nethermind) Critical Fixed :white_check_mark: A malicious contract could revert the payout and halt the withdrawal queue processing entirely. Patched by changing withdrawal finalization to avoid per-user external calls in a single loop.
Precision loss on cross-chain transfers causes payout shortfall (Nethermind) Low Acknowledged :warning: Converting HYPE between 18-decimal (EVM) and 8-decimal (HyperCore) can truncate amounts, possibly leaving a shortfall for the last withdrawing user. Team accepted this risk, noting they can perform an extra withdrawal to cover any remainder.
Instant withdrawal rate limit stricter than intended (Nethermind) Low Acknowledged :warning: The math/logic for enforcing instant withdrawal limits could prevent some legitimate withdrawals (config issue). Accepted as a minor issue.
“Just-In-Time” stake can capture rewards (Nethermind) Low Acknowledged :warning: An attacker could front-run an updateExchangeRatio() call by staking a large amount of HYPE, sharing in rewards they didn’t earn. This sandwich attack is feasible if rewards accrue over long periods. The team is aware and plans to update the ratio frequently to mitigate.
Withdrawal request amount fixed pre-slashing (Nethermind) Informational Acknowledged :warning: The HYPE amount for a standard withdrawal is fixed when requested, so a slashing event could leave the protocol unable to fully pay that amount. (Hyperliquid has no slashing currently, so considered acceptable).
finalizeWithdrawals lacked Checks-Effects-Interactions (CEI) (Nethermind) Best Practice Fixed :white_check_mark: The original finalizeWithdrawals loop made external calls (HYPE transfers) before updating all internal state, contrary to CEI pattern. This was cleaned up in code updates to prevent theoretical reentrancy issues.

Certora Audit Findings:

Issue (ID by Certora) Severity Status Description
L-01: Truncation loss in depositToHyperCore Low Fixed :white_check_mark: Sending HYPE to HyperCore truncated to 8 decimals could lose tiny fractions of HYPE. Fix: validate amount’s lower 10^10 portion is zero before transfer.
L-02: Small withdrawals could be “dust-locked” Low Acknowledged :warning: If a user’s beHYPE balance is below minWithdrawalAmount, they can’t withdraw on-chain and must use a DEX to exit, which is an inconvenience. Accepted to prevent spam withdrawals.
L-03: Missing slippage parameter in withdraw Low Fixed :white_check_mark: Without a min-out parameter, a user’s instant withdrawal might execute after an exchange rate update and return less HYPE than expected. Fix: added a minHypeOut to protect against exchange rate changes mid-transaction.
L-04: Non-CEI pattern in withdrawal finalization Low Fixed :white_check_mark: Same issue as Nethermind’s CEI finding – external calls in loop before state updates. Fixed by burning beHYPE in-loop before sending HYPE.
L-05: L1 block latency can skew exchange rate Low Fixed :white_check_mark: If updateExchangeRatio is called in the same block as a large deposit or withdrawal, it might read stale HyperCore data. Fix: enforce a minimum delay (blocks) after cross-chain ops before updating the ratio.
I-01: Missing events on parameter changes Info Fixed :white_check_mark: Several admin functions (setting fees, limits) did not emit events, reducing transparency. The team added events for these state changes.
I-02: Wrong revert message on auth check Info Fixed :white_check_mark: A require in withdrawFromHyperCore incorrectly used NotAuthorized error for an excess amount check (misleading). Corrected to a more appropriate error message.
I-03: Inconsistent parameter names (nits) Info Fixed :white_check_mark: Some function parameters were misnamed (_minStakeAmount that actually meant _minWithdrawAmount, etc.). Renamed for clarity.
I-04: Redundant check in withdrawFromHyperCore Info Fixed :white_check_mark: The check “amount <= hypeRequestedForWithdraw” could be bypassed by multiple small calls and didn’t account for already available funds. The team removed this check to simplify logic (since Protocol Admin will manage amounts properly).
I-05: Withdrawal fee rounds down (protocol loses dust) Info Acknowledged :warning: The fee calculation uses floor division, so a few wei of HYPE fee might be uncollected (leaving slightly more to the user). This minor loss was deemed acceptable for simplicity.
I-06: Vault inflation attack via donation Info Acknowledged :warning: An attacker could donate a large amount of HYPE to the contract without minting beHYPE, inflating the exchange rate artificially when admin next updates it. Because updates are admin-only and guarded by an APR limit, the impact is minimal and accepted as a theoretical risk.
I-07: Missing input validation (sanity checks) Info Acknowledged :warning: Certain functions didn’t check for address(0) assignments or extremely high fee values, relying on the admin’s care. The team acknowledged this, trusting their deployment process and multisig governance to prevent misconfiguration.
I-08: APR guard sensitivity to short intervals Info Acknowledged :warning: If updateExchangeRatio is called very frequently, the annualized APR change calculation could be exaggerated (small time delta). This is a note for operational caution – ensure updates are scheduled at sensible intervals.
I-09: Anyone can call claimWithdrawal for others Info Acknowledged :warning: The design allows any address to call the claimWithdrawal()function to finalize a user’s standard withdrawal (after it’s been finalized by admin). This could interfere with protocols expecting only the user to claim; however, Hyperbeat intentionally allows it for “batched” or delegated claims, accepting the risk of confusion.
I-10: Emergency undelegation restricted by queue check Info Fixed :white_check_mark: An added guard in withdrawFromStaking() prevented the admin from emergency withdrawing more HYPE from HyperCore than the pending queue amount. This could hinder emergency response. The team adjusted this in an update to allow a bypass in true emergencies.

Key Takeaways: Both audits found no critical vulnerabilities remaining after fixes; most issues were Low-severity or informational. The Nethermind audit highlighted a critical Denial-of-Service vector in withdrawal finalization, which was promptly fixed by changing the withdrawal logic to a “pull” model (avoiding in-loop external calls). The majority of Nethermind’s other findings relate to edge-case scenarios and protocol design choices: for example, handling of donations, precision loss due to different decimal systems, and the fact that withdrawals don’t account for slashing by design. These were acknowledged as known trade-offs rather than immediate threats.

The Certora audit (combined with formal verification) corroborated many of these and added a number of low-level improvements. Importantly, the Certora team used formal methods to reason about scenarios like the “vault inflation attack” (donation of assets) and concluded the risk was “extremely limited” by the admin-only update and exchangeRateGuard. Certora’s findings led to several fixes that harden the protocol: adding slippage protection, block-delay for exchange rate updates, more event emissions, and removing a redundant check that could prevent liquidity management. All of Certora’s low findings were addressed, and the informational ones were either fixed or consciously left as acceptable risks (with reasoning documented by the team).

In summary, Smart Contract Risks for beHYPE have been reduced via these audits to mostly design-level nuances. After patching, the contracts do not contain known reentrancy, math overflow, or access control bugs. Remaining risks lie in areas like: the timing of updates (the admin must operate correctly to avoid JIT attacks or stale reads), the assumptions (no slashing events, donation unlikely) and intentional limits like minWithdrawalAmount which could inconvenience small holders. These are non-exploitable by attackers in the current context, but they highlight that the protocol’s safety and efficiency depend on proper governance and operation. Both auditors rated the code quality and documentation highly, and with the critical logic fixed, the smart contracts themselves are considered robust.

B. Off-chain Component Risk

Off-chain components introduce trust and operational risks that are distinct from the on-chain code vulnerabilities. In Hyperbeat’s beHYPE, the primary off-chain dependencies are the admin roles and processes rather than external price oracles (since the system mostly uses on-chain data). We outline the main off-chain risks:

