Renzo

Tier 2

Security Analysis Report

TVL

$102.09M

Sections

7

report sections

Citations

1107

evidence references

Version

v6

11/5/2026

5.3

Moderate Risk

Safety Score · out of 10

📋

Executive Summary

5.3/10

Executive Summary: Renzo

Renzo’s main holder risk is needing cash before its restaked assets can be withdrawn. A market discount can become a loss for forced sellers or users borrowing against their tokens. Historical ezETH audits provide useful assurance, but current deployment coverage and safeguards differ by product.

Safety Rating

Overall Score: 5.3/10 Dimension scores: Code security 6.0 (provisional) · Economic security 4.2 (provisional) · Team transparency 6.4 (provisional) · Governance 3.7 (provisional) · Track record 6.1

Scores by dimension, points out of 10 (points)
Code security
6.0 points
Economic security
4.2 points
Team transparency
6.4 points
Governance
3.7 points
Track record
6.1 points

Assessment: Renzo suits risk-tolerant users who can wait through withdrawals and assess each token separately. The evidence supports specific safeguards, while leaving important gaps in current backing, execution controls and stress liquidity.

Scores by Dimension

The dimension findings reflect the strengths and limits of the evidence.

DimensionKey Finding
Code SecurityezETH has substantial historical review; equivalent coverage across products and current versions is not established.
Economic SecuritySlow native exits, weak sampled trading routes and immediate pzETH withdrawal-pause authority limit dependable access to funds.
Team TransparencyNamed leadership, a foundation and public technical material support accountability; complete leadership histories and community health remain unverified.
GovernancePrincipal Ethereum upgrades wait seven days, while other upgrades and sensitive settings can change immediately; vote execution remains unclear.
Track RecordOperation since 2023-12-18 includes an ezETH depeg; final losses, comprehensive reimbursement and recovery duration remain unknown.

What This Protocol Does

Renzo pools assets and issues receipt tokens that users can hold or use elsewhere in decentralized finance, or DeFi. Restaking uses staked assets to secure additional services, adding rewards and loss risks. Its wider catalog includes the Staking Suite, Reserve Vaults and Enterprise Suite.

The assessed TVL, meaning deposited asset value, covers ezETH and its cross-chain representations, pzETH, ezSOL, bzSOL and ezJTO. REZ supplies governance and incentives. Related ezREZ and ezEIGEN matter because they share upgrade control; their collateral remains separate.

Reserve Vaults, ezBTC, Opolis bonds and Enterprise remain other-product context where shared exposure is not established. Their balances are not attributed to this TVL.

The captured TVL was $102.09 million on 2026-11-03; the adapter counts some capital also used elsewhere in DeFi. It is neither a live balance nor unique system-wide capital.

The 2026-11-05 product captures discontinue ezSOL, bzSOL and ezJTO minting and retain withdrawal instructions. Blast was deployed earlier and has no current TVL; that deployment is not assessed.

Key Strengths

Renzo’s strongest safeguards apply to named products and control paths.

  • Independent review of ezETH core and bridges. Sigma Prime’s 2024-06 review and Nethermind’s 2025-04-14 core and 2025-05-01 bridge reviews support historical review depth. Their auditor-controlled publications are established. That assurance does not transfer to pzETH or Solana programs.
  • Delayed principal Ethereum upgrades. At Ethereum block 26,345,115 on 2026-11-05, principal proxy upgrades wait seven days. Shared ezREZ and ezEIGEN beacon upgrades also wait seven days. Separate immediate powers remain.
  • Native redemption offers an alternative to selling. Documentation gives up to 15 days for ezETH, 14–21 days for pzETH and up to four days for ezSOL. These descriptions do not guarantee exits during stress.
  • Named accountability and technical disclosure. Public addresses, audit links and continuing commits enable scrutiny. Renzo documents a $250,000 Immunefi bounty, but its current terms and asset scope remain unconfirmed.

Key Risks

The first three risks follow the holder-risk ranking in the cross-risk analysis.