  • Protocol Admin and Guardian Keys: The Protocol Admin has sweeping control over core functions – they alone can trigger reward accrual updates and finalize withdrawal queues. If an Admin key is compromised or a malicious actor gains control, they could disrupt the system. For instance, a rogue admin could halt withdrawals (by not calling finalize, or by pausing contracts via RoleRegistry) which would freeze user funds. They could also set extreme fee parameters (Protocol Guardian could set an unreasonably high instant withdrawal fee, effectively stealing a portion of withdrawals). The audits specifically mention that some safety checks (like a maximum cap on the fee in bps) were not enforced in code, with the team instead relying on multisig governance and manual review to prevent abuse. Thus, users must trust that the Admin/Guardian multisigs are secure and managed by honest actors. The risk is mitigated by using a secure multisig (making unilateral malicious actions less likely) and by transparency – any parameter changes or strange admin behaviors would be visible on-chain, likely sparking community intervention.
  • Operational Liveness: The design assumes the admins will promptly perform periodic tasks. Staking reward updates via updateExchangeRatio() must be called regularly; if the Admin fails to do so for an extended period, beHYPE’s exchange rate would lag behind the actual accrued HYPE. This could make the system vulnerable to a JIT attack (because a large amount of rewards would accrue unaccounted, making it profitable for an attacker to stake just before an update). It would also mean beHYPE holders aren’t seeing their token appreciating as it should, undermining confidence. Similarly, withdrawal finalizations require the admin to act once the 7-day unstaking period is over. If they delay, users face extended illiquidity. In an extreme case, if Admins became unresponsive or lost keys, the queued withdrawal funds could be locked indefinitely (there is no trustless timeout mechanism for users to withdraw themselves). This liveness risk is common to validator-based staking systems – effectively, Hyperbeat operates a semi-manual bridge between HyperCore and HyperEVM. The team has acknowledged these timing sensitivities (e.g., ensuring not to update too fast or too slow), so robust monitoring and runbooks are needed off-chain. Community oversight is also a factor: since all actions are on-chain, if an update or finalize is overdue, users can publicly flag it.
  • Upgradeability and Change Management: The RoleRegistry includes an onlyProtocolUpgrader role that can authorize contract upgrades. This suggests the contracts are upgradeable (EIP1967 proxy with UUPS pattern). While upgradeability is useful for fixing bugs (indeed, the team did one round of upgrades during audits as seen in commit hashes), it is an off-chain risk. A malicious or compromised upgrader could deploy a new implementation that, for example, siphons funds. Even if not malicious, an upgrade could introduce a bug if not carefully audited. Hyperbeat will need stringent off-chain governance (perhaps time-locked upgrades, community review) to mitigate this. The Certora audit’s recommendation to add sanity checks on new configurations underlines that human error in upgrades or parameter tuning is a concern; the team’s response is that they have a “strict review process” for any such changes. Trust in that process is part of the risk profile for users.
  • External Dependencies: Uniquely, Hyperbeat relies on the Hyperliquid network itself as a dependency. If Hyperliquid’s consensus or HyperCore were to fail, beHYPE would be affected (e.g. if HyperCore stopped producing blocks, withdrawals couldn’t finalize, and staking rewards would pause). However, this is a broader blockchain risk not specific to Hyperbeat’s design. More specifically, beHYPE does not rely on external price oracles for its core logic – the exchange rate is computed from on-chain balances, and the withdrawal logic uses internal accounting. This reduces the risk of oracle manipulation or off-chain price feeds. One external integration planned is using beHYPE on other chains (like Ethereum mainnet via bridging); that will introduce typical cross-chain risks (bridge security, liquidity on other chains), but those are beyond the scope of the core protocol risk on Hyperliquid.
  • Economic Risks and Liquidity: Some risks are off-chain in nature but have on-chain impact – for example, liquidity for instant withdrawals. The protocol sets an instant withdrawal capacity and a refill rate, but it’s essentially up to the Hyperbeat team to seed and manage that liquidity. If many users try to exit and the capacity is depleted, Hyperbeat might choose off-chain to inject more HYPE (perhaps from treasury or new stakers) to honor further instant withdrawals. If they do not, users will fall back to the 7-day standard path. Then the risk of “Just-in-time” staking reward capture becomes higher (leverage long beHYPE in a platform like Hyperlend, then immediately unwind using the available instant liquidity for redemption). We indeed saw similar behavior happening in wstHYPE, which at one point had 21M USD worth of HYPE available for redemption.

In conclusion, the off-chain component risks revolve around governance and operations. Hyperbeat has attempted to minimize these by keeping core logic on-chain and simple (no complex oracles, no custodial key holding of user funds – users mint/burn themselves). Nonetheless, users must trust the Hyperbeat team’s multisig to act diligently and honestly. The audits explicitly note scenarios where the team’s actions are the mitigating factor (for example, donation attacks being limited by admin control of ratio updates, or emergency withdrawal needing a bypass which they implemented). Any centralized failure (compromised admin, insider misuse, or even just negligence) could lead to outcomes ranging from temporary DOS of withdrawals to, in the worst case, loss of funds (through malicious upgrade). These are standard risks for a protocol in its early stages with admin keys. Over time, one would expect Hyperbeat to further decentralize these components – for instance, handing over controls to a governance DAO or implementing trust-minimized automation for withdrawal finalization – to reduce the off-chain risk surface. For now, the design and audits suggest the team is aware of these issues and has put basic safeguards (multisigs, transparent on-chain records) in place, but the ultimate security of beHYPE depends on vigilant off-chain governance in tandem with its audited smart contracts.

1 Like

Unit Protocol Risk Assessment by BA Labs

Part I · Protocol Overview

A. Protocol Details

A.1 Protocol Description

Unit Protocol is a cross‑chain asset‑tokenization layer purpose‑built for Hyperliquid L1. A three‑member Guardian Network employs 2‑of‑3 MPC (secp256k1) to escrow native BTC and ETH (with SOL forthcoming) in on‑chain multisig vaults. Once sufficient confirmations accumulate on the source chain, the Guardians co‑sign a transaction that mints an equal number of ERC‑20‑style uAssets on Hyperliquid. The design is intentionally minimalistic: every uAsset is fully backed by its underlying coin at a 1 : 1 ratio; there are no yield strategies, rehypothecation loops, or algorithmic stabilisers that might introduce endogenous risk.

A.1.1 Underlying Collateral

Each vault address—Taproot multisig for BTC, externally owned multisig for ETH—is published in the Key Addresses section of the documentation, permitting public, on‑chain verification of collateral balances. Guardians periodically execute cold‑sweep transactions to consolidate small UTXOs, thereby improving gas efficiency. An on‑chain circuit‑breaker allows any Guardian to halt further minting or burning if the reported vault balance diverges from internal accounting.

A.1.2 Asset Status & Parameters (April 2025)
uAsset Status Source chain Vault type Min deposit Mint finality*
uBTC Main‑net (since Jan 2025) Bitcoin P2TR 2‑of‑3 MPC 0.002 BTC 2 BTC blocks + 10 HL confs
uETH Main-net Ethereum EOA 2‑of‑3 MPC 0.05 ETH 14 ETH blocks + 10 HL confs
uSOL In development Solana Multisig (TBD)

*Elapsed from first source‑chain confirmation to ERC‑20 liquidity on Hyperliquid.

A.1.3 Minting & Redemption Flow

The movement of value across chains follows a two‑stage protocol pipeline: (i) consensus finality on the source chain establishes that the deposit is irreversible; (ii) Guardian consensus—expressed via a 2‑of‑3 MPC signature—authorises the state transition on Hyperliquid. Only after both checkpoints does the economic ownership of the underlying coin transform into a uAsset claim.

Illustrative operational steps

  1. Deposit address generation — A user requests a fresh deposit address from the public API; the request is routed through the relay to all Guardians, which deterministically derive the same address.

  2. Source‑chain transfer — The user sends at least the minimum amount to the generated address.

  3. Confirmation window — Guardians wait for 2 Bitcoin blocks (≈ 20 min) or 14 Ethereum blocks (≈ 3 min) to guard against re‑orgs.

  4. MPC signing — Once the threshold is met, any two Guardians jointly sign a mint message; the third Guardian later appends its signature for audit completeness.

  5. Hyperliquid finality — The mint transaction requires 10 Hyperliquid confirmations before the uAsset becomes transferable.

  6. Redemption mirror‑flow — A burn request undergoes the same checkpoints in reverse, with an intermediate 5 000‑block HL (≈ 6 mins) lock‑up before the withdrawal batch is executed on the native chain.

A.1.4 Fee Structure & Incentives

Unit Protocol adopts a pure pass‑through economic model. The only costs borne by users are the gases intrinsic to the originating and destination chains. Specifically, the destinationFeeAmount covers the on‑chain fee for the final leg of the transaction, while sweepFeeAmount estimates the cost of vault consolidation. No spreads, performance fees, or token incentives are levied; consequently, the protocol accrues no revenue.

A.1.5 Guardian Operations & Node Management

The Guardian Network—currently composed of Unit Labs, Hyperliquid, and Infinite Field—operates a homogenised software stack known as the Agent. Each Agent instance runs inside an AWS Nitro Enclave, isolating the MPC key‑share from the host kernel and from other cloud tenants. A stateless relay mediates encrypted message exchange, ensuring that no single Guardian can unilaterally advance state. The deterministic Agent state machine indexes the relevant blockchains, enforces rate limits, and invokes circuit‑breakers when invariant checks fail. Although this architecture minimises operational variance across Guardians, it also introduces a single‑implementation risk: a latent bug or vulnerability in the closed‑source Agent binary would propagate uniformly across the quorum.

A.1.6 Guardian Selection & Governance

Guardian membership is presently permissioned: the founding team at Nexus Research Labs, Inc. appointed the initial three operators and retains unilateral authority to add, rotate, or remove members. The documentation outlines an aspirational roadmap to expand the quorum to a 5‑of‑7 threshold, but provides no formal criteria, bonding requirements, or timeline. Broader governance of Unit Protocol—including contract upgrades, parameter tuning, and incident response—likewise resides with the corporation rather than a token‑holder DAO. Decisions are therefore channeled through a traditional corporate framework, subject to Delaware law and the binding‑arbitration clause embedded in the Terms of Service. This structure streamlines accountability but reinforces centralisation risk.

B. System Design

B.1 Bridging Finality & Latency (median April 2025)

Path Source‑chain confirmations HL confirmations Wall‑clock latency
BTC → uBTC 2 10 ~25–30 min
ETH → uETH 14 10 ~4–5 min
HL → BTC 5 000 HL 2 BTC blocks ~6 min HL + 20 min BTC

B.2 Monitoring & Observability

While the public REST API exposes per‑operation state transitions, Unit provides no real‑time dashboard detailing Guardian uptime, signature latency, or quorum‑failure counts. Security logs and audit trails remain private, leaving external stakeholders unable to quantify operational performance or service‑level breaches.

B.3 System Architecture (Textual Overview)

  • Guardian Agents — off‑chain MPC nodes indexing both chains.

  • Relay Server — stateless message bus for encrypted Guardian traffic.

  • Vaults — MPC vaults holding backing assets.

  • Gateway Contracts — minimal ERC‑20 mint/burn logic on Hyperliquid (code closed‑source).

  • Public API — endpoints for deposit‑address generation, operation tracking, and withdrawal‑queue status.

Part II · Market & Liquidity Risks

A. Price Stability

Since launch, uBTC/BTC has maintained a very volatile peg (up to 2% deviation at times) in the HyperEVM ecosystem. This tight band is sustained primarily by the deterministic redemption pathway: arbitrageurs can lock a costless basis by purchasing discounted uBTC and redeeming for native BTC. The mechanism, however, is only as robust as the Guardian Network. A prolonged halt—whether from collusion, cyber‑intrusion, or regulatory freeze—would sever the arbitrage loop and could precipitate a more pronounced de‑peg. Additionally, elevated Bitcoin network fees inflate redemption costs and delay withdrawal settlement, intermittently widening the discount during fee spikes. We can already see that the 25-30 min waiting duration to bridge already makes it very difficult to keep uBTC/BTC feed less volatile.

B. Liquidity

Depth for uAssets is heavily concentrated on Hyperliquid. Absent meaningful liquidity on centralised exchanges or cross‑chain DEXs, price discovery and arbitrage capacity are effectively siloed. Redemption throughput is further bounded by protocol‑level batching: the withdrawal queue executes every ~3 BTC blocks (~30 min) or ~21 ETH slots (~4 min). During stress events, exit demand could exceed this capacity, elongating queues and magnifying slippage. Finally, minimum deposit sizes (0.002 BTC; 0.05 ETH) impose granularity constraints that may complicate liquidation cascades or portfolio rebalancing for smaller actors.

Part III · Technical Risk

A. Smart‑Contract Risk

Unit’s Hyperliquid gateway contracts are closed source; only their ABIs are published, and no independent security audits have been released as of April 2025. Uncertainty persists regarding upgrade mechanisms and admin‑key custody. The absence of a public bug‑bounty program further diminishes the incentive for white‑hat scrutiny. Taken together, these factors mean that critical invariants—such as the requirement that minted supply never exceed vault reserves—are unverifiable by external analysts.

Another source of concern is that, the smart contracts are not publicly available, and doesn’t seem to be have an open smart contract audit. It could be speculated that the reasoning for such a decision by Hyperunit team could be due to the fact that they might be using custom pre-compile functions that are not known. Which is a textbook case of security by obscurity.

B. Off‑Chain Component Risk

By design, any two Guardians constitute a quorum, enabling them to mint unbacked uAssets or drain vaults. Although AWS Nitro Enclaves harden key material against many threat vectors, they do not eliminate insider risk, social engineering, or supply‑chain compromise. Moreover, the Guardians all execute the same proprietary Agent binary; a logic error or latent vulnerability therefore carries systemic impact. Operational resilience is likewise dependent on the relay’s availability: a sustained relay denial‑of‑service would pause bridge activity, and fallback pathways remain undocumented. Compounding these exposures, the lack of public service‑level telemetry hampers early detection of anomalies, increasing the time‑to‑response during incidents.

Part IV · Consolidated Risk Narrative

Unit Protocol succeeds in delivering a deterministic, low‑cost bridge of BTC, ETH, and (prospectively) SOL onto Hyperliquid with complete on‑chain collateralisation. Nevertheless, the system’s security guarantees hinge on a small, non‑transparent operational core. The centralised Guardian quorum, opaque codebase, and limited exit bandwidth constitute the principal vectors through which peg integrity might be compromised during market stress. Until the protocol achieves Guardian decentralisation, publishes audited open‑source contracts, and diversifies external liquidity venues, participants must accept a regime in which counterparty, technical, and liquidity risks remain materially above those of native asset custody.

1 Like

uSOL Risk Assessment by BA Labs

Part I: Protocol Overview

A.1 Protocol Details

Unit Solana (uSOL) is a wrapped representation of Solana’s native token (SOL) on the Hyperliquid platform. It is issued by the Hyperunit protocol via a lock-and-mint bridging mechanism. In essence, when users deposit SOL from the Solana blockchain into Hyperunit, an equivalent amount of uSOL is minted on Hyperliquid for use in trading and DeFi, and when users withdraw, uSOL is burned and the corresponding SOL is released back on Solana. The design of uSOL emphasizes a fully collateralized peg – every uSOL in circulation should be backed 1:1 by actual SOL held in reserve.

A.1.1 Underlying Collateral

The underlying collateral for uSOL is native SOL tokens held in a designated treasury account on the Solana blockchain. Whenever a user deposits SOL through Hyperunit, those SOL coins are locked in a Solana address controlled by the protocol’s guardian network, and an equivalent amount of uSOL is created on Hyperliquid. According to Hyperunit’s documentation, the main Solana treasury address for uSOL on Solana mainnet is publicly known (9SLPTL…KpKS, as stated in their docs). This transparency allows anyone to verify the total SOL reserves on Solana that back the uSOL tokens in circulation. In other words, uSOL is fully collateralized by SOL – it is not an algorithmic or fractional stablecoin, nor a synthetic derivative. It represents a claim on real SOL, held in a secure, multi-signature-like custody (via threshold signature) by the Hyperunit guardians.

Importantly, Hyperunit’s architecture is designed to avoid single-custodian risk. Unlike some wrapped tokens that entrust reserves to a single entity, uSOL’s reserve of SOL is managed by a distributed set of operators (Guardians) who must jointly authorize any movement of those funds. Hyperunit emphasizes that it is “neither custodian-based nor reliant on synthetic assets,” but rather a native tokenization layer bridging SOL into Hyperliquid. This means users maintain self-custody in the sense that no centralized exchange or third-party holds their SOL; instead, a decentralized protocol (the guardian network) locks the SOL on-chain and issues uSOL. As a result, each uSOL is always redeemable for 1 SOL, assuming the system operates correctly and guardians faithfully hold the collateral. The known collateral addresses and on-chain verifiability of reserves provide some assurance that the uSOL supply is fully backed at all times (users can observe the Solana address balance to confirm that it matches total uSOL outstanding).

A.1.2 Fee and Revenue Sources

Hyperunit does not charge explicit fees for minting or redeeming uSOL beyond the necessary network transaction costs. In other words, the protocol’s goal is to provide the bridge at minimal cost to the user, only passing along the underlying blockchain fees. When a user performs a deposit or withdrawal, they are responsible for the standard transaction fees on the source and destination blockchains (e.g. Solana network fee, Hyperliquid network fee), but no additional fee or spread is taken by Hyperunit itself. The documentation explicitly states that “Unit does not collect revenue from deposits or withdrawals – we aim to offer the lowest cost possible. The ‘fees’ associated with an operation are those required to process the transactions on their respective networks.”.

This fee structure implies that Hyperunit’s revenue model is not based on bridge tolls. Instead, the incentive for the Hyperliquid ecosystem is indirect: by making it cheap and easy to bring assets like SOL onto Hyperliquid (as uSOL) for trading, the platform can attract higher trading volume and user activity, which in turn generates revenue through trading fees on the exchange. For the users, the only costs incurred during a deposit or withdrawal are: (a) the fee to send SOL on Solana to the deposit address, and (b) the fee for the guardians to send the redemption transaction on Solana (for withdrawals). These fees are typically minor network fees; for example, current fee estimates show only a few million lamports (fractions of a cent in SOL) are required for Solana transaction processing. Hyperunit provides an API endpoint to estimate the prevailing fees on each network, so users know the expected cost and timing for deposits/withdrawals.

Because Hyperunit itself does not impose extra charges, there are no direct revenue streams like seigniorage or profit from uSOL issuance. All uSOL are issued and redeemed at parity (1 uSOL for 1 SOL), and the protocol’s objective is to minimize friction. This fee-less model can be seen as an incentive for users to utilize uSOL, as they can move assets in and out without worrying about heavy bridge fees. It’s worth noting that Hyperliquid may run trading reward programs or loyalty point schemes (e.g. users can earn Hyperliquid points for activity), which could indirectly incentivize bringing assets like SOL into the ecosystem, but the bridge itself remains simply a utility and not a profit center.

A.1.3 Minting and Redemption Flow

Minting uSOL (Deposit Flow): To mint uSOL, a user deposits SOL from Solana into Hyperunit’s system. This process is initiated by the user in the Hyperliquid or Hyperunit interface by selecting Solana (SOL) and generating a unique deposit address. Hyperunit’s guardian network uses a threshold signature scheme to derive a unique Solana deposit address tied to the user’s Hyperliquid account. The user then transfers their SOL to this deposit address, just like sending a regular SOL transaction. Once the deposit transaction is detected by the guardians, it goes through a confirmation and verification process. The protocol waits for a sufficient number of Solana confirmations (currently 32 blocks on Solana) to ensure the deposit is final and cannot be reverted. Each guardian independently monitors Solana’s blockchain (via their own Solana node/indexer) and validates that the deposit to the correct address has occurred and is irreversible.

After finality, the guardians coordinate to mint the corresponding uSOL on Hyperliquid. On Hyperliquid’s side (the destination), a transaction is constructed to credit the user’s Hyperliquid wallet with an equivalent amount of uSOL. All guardians perform additional checks (e.g. ensuring the source address is not sanctioned or blacklisted) before approval. The guardians then reach consensus and collectively sign the mint transaction using a 2-of-3 threshold signature (MPC) scheme. Once the threshold signature is produced, the mint (deposit credit) transaction is broadcast to the Hyperliquid network. Hyperliquid itself is a high-throughput chain; however, Hyperunit still enforces a confirmation period on Hyperliquid (currently 10 Hyperliquid blocks) to consider the operation finalized. After this finalization, the user’s Hyperliquid balance shows the new uSOL, and the deposit operation is marked complete. At this point, the user effectively has an IOU for their original SOL, now in the form of uSOL, which can be traded or used on Hyperliquid as if it were native SOL.

Behind the scenes, after a deposit is credited, Hyperunit will often sweep the deposited SOL from the one-time deposit address into a central treasury address for better security and management. This means the SOL originally sent by the user doesn’t remain isolated at the deposit address indefinitely; it is consolidated (via another on-chain transaction) into the main Solana reserve address controlled by the guardians. This sweep is also performed by the guardians under consensus and ensures that the collateral is held in a small number of secure addresses. The sweeping process does not affect the user’s credited uSOL, but it optimizes custody on the Solana side.

Redemption of uSOL (Withdrawal Flow): Redeeming uSOL for SOL is essentially the reverse process. The user initiates a withdrawal by specifying a Solana recipient address (where they want the SOL delivered) and the amount of uSOL to redeem. On Hyperliquid, the user will execute a transfer of uSOL to a designated withdrawal address or smart contract that signals the guardians to process a withdrawal. The Hyperunit system identifies a withdrawal request by detecting this on-chain event in Hyperliquid (a transfer of uSOL to the Unit protocol’s withdrawal account). Just like deposits, withdrawals require confirmations on the source side (Hyperliquid chain) – Hyperunit currently waits for 5000 block confirmations on Hyperliquid to ensure the withdrawal request is well-finalized and not subject to reorgs. Once finalized, the guardians prepare the corresponding SOL transaction on the Solana network: this will transfer the specified amount of SOL from the treasury to the user’s provided Solana address.

However, to manage throughput and network load, Hyperunit employs a withdrawal queue and batching mechanism for many blockchains. Solana has fast block times, but for consistency Hyperunit may still queue Solana withdrawals in timed batches (the documentation explicitly mentions batching every ~3 blocks for Bitcoin and ~21 slots for Ethereum; Solana’s interval would be analogous, likely a short interval given Solana’s speed). When the withdrawal comes up in the queue, the guardians collectively sign the Solana transaction releasing the SOL. The 2-of-3 MPC signature is applied to the Solana treasury’s private key shares, so that a valid Solana signature is produced without any single guardian ever holding the full key. The SOL is then sent on-chain to the user’s Solana address, and the guardians mark the uSOL withdrawal as completed.

During this process, Hyperunit subtracts the Solana network fee for the transaction from the withdrawn amount (or otherwise ensures the fee is covered), since the user must ultimately pay for the on-chain execution. In practice, the SOL withdrawal fee is tiny (~0.000008 SOL as of recent estimates). After broadcasting the SOL transaction, Hyperunit will also update the uSOL supply (burning the uSOL that was redeemed). The user should then see the SOL appear in their wallet on Solana after the standard Solana block confirmations (Solana confirmations are near-instant finality due to its design, so effectively the user receives SOL almost immediately once guardians broadcast).

Both deposit and withdrawal flows are orchestrated by a deterministic state machine running off-chain within the guardian network. Each step (detection, waiting for confirmations, signing, broadcasting, etc.) is performed in order by all guardians, ensuring that at least a majority have validated every condition before proceeding. This design minimizes race conditions and ensures consistency (e.g., a withdrawal will not be signed unless the corresponding uSOL burn on Hyperliquid was seen by all guardians, etc.). If anything abnormal is detected (such as mismatched amounts, or a potential double spend), guardians will not sign and can even trigger safety halts (see circuit breakers in Section B.2).

In summary, minting involves locking SOL and minting uSOL on Hyperliquid, and redemption involves burning uSOL and releasing SOL on Solana. Both operations are atomic from the user’s perspective (deposit → receive uSOL; redeem → receive SOL) but are underpinned by a robust cross-chain coordination protocol to ensure security and correctness at each stage.

A.1.4 Fee Structure and Incentives

Fee Structure: As noted above, Hyperunit’s fee structure is very straightforward – it is essentially fee-free aside from mandatory network costs. Users do not pay any spread or commission to Hyperunit when converting between SOL and uSOL. Instead, the user only pays the on-chain transaction fees for the deposit and withdrawal operations. For a deposit: the user pays the Solana transaction fee to send SOL into the deposit address (this is paid in SOL to Solana validators). For a withdrawal: the user effectively pays the Solana network fee for the guardians’ transaction out of the reserve (this may be deducted from the withdrawn amount or requires the user to leave a tiny bit to cover it). Additionally, on the Hyperliquid side, the deposit and withdrawal trigger transactions on Hyperliquid’s ledger — Hyperliquid’s chain currently has negligible fees (or the protocol might subsidize them) so this cost is minimal. Hyperunit provides an API to estimate fees and processing times on each supported chain, reflecting current network conditions. For instance, if Solana is congested or Bitcoin fees spike, the API would show higher fees and perhaps longer estimated times. This transparency helps users plan their transfers.

Because Hyperunit’s own fee is zero, there is no direct monetary incentive extracted from users at the bridge level. This aligns with Hyperliquid’s strategy: the presence of assets like uSOL on the exchange boosts trading volume and platform liquidity, which indirectly benefits Hyperliquid (trading fees, ecosystem growth) without needing to nickel-and-dime the deposit/withdrawal process. It is a user-friendly approach that lowers the barrier for on-boarding capital.

Incentives: Currently, there are no native token incentives or yield rewards specific to holding or using uSOL – it is intended to track SOL 1:1, not to provide an extra yield or staking reward. (Notably, uSOL itself is not a staked SOL derivative; it’s a liquid, instant redeemable asset, so it does not earn Solana staking returns.) The primary incentive for users is the utility uSOL provides: it enables participation in Hyperliquid’s high-performance on-chain order book and DeFi ecosystem using SOL value, with the ability to easily go back to native SOL. In a sense, uSOL inherits whatever incentives Hyperliquid offers for trading and liquidity provision. For example, if Hyperliquid has reward programs (like trading competitions or point systems for active traders), users bringing SOL into Hyperliquid (as uSOL) to trade can indirectly benefit from those. In one example from a DeFi strategy built on Hyperliquid, participants depositing uSOL to a yield strategy also accrued HyperUnit points as a reward, highlighting that active use of uSOL can tie into Hyperliquid’s broader incentive framework. These points or rewards are part of Hyperliquid’s exchange incentive program, not a property of uSOL itself.

From the protocol’s perspective, guardians and operators are incentivized to maintain the system’s integrity because Hyperunit’s success drives Hyperliquid’s success. The initial guardians include the Hyperliquid team and partners (discussed later), who have a vested interest in the platform growth rather than direct fee revenue. It’s possible that in the future, if the guardian network decentralizes further, there could be a fee model or a token to reward guardian operators for their service. As of now, however, no such token or fee exists; the service is run presumably by stakeholders funded through other means (e.g., Hyperliquid’s revenues or investment backing).

In summary, Hyperunit’s fee structure is extremely low-cost for users, and the incentives are aligned with increasing Hyperliquid usage rather than extracting fees. This user-aligned model encourages more SOL holders to bring liquidity on-platform via uSOL, which benefits the overall ecosystem (even if the bridge itself isn’t a profit center). It’s a trade-off where security and trustworthiness of the bridge must remain very high, since the protocol isn’t compensated per transaction; any loss of trust (e.g., an incident) would harm Hyperliquid’s reputation and volumes, which is presumably a far greater cost.

A.1.5 Off-Chain Components

While the end result of using uSOL is visible on-chain (SOL locked on Solana, uSOL tokens minted on Hyperliquid), much of the Hyperunit protocol’s logic resides off-chain in a coordinated network of servers called the Guardian Network. These off-chain components are critical to the system’s design and introduce certain trust assumptions (which we will evaluate in risk sections). Here we describe the key off-chain elements:

• Guardian Network:
The core of Hyperunit is a small set of independent guardians (initially three) running the Hyperunit Agent on secure servers. They monitor supported blockchains and collectively execute bridge actions. One guardian acts as leader to propose operations; the others independently verify. Communication is broadcast via a lightweight relay that only forwards encrypted messages and holds no keys. Even if the relay fails or is compromised, secrets are not exposed and funds remain protected, since all sensitive operations occur inside enclaves.

• Blockchain Nodes & Indexers:
Each guardian runs its own full nodes and indexers for every supported chain (e.g., Solana, Ethereum, Bitcoin). This lets every guardian independently detect deposits, track confirmations, and validate events. No single oracle or watcher is trusted. A false deposit would require multiple guardians to be deceived in the same way, materially reducing the risk of spoofed or incorrect credits.

• Agent Software (Protocol Engine):
Each guardian runs identical agent software that enforces the protocol:

Chain Service: watches blockchains, verifies finality, and constructs outgoing transactions.
State Machine / Flow Manager: enforces the ordered lifecycle of each operation (detect → confirm → agree → sign → execute), preventing shortcuts or unilateral actions.
Consensus Service: enforces the threshold quorum (currently 2-of-3) on both data and execution; if guardians disagree, the operation halts.
Wallet Manager (MPC): controls signing via threshold cryptography. Private keys are split into shards held by guardians and used only inside secure enclaves (e.g., Nitro). Signatures are produced jointly without ever reconstructing the full key.

• Off-Chain Compliance Layer:
Guardians apply sanctions screening, geo-blocking, and related policy checks before processing deposits or withdrawals. These controls are enforced off-chain using external lists and heuristics. This supports regulatory alignment but makes the system explicitly permissioned and not fully censorship-resistant.

• User Interface & APIs:
Hyperunit is accessed through a centralized web UI and guardian APIs, which handle deposit address generation, status updates, and error reporting. Advanced users can integrate directly with the APIs, but they remain a central dependency. For most users, the UI and Hyperliquid interface abstract away the off-chain complexity while aggregating confirmations, queues, and operation states.

Summary of Off-Chain Trust: All these off-chain components mean that uSOL’s issuance and redemption rely on an external consensus (the guardian network) rather than a trustless on-chain contract. The upside is flexibility and performance – deposits can be from non-EVM chains (Bitcoin, Solana), and the process can include sophisticated logic (threshold signatures, batching) that might be impossible or inefficient to do purely on-chain. The downside is that users must trust that the guardian network is honest and secure. The guardians collectively act as the decentralized custodian of the SOL collateral and the oracle informing Hyperliquid of deposits. Thus, the design is often described as “non-custodial” in that no single party holds funds, but it’s not trustless in the same way a smart contract on Ethereum would be – trust is distributed among the guardians. This will be discussed more in the risk analysis (Part II).

B. System Design

B.1 Monitoring and Observability

A critical aspect of any asset-backed token system is the ability for users and third parties to monitor the backing collateral and overall system health. Hyperunit provides certain avenues for transparency, though some components remain opaque by nature of being off-chain.

First, as mentioned, the reserve addresses on each source chain are publicly known, which is a cornerstone of observability. For uSOL, one can look up the Solana treasury address in a Solana block explorer and see the balance of SOL held. Likewise, Hyperunit publishes addresses for its Bitcoin and Ethereum reserves in its docs. This means that at any time, the community can audit that the amount of SOL held in custody is at least equal to the total uSOL supply. On Hyperliquid’s side, uSOL exists as a token on both the HyperCore (native exchange chain) and HyperEVM environments. The total supply of uSOL on HyperEVM is visible via the token contract (address 0x068f…A29) which anyone can query. On HyperCore (the non-EVM chain), uSOL is a native asset with a known token index (254) and a maximum supply parameter (500 million set as an upper bound). While HyperCore doesn’t have a public block explorer in the traditional sense, Hyperliquid does expose APIs and data on asset balances, so one can observe how much uSOL is in circulation. In short, the peg backing is verifiable: e.g., if there are X uSOL in existence, the Solana address should also hold ~X SOL (minus any slight differences due to fees or timing of sweeps).

Furthermore, Hyperunit and community contributors have built dashboards for real-time monitoring. One notable example is the HyperDash Unit dashboard, which compiles data on deposits, withdrawals, and total flows. According to an analysis by OAK, “Thanks to the HyperDash dashboard (unit.hyperdash.info), users can monitor all flows handled by Unit.”. This dashboard visualizes metrics like cumulative inflows of each asset, recent volumes, and even comparative stats (showing, for instance, that Hyperliquid became the #1 DEX for native BTC spot volume largely due to Unit bridging BTC onto the platform). For uSOL specifically, one could see the total SOL brought in and out via Unit over time. As of mid-2025 (a few months post-launch), the dashboard indicated substantial usage – for example, over 1,182 BTC and 9,619 ETH had been bridged into Hyperliquid in the first four months for trading, and Solana integration (uSOL) was also live, likely contributing similarly. Such statistics demonstrate both the utility and the scale of assets under management, which is directly relevant to risk (higher TVL makes it a bigger target for exploits, and necessitates stronger monitoring).

Figure 1: Simplified architecture of the Hyperunit system, showing the off-chain Guardian Network coordinating between external blockchains (like Solana) and the Hyperliquid platform. Guardians independently monitor deposits on L1 chains and reach consensus to mint or burn tokens on Hyperliquid.

User-facing observability:
Users get real-time status updates via the UI and APIs. After depositing SOL, they can track stages like “waiting for confirmations” or “minting in progress.” Hyperunit also exposes endpoints such as GET /withdrawal-queue, showing queued withdrawals and the last processed transaction per network. This gives users visibility into whether their withdrawal is batched and roughly where it sits, reducing uncertainty and support overhead.

Structural opacity:
Guardian operations are off-chain and not directly observable. Users don’t see individual guardian checks or signatures—only the final on-chain result (mint or release). Deterministic execution means a completed operation implies consensus was reached; failures usually surface only as delays or stuck transactions. In those cases, users must rely on support channels or status announcements, unlike fully on-chain bridges where failure modes are visible in smart contract events.

Auditability limits:
While Hyperunit has disclosed its high-level security design (MPC, enclaves, guardian model), the guardian software is not open source. This constrains independent verification and shifts trust to external audits and team credibility. The Zellic audit provides transparency on the on-chain contracts, but the off-chain guardian logic remains largely opaque. Guardians maintain internal logs and audit trails, but these are not publicly accessible in real time.

In summary, Hyperunit offers a moderate level of transparency: the collateral is on-chain and checkable, key metrics and flows are tracked by dashboards, and users have tools to check the status of their transactions. However, the core consensus process among guardians is opaque by necessity, which requires a level of trust. The current guardian set is known (Hyperliquid and partners), which at least means the actors are accountable to their reputations. But from an exterior point of view, the opaqueness is that one must trust that these known entities are following the protocol and not colluding or deviating, since one cannot “watch” the guardian consensus on-chain. This is a typical trade-off in such bridge designs – they prioritize performance and cross-chain flexibility at the cost of some transparency, unlike purely on-chain light-client bridges that put all logic on-chain (but those are limited to certain ecosystems and often slower). Users and observers should therefore monitor both the on-chain reserves and any available off-chain status indicators (like official dashboards or announcements) to remain confident in uSOL’s backing and operation.

B.2 System Architecture

The Hyperunit / uSOL architecture is security-first and structured in layers: on-chain assets (SOL, uSOL), an off-chain guardian network, and hardened security infrastructure (MPC, enclaves). Together, these balance decentralization with operational control.

• Distributed Guardians & Consensus:
The system currently runs with N=3 guardians (Unit, Hyperliquid, Infinite Field) using a 2-of-3 threshold. Any mint or release requires two independent approvals, preventing unilateral control and tolerating one faulty or malicious node. A threshold signature scheme (TSS) ensures no single guardian can forge transactions. Guardians run a deterministic state machine in lockstep; if one deviates, the others refuse to sign, halting the operation.

• Secure Key Management (MPC + Enclaves):
Reserve keys are split into shards held by guardians and never reconstructed. Signing is done jointly via MPC inside hardware enclaves (e.g., AWS Nitro), so even a compromised server cannot extract key material. No machine ever sees the full private key, significantly reducing key-theft risk versus standard multisigs.

• Strict Protocol Logic & Circuit Breakers:
Guardians independently verify deposits (confirmations, formatting, blacklists, minimums, etc.). Only if all checks pass does processing continue. The system includes rate limits and emergency halts: if invariants break or suspicious behavior appears, the bridge can automatically pause and require manual intervention.

• Secure Communication:
All guardian-to-guardian communication is end-to-end encrypted with pre-shared identities. A relay may forward messages but cannot read or forge them. Even if the relay fails, safety is preserved (though liveness may degrade).

• Redundancy & Roadmap:
With 2-of-3 consensus, the bridge continues operating if one guardian is offline. It cannot tolerate two failures. Future plans include expanding the guardian set (e.g., 3-of-5) to improve decentralization and fault tolerance.

• Deep Integration with Hyperliquid:
uSOL is native on HyperCore (for spot trading) and mirrored as an ERC-20 on HyperEVM. Guardians mainly handle external L1 ↔ HyperCore bridging; transfers between HyperCore and HyperEVM are handled by Hyperliquid’s own chain logic. This tight integration enables fast settlement and native use of uSOL as collateral.

• Compliance & Policy Layer:
Guardian software embeds sanctions and geofencing checks. Deposits from blocked entities are not processed. These controls are enforced collectively by guardians rather than a single admin key. This makes the system explicitly permissioned at the policy level, aligning it with regulatory requirements but reducing neutrality and transparency.

In summary, Hyperunit’s architecture is a carefully balanced combination of decentralization (a small consortium of nodes), advanced cryptography (MPC, TSS, enclaves), and practical engineering (state machines, batching). It is designed to maximize security (by eliminating single points of failure in key storage and validation) and to maintain high performance (by using off-chain coordination to avoid expensive on-chain verification of external chains). The system architecture succeeds in making uSOL feel like a native part of Hyperliquid – deposits and withdrawals are relatively fast (finalized in minutes), and within Hyperliquid, uSOL behaves just like SOL would (fast trading, instant settlement). The cost is that the architecture introduces a trusted intermediary layer (the guardians), which we will analyze for risks in Part II. Nonetheless, as a piece of infrastructure, Hyperunit is a novel design pushing toward decentralized custody of non-native assets. It operates with a philosophy that even though it’s not fully trustless, it’s transparent and robust enough that users can trust it more than a centralized custodian, and it can be progressively decentralized over time.

1 Like

uSOL Risk Assessment by BA Labs

Part II: Technical Risk

A. Smart Contract Risk

uSOL’s lifecycle involves interactions with on-chain smart contracts (particularly on Hyperliquid’s side). Any vulnerabilities in those contracts or in the bridging logic could pose a risk to the peg and user funds. To assess this, we consider known audits and potential smart contract issues.

Audit Findings: The Hyperliquid/Hyperunit bridge contracts underwent a security audit by Zellic in mid-2023, which scrutinized the smart contract component of the bridging system. According to the published Zellic report, no critical vulnerabilities were found, but there was 1 high-severity issue and 1 medium-severity issue, along with several informational findings. The Hyperliquid team addressed all identified issues and implemented fixes before launch. Below is a summary of the audit findings, their severities, and their resolution status:

Issue Identified (Zellic Audit) Severity Status (Amendment)
Withdrawal finalization did not work as intended – A logic bug in the bridge contract’s finalizeWithdrawals function could prevent proper finalization of certain withdrawals. This meant withdrawals might not complete under specific conditions. High (Critical impact on withdrawal process) Fixed. Logic corrected in code; added test cases to ensure withdrawals finalize correctly.
Disputed actions not blocked by validator rotation – The contract did not properly handle a scenario where an action (like a withdrawal) is disputed and a guardian validator set change occurs; the disputed action could still potentially proceed. This was a nuanced state issue in the consensus/bridge handoff. Medium (Labeled High severity in report but Medium impact) Fixed. Enhanced checks to ensure any ongoing disputed operation is halted if a guardian set rotation happens, closing the inconsistency.
Missing message validation enabling griefing – The contract didn’t verify that a withdrawal-finalization message actually corresponded to an existing withdrawal request. An attacker could send an arbitrary message that would be treated as “finalized” (often resulting in a 0 transfer or revert), potentially blocking the real withdrawal message later (denial-of-service via message collision). Low (Informational) Fixed. Added proper validation to ensure only genuine pending withdrawals can be finalized, eliminating the possibility of finalizing a non-existent request.
Signatures could be reused across contracts – The signature scheme (EIP-712 domain separator) did not include contract-specific identifiers, meaning the same guardian signatures might be valid on a different contract on the same chain if an identical protocol were deployed. In the current context it wasn’t exploitable (only one bridge contract exists), but it posed a maintenance risk if code were reused for another token without changes. Low (Informational) Fixed. Domain separator or signed message was amended to include contract address or token identifier to prevent any cross-contract replay.
Withdrawal and validator-update signatures lacked an “action” field – The signatures did not encode what action the guardians were approving (withdrawal vs. validator set update), so theoretically a valid signature for one type of action might be replayable as another if other conditions aligned. Essentially, context separation was not strict. Low (Informational) Fixed. The signing schema was revised to incorporate an explicit field distinguishing different action types, ensuring signatures cannot be misinterpreted for the wrong action (acknowledged and patched by the team in commit updates).
Unchecked return value of ERC-20 transferFrom in deposit – In the bridge’s deposit function (for Ethereum/Arbitrum USDC), the code did not check the boolean result of usdcToken.transferFrom. If a token didn’t revert on failure (some ERC-20 tokens return false on failure instead of reverting), this could lead to an event being emitted without actual transfer, causing accounting issues. (In current deployment with standard USDC, failure results in revert, but the code wasn’t future-proof.) Low (Informational) Fixed. Replaced with safe transfer functions or added checks so that the deposit function verifies the transfer succeeded before emitting events.

Citations: Zellic Audit Report, which notes one high-impact and one medium-impact issue among others, all of which were acknowledged and fixed by Hyperliquid’s team.

The resolution of all these issues was confirmed in a follow-up patch review audit (Zellic’s final report), which found that the Hyperliquid contributors had implemented the recommended fixes and added test cases to cover the scenarios. This indicates a responsive approach to security: potential flaws in the bridging contracts were proactively identified and eliminated prior to mainnet launch.

Residual Smart Contract Risks: Given the audit and fixes, what smart contract risks remain for uSOL?

  • One risk is the generic risk of any bug that escaped audits. While Zellic’s audit was thorough, no audit can guarantee absence of bugs. The bridge contracts (especially on HyperCore/HyperEVM side) are fairly specialized code (handling things like consensus of signatures, tracking nonces/messages, etc.). If there were an unknown bug, worst-case it could lead to mis-accounting of uSOL or a situation where withdrawals/deposits get stuck. The critical nature of the code means even “medium” bugs could disrupt service or, in a nightmare scenario, allow an attacker to withdraw funds without proper authorization. That said, the core logic is not extremely complex (mainly checking signatures and moving tokens), and multiple eyes (internal, Zellic) have reviewed it, reducing likelihood of a severe latent bug.
  • Bridge Contract Upgradability: It’s not explicitly stated in docs whether the bridge contracts (especially the HyperEVM ERC-20 contracts or any coordinating contract) are upgradeable or controlled by a multi-sig. If they are, that introduces a trust element – those who control the upgrade could potentially alter how uSOL works. Given Hyperliquid’s architecture, it’s possible the HyperCore chain logic can be upgraded via governance or the founding team (Hyperliquid is new and somewhat centrally managed at the moment). If so, users have to trust that no malicious or faulty upgrade will occur that impacts uSOL. For instance, a contract upgrade could theoretically pause withdrawals or add fees if done. This is more of a platform risk than a flaw, but relevant to mention in smart contract risk context.
  • Interaction with Hyperliquid’s core ledger: uSOL relies on HyperCore’s native token implementation (for trading, etc.). If there were a bug in Hyperliquid’s core handling of assets (not specifically in Unit’s code but in the chain’s asset management), that could affect uSOL. For example, a bug in Hyperliquid’s matching engine or balance accounting could conceivably corrupt the uSOL balances. However, Hyperliquid’s core has also been audited and is backed by the team’s expertise (they built a custom chain for performance), so we assume those mechanics are sound. Still, the complexity of a custom chain plus an EVM sidecar means the surface area is large. The Hyperliquid core chain’s consensus and validity are also crucial – if Hyperliquid L1 had a consensus failure or rollback, it could affect uSOL balances (though SOL reserves on L1 would remain intact, users might have discrepancies in what Hyperliquid thinks they are owed).
  • Dependence on External Protocols: uSOL smart contract risk is also tied to Solana’s SPL and system program reliability. Although SOL transfers are straightforward, if Solana had a bug in its token program (for SPL tokens in case of other assets) or some catastrophic failure, those could indirectly impact bridging. For SOL itself, the risk is more about Solana’s network uptime (which is an off-chain risk). On Hyperliquid EVM side, uSOL is an ERC-20 – one minor risk could be lack of standard features like permit, etc., but those don’t impact security, more UX.

In conclusion, the smart contracts associated with uSOL have been audited and currently have no known vulnerabilities. The team’s prompt fixing of identified issues and the relatively simple function of the contracts (holding and transferring tokens under multi-sig control) bode well for their security. The main assurances are that (a) guardian signatures are always required for mint/burn (so an attacker can’t mint uSOL arbitrarily unless they break the guardian security), and (b) the contracts correctly implement all necessary checks (which the audit helped ensure).

The biggest smart contract-related risk is if the guardian logic is somehow bypassed or if invalid signatures are accepted due to a bug – but the audit’s high-severity findings were exactly about those kinds of issues (finalization logic, signature domain separation), and those were fixed. As long as the contracts require the proper threshold signatures and map one-to-one with locked collateral, then even if Hyperliquid’s chain or contract is attacked, an attacker shouldn’t be able to mint uSOL without depositing SOL or withdraw SOL without burning uSOL. This is key: any break in that one-to-one mapping due to a contract bug would directly compromise the peg. Thankfully, no such break is known at this time.

B. Off-Chain Component Risk

Off-chain components – primarily the Guardian network – are the heart of Hyperunit’s security model, and they also represent its greatest sources of systemic risk. Unlike a fully on-chain system where one trusts code and decentralized consensus, here users must trust that the off-chain protocol is run honestly and robustly by the guardians. We analyze several facets of off-chain risk: guardian collusion or compromise, operational failures, network attacks, key management risks, and censorship/regulatory risks.

  1. Guardian Collusion or Malicious Behavior: Currently, there are three Guardians (Unit, Hyperliquid, and Infinite Field) operating the network. Security relies on at most one of them being malicious at any time. If two of the three collude or are taken over by a malicious actor, they gain full control of the system. With a 2-of-3 threshold, a colluding majority could sign any transaction they want – for example, they could mint arbitrary uSOL to an attacker’s account on Hyperliquid or withdraw all SOL from the treasury to a wallet of their choosing. Such an event would be catastrophic, as it would effectively steal the underlying collateral, leaving uSOL holders unbacked. This is the classic threat in any multisig-based bridge: the system is only as secure as the honesty of the majority of signers. In Hyperunit’s case, the three entities are presumably trusted collaborators (Hyperliquid’s own team and close partners). The likelihood of collusion is low in terms of intent – Hyperliquid has a strong incentive not to defraud users (it would destroy their business) and the third guardian (Infinite Field) is likely a reputable institution whose incentives align as well. However, collusion could also occur via external coercion or hacking (discussed below).

Because the guardian set is small, it also might be a target for bribery or coercion. Imagine a state-level actor or criminal organization attempting to compromise the bridge: bribing or extorting two key personnel could, in theory, break the system’s security. This is a centralization risk inherent in having only three guardians. Oak Research notes that “the infrastructure currently relies on a restricted quorum — composed of just three Guardians… While this design enables a secure and controlled launch, the next key milestones will be expanding the Guardian network and opening governance.”. Indeed, plans to add more guardians and decentralize governance would mitigate collusion risk by requiring a larger conspiracy to break security (e.g., 4 of 6 guardians colluding is harder than 2 of 3). Until then, users are trusting these specific entities.

It’s worth noting there is some recourse if only one guardian turns rogue: with 2-of-3, a single malicious guardian cannot by itself forge signatures. They could, however, refuse to sign anything (a denial-of-service) or try to mess with deposit detection. The protocol’s design can handle one offline or uncooperative guardian – it will simply rely on the other two for consensus. A rogue guardian could at best delay operations (if they’re the leader and start proposing invalid actions, the others would reject those and likely switch leadership or pause until the situation is handled). The honest majority could then potentially remove the bad actor (though this requires a rotation procedure, possibly involving an on-chain governance or an update to the guardian list – which itself likely needs a quorum of signatures to enact). The Zellic audit’s issue about disputed actions during validator rotation hints that there is a mechanism to update the guardian set. This suggests if one guardian is identified as malicious, the system can be reconfigured to exclude them (Hyperliquid would likely do this in an emergency). However, during the rotation window, things must be carefully managed to avoid the malicious guardian exploiting the handover – the fix for that issue was implemented.

  1. Key Compromise and Cybersecurity: Each guardian node runs in cloud or on-premise environments, and although keys are held in secure enclaves, there’s still risk of hacking or insider threats. If an attacker compromises the servers of two guardians, they could potentially execute fraudulent transactions. The secure enclave setup means even with server control, extracting the key share is non-trivial, but the attacker might instead try to hijack the signing process (e.g., input their own transaction to sign). Guardians likely require some authentication to trigger a signing (it may not sign arbitrary requests unless the protocol’s state machine deems it valid). But a sophisticated attacker on a guardian’s machine might attempt to manipulate the agent software or simulate the other guardian’s agreement. The protocol’s multi-party nature makes this hard – one would need to simultaneously subvert two independent systems and the coordination between them. The use of encryption in inter-guardian communication also means compromising one node doesn’t let you eavesdrop or inject easily into the consensus without the other’s keys.

A more direct key risk is if there’s a flaw in the MPC or enclave. For example, if AWS Nitro Enclaves had an exploit that allows dumping memory, an attacker might steal a guardian’s key share from memory. With two shares, they could reconstruct the full private key for the Solana reserve (since it’s a 2-of-3 scheme, any 2 shares can sign – effectively the key can be recombined). This is a critical risk: compromise of any 2 guardians’ secure environments = loss of funds. The security measures (enclaves, KMS, etc.) significantly raise the bar for attackers, but we have seen in the past that even Intel SGX and other enclave tech have had vulnerabilities. It requires state-of-the-art attacker capabilities, but given the value locked (possibly tens of millions of dollars in assets across BTC, ETH, SOL), it’s a prize that could attract advanced persistent threats. The guardians must keep their systems patched and ideally geographically and technologically diverse to reduce common-mode failures. It’s not public how the guardians are run (likely each by a separate team on separate infrastructure), which is good for decentralization.

  1. Operational Failures and Downtime: The reliance on a few off-chain servers means the bridge can experience downtime or delays if guardians face issues. For example, if one guardian’s indexer falls behind or a node crashes, the protocol might halt new operations until it recovers or the others decide to continue without it. The design allows for one to be down, but consider if an issue like a network partition occurred – e.g., Guardian A cannot communicate with B and C (maybe cloud outage or censorship in one region). The other two could still form quorum, but if the partition makes it unclear who’s leader or splits them 1-2, it might cause a stall. There’s also the scenario of software bugs or crashes in the agent code that weren’t caught. If all guardians hit the same bug (say a deposit with a weird edge-case causes an exception in the code), the system could deadlock or repeatedly crash at that transaction. The deterministic design means they’d all likely experience the bug similarly. In such a case, human intervention is needed to patch the software and possibly manually handle the stuck operation. During that time, deposits/withdrawals might be paused.

Another operational risk is the withdrawal queue backlog. The system intentionally batches withdrawals to avoid spamming underlying networks. If there is a sudden surge in withdrawals (e.g., panic event where many users try to redeem uSOL at once), the queue could grow and users might face longer wait times. This is not a security failure per se, but it is a risk to user experience and potentially peg stability if people become anxious about delays. In extreme conditions, a long queue could be seen as a bank run scenario. The guardians might have to decide whether to process larger batches faster (paying more in fees) or stick to limits. There’s a risk of withdrawal throttling– while likely meant for technical reasons, it does mean users can’t all exit instantly by design. This could be exploited by arbitrageurs or cause a situation where uSOL trades at a slight discount if people fear a slow redemption. So far, volumes and usage have been manageable, but this is something to monitor.

  1. Censorship and Opaqueness: As noted, Hyperunit includes compliance features that can block certain users. From a risk perspective, this introduces centralization and fairness concerns. The guardians could be compelled by regulators to freeze certain funds or deny service broadly (e.g., in an extreme scenario, if a government asserted that Hyperunit is facilitating something illicit, they could pressure the known entities to shut it down or block all users from a region). Unlike a permissionless protocol, the guardians have the power to comply. Indeed, they have geofencing and can refuse deposits/withdrawals from blacklisted addresses. The risk for users is that access is not guaranteed: your funds could get stuck if your address gets flagged or if you fall under a jurisdictional ban after you’ve deposited. For example, if tomorrow US authorities sanction Hyperliquid and the guardians refuse to serve US-based users, a user in the US holding uSOL might not be able to directly withdraw to Solana (until they prove they aren’t US, or use a third party). This is not a traditional “technical risk” like a hack, but a socio-technical risk of the design. It trades some decentralization for compliance. Some analysts have pointed out this could hurt Hyperliquid’s competitiveness versus fully open platforms.

From an “opaqueness” standpoint, the guardians’ decision-making and system status are not fully transparent (as discussed in Part I.B.1). This means if something goes wrong, users might be in the dark. There’s a trust in the operators to communicate issues promptly and truthfully. If, say, a partial key compromise happened but funds weren’t yet stolen, would they pause and announce it? Would users know to stop depositing? The system’s central coordination means the team can and likely would inform users via official channels, but it’s a point of trust. The audit logs the guardians keep are for internal/post-mortem use, not for public real-time verification.

  1. External Blockchain Risks: Off-chain risk also includes the risks of the underlying blockchains that Hyperunit connects. In uSOL’s case, Solana blockchain risk comes into play. If Solana were to experience a severe outage (which it has in the past) or a deep consensus failure, Hyperunit might be unable to process new deposits or withdrawals of SOL. For instance, a Solana outage could freeze the ability to redeem uSOL for a period, meaning uSOL holders are stuck on Hyperliquid. If Solana had a catastrophic failure or was 51% attacked such that finality is no longer reliable, the guardians might have to suspend operations to avoid accepting deposits that could be reversed. Similarly, if Solana’s value or functionality collapsed, uSOL would be affected (though that’s more of a market risk). These are not caused by Hyperunit, but the protocol has to handle them. The guardian design does allow halting in such cases (they’d likely pause if Solana is unstable).

Another external factor: since Hyperunit also deals with BTC, ETH, etc., a problem in any one chain (like Bitcoin extreme backlog) could indirectly affect overall operations (guardian resources, etc.). But for uSOL specifically, Solana’s health is the key external dependency.

  1. Recovery and Governance: A subtle risk is what happens if Hyperunit guardians fail or the protocol needs emergency intervention. Is there a way for users to recover funds if the guardians all disappeared? Currently, there is no trustless escape hatch – if Hyperunit guardians went AWOL, the SOL in the treasury would remain stuck since no one can sign withdrawals. Users holding uSOL would be unable to redeem (though Hyperliquid could potentially allow trading it if someone speculates on recovery). This is a low-probability scenario (all operators vanish), but not impossible (e.g., legal injunction could shut them down). It highlights that the system is custodial in aggregate – funds depend on guardians being there to sign. Decentralizing to more guardians and perhaps introducing a robust governance (maybe involving HYPE token holders or others) is planned, but currently it’s a closed set. There’s a risk that governance decisions (like adding a new guardian) are effectively under the control of the founding team and could be done in ways users might not easily influence or know of until after the fact.

  2. Scenario Analysis (Worst-Case): The absolute worst-case scenario for uSOL would be a compromise of two guardians leading to theft of the SOL reserve. If that happened, the attacker could drain the Solana treasury (e.g., send all SOL to their own address). Users on Hyperliquid would still see their uSOL balance, but it would no longer be backed by anything. Hyperliquid would presumably halt trading on uSOL if such an event were detected to prevent further harm, but it would be a major loss event. The bridge could become insolvent. This is analogous to a bridge hack (many have occurred on other platforms, often due to smart contract bugs or key compromises) – here the likely vector would be key compromise or insider betrayal. The Hyperunit team’s security measures (enclaves, etc.) are specifically aimed at making this extremely difficult. Also, having a small, known set of guardians means it’s easier to secure (enterprise-grade security on 3 nodes is easier than on 15 unknown nodes). But it’s also a central point of failure in that sense.

Another bad scenario is a guardian disagreement or fork that leads to inconsistent state. If, say, one guardian saw a deposit and minted uSOL, but another did not and thus doesn’t recognize it (this shouldn’t happen with proper consensus, but imagine a bug), it could cause a balance discrepancy. The deterministic consensus normally prevents this, but hypothetically if there was a split-brain situation (perhaps due to a network partition where two guardians sign something and the third signs a different thing), Hyperliquid might briefly see conflicting transactions. However, Hyperliquid’s chain likely would not allow double-minting due to nonce/message tracking – one would override the other. The design likely ensures finality by requiring the threshold signatures up front. So, this scenario is unlikely.

Mitigations and Outlook: Many of these off-chain risks are mitigated by the architecture choices:

  • Collusion is mitigated by having multiple independent guardians and plans to add more (increasing decentralization).
  • Key compromise is mitigated by MPC and secure enclaves, plus likely rigorous operational security (guardians presumably have intrusion detection, etc.).
  • Downtime is mitigated by redundancy and the ability to continue with 2-of-3 guardians. The system also has the circuit-breaker to pause on anomalies, which, while a form of downtime, at least prevents chaos.
  • Censorship is a deliberate trade-off; the mitigation for users is simply being aware of it and avoiding triggering those conditions. Over time, if governance opens up, there might be community input on those policies.
  • Transparency could improve if the software becomes open source or if proof-of-reserve is regularly attested (though users can self-attest by checking addresses).
  • Expanding guardians will be the big improvement: if guardians go to, say, 5-of-7 or even a larger decentralized set, the system becomes much more resilient to both collusion and individual failures. The team explicitly acknowledges this as a goal.

Finally, insurance and recourse: as an emerging protocol, Hyperunit doesn’t have a formal insurance fund for bridge failures (unlike some bigger exchanges or bridges that set up insurance or rely on tokens to socialize losses). Users currently assume the risk. It’s possible the Hyperliquid team, if a small incident happened, might cover it out of pocket to maintain trust (as some exchanges do). But in a severe hack, users might not be fully compensated. This is a risk one accepts when using any cross-chain bridge.

In conclusion, the off-chain risks of uSOL via Hyperunit boil down to trusting the integrity and security of the Guardian Network. At present, the network is small and semi-permissioned, which is both a weakness (centralization) and a strength (easier to secure and manage). The technical design is strong – employing industry best practices like MPC, threshold sigs, and independent validation – which significantly reduces the likelihood of accidental faults or simple exploits. However, no system with human-operated signers is immune to attack or betrayal, so users must weigh that risk. Hyperunit’s approach can be seen as an attempt to maximize security given those constraints: multiple lines of defense are in place (deterministic state machine, encryption, enclave isolation, audit logs). The track record so far (since launch in Feb 2025) has been incident-free, and large volumes have flowed through safely.

That said, uSOL holders should remain vigilant. Key things to monitor include: the Solana reserve balance (to ensure it matches uSOL supply), any news of guardian changes or pauses (which might indicate issues), and the progress of decentralization efforts. As Hyperunit evolves – hopefully adding more guardians and perhaps community oversight – the risks will gradually reduce on the decentralization axis, though operational complexity will increase. In summary, technical risks exist but are being actively managed. The design choices reflect a mature security ethos (learned from many past bridge hacks in the crypto space), and while trust in the operators is required, that trust is buttressed by transparency measures and the aligned incentives of those operators to keep the system solvent and secure. Users of uSOL should be aware that they are effectively in a 2-of-3 multisig trust model and size their exposure accordingly – a model that, in academic terms, shifts the security assumption from an honest majority of hashpower (like in a blockchain) to an honest majority of a small committee. This model has both strong security (if that committee is honest and competent) and clear pitfalls (if it is compromised), making ongoing risk assessment essential.

1 Like

AI produced TLDR

kHYPE Risk Assessment Summary

Protocol Overview

kHYPE is a liquid staking token (LST) issued by Kinetiq protocol for the Hyperliquid network. It represents staked HYPE tokens at a 1:1 rate initially, with the exchange rate increasing as staking rewards accrue. The protocol allows users to stake HYPE and earn yield while maintaining liquidity through the kHYPE token.

Key Components:

  • Underlying Collateral: Native Hyperliquid token (HYPE)
  • Yield Sources: Staking rewards (block rewards + transaction fees) from Hyperliquid validators
  • Minting Flow: Users deposit HYPE → receive kHYPE tokens
  • Redemption Flow: Users burn kHYPE → receive HYPE after unbonding period
  • Fee Structure: 0.10% unstaking fee (default)

System Architecture:

  • StakingManager: Handles user interactions and HYPE delegation
  • kHYPE Token: ERC-20 token representing staked HYPE
  • StakingAccountant: Tracks total staked HYPE vs. kHYPE supply
  • ValidatorManager: Manages validator selection and delegation
  • OracleManager: Bridges off-chain validator data to on-chain
  • PauserRegistry: Emergency pause functionality

Technical Risk Assessment

Smart Contract Risks

Multiple independent audits identified and addressed critical vulnerabilities:

  1. Accounting Precision Issues:

    • Decimal mismatch between Hyperliquid (8 decimals) and EVM (18 decimals)
    • Fixed by aligning kHYPE minting with actual staked HYPE amounts
  2. Upgradeability Risks:

    • Initialization problems in proxy contracts
    • Resolved by proper implementation of initialize() functions
  3. Withdrawal Logic Flaws:

    • Ineffective cooldown mechanism (removed)
    • Unfair slashing distribution (fixed)
    • Queue processing optimization implemented
  4. Access Control Vulnerabilities:

    • Unprotected setter functions in ValidatorSanityChecker
    • Missing pause modifiers on critical functions
    • All resolved with proper role-based access controls
  5. Oracle and Validator Selection Risks:

    • Potential for outdated balance usage in rebalancing
    • Missing fund withdrawal on validator deactivation
    • Mitigated through enhanced oracle integration and validation

Audit Summary:

Audit Firm High Risk Medium Risk Low Risk Status
Spearbit Guild 2 (fixed) 7 (5 fixed) 5 (3 fixed) Resolved
Pashov Audit Group 5 (fixed) 6 (fixed) Several Resolved
Code4rena Contest 3 (fixed) 5 (fixed) 19 Resolved

Remaining Risk Factors:

  1. Admin Key Risk: Multi-sig controls critical functions
  2. Oracle Dependency: Off-chain components feed validator performance data
  3. Cross-Chain Complexity: Bridging between EVM and Hyperliquid consensus
  4. Slashing Exposure: Though not currently live on HyperEVM, oracle implementation includes slashing logic

Mitigation Measures:

  • Multi-sig admin controls
  • Emergency pause functionality
  • Validator performance scoring system
  • Buffer strategy for liquidity management
  • Comprehensive monitoring and alerting systems

The protocol demonstrates thorough security preparation with multiple audits addressing critical vulnerabilities before mainnet launch. The modular architecture allows for isolated security of each component while maintaining system-wide integrity through proper accounting and access controls.

1 Like