  1. A discounted exit becomes a liquidation loss: ezETH, bridged ezETH and pzETH. Magnitude: forced sellers can lose value, and leveraged holders can lose their posted equity. On 2026-11-05, Ethereum routes to wrapped ETH (WETH) through KyberSwap quoted 15.85% impact for a $100,000 ezETH trade and 97.87% for pzETH. These are route-specific quotes, not executed losses; smaller trades and other venues remain unresolved. Likelihood: medium. An exit wave needs no privileged access, while queues slow purchases and redemptions that could close the discount. REZ and ezREZ also face confidence and token-price exposure; core receipts are not documented as REZ-backed.
  2. Administrative compromise or error outlasts the exit window: ezETH, pzETH, bridged ezETH and shared-control ezREZ/ezEIGEN. Magnitude: a harmful implementation can endanger the affected product’s principal; a pause can freeze exits without an established maximum suspension duration. Seven days’ upgrade notice may expire before a withdrawal completes. Likelihood: low, with limited confidence. The malicious case requires privileged compromise or collusion; signer independence and monitoring remain unknown.
  3. Backing, slashing or settlement failure: ezETH and bridged ezETH, pzETH, ezSOL, bzSOL and ezJTO. Slashing means assets can be taken as a penalty for failures by the services they secure. Magnitude: permanent backing loss or extended redemption interruption in the affected allocation. The evidence cannot size a protocol-wide loss. Likelihood: medium. Operators, external administrators or infrastructure can disrupt required layers; current allocation and isolation limits remain incomplete. Solana minting has stopped, but withdrawal exposure remains.
  4. Current audit-version assurance is incomplete. Most deployed source was not compared with audit revisions; two components have measured differences. That is an evidence limit, not a demonstrated live vulnerability. Several project-hosted audit copies have unverified auditor publication provenance; this lowers confidence rather than imposing a provenance-only penalty.

Documented terms and measured controls

Older descriptions do not establish current settings or universal upgrade protection.

The LlamaRisk assessment discussing 2024-04 conditions says the timelock is three days and protects all upgrades; on-chain on Ethereum at block 26,345,115 (2026-11-05), principal upgrades wait seven days. Separate immediate paths remain.

At that block, the main Safe wallet requires three of five signatures. It can upgrade RiskOracleMiddleware immediately. pzETH has a separate five-of-eight upgrade wallet with no delay, and authorized single keys can pause its withdrawals immediately. The exact fastest delay of the underlying non-Ethereum proxy paths remains unknown, despite immediate upgrades of their local administrator wrappers.

The undated ezETH product page says 100% of EigenLayer rewards pass through and also states a 10% restaking-reward fee, split equally between treasury and operators; on-chain on Ethereum at block 26,345,115 (2026-11-05), DepositQueue.feeBasisPoints is raw 1500. The deployed denominator and fee scope are unverified, so 1500 cannot be presented as a percentage.

The instant-withdrawal docs describe anomaly safeguards as future work; on-chain on Ethereum at block 26,345,115 (2026-11-05), RiskOracleMiddleware is deployed. Its deposit and withdrawal pause flags are false. This does not confirm Chaos Labs as its provider or remove queue delays.

Withdrawal documentation also mixes seven-day buffered and 10–15-day unbuffered estimates with a fourteen-day EigenLayer minimum. These figures remain unreconciled and do not establish a guaranteed end-to-end wait.

pzETH’s FAQ describes services as not yet functional, but an official 2025-12-19 blog listing announces pzETH underwriting borrowing on Cap. Current allocation and realized underwriting income remain unconfirmed.

Critical Findings

Audit-status counts describe historical issue groups, not a count of live vulnerabilities.

Code Security counts 26 critical, high and medium groups once after cross-audit deduplication, using their latest audit status. Code checks do not change the auditor-status partition.

SeverityTotal issue groupsFixed per the auditorNot marked fixed by the auditor
Critical770
High1091
Medium963

The published product split below describes reviewed code surfaces, rather than confirmed coverage of every deployment. Each severity cell gives total groups, fixed per the auditor, then not marked fixed. EzRVault code is shared by ezREZ and ezEIGEN.

Reviewed code surfaceIncluded audit entriesCritical groupsHigh groupsMedium groups
ezETH core and bridges, including contest link64 total; 4 fixed; 0 not fixed9 total; 8 fixed; 1 not fixed7 total; 5 fixed; 2 not fixed
EzRVaults and LiquidVaults, including targeted integrations53 total; 3 fixed; 0 not fixed1 total; 1 fixed; 0 not fixed2 total; 1 fixed; 1 not fixed
REZ-token reviews20 total; 0 fixed; 0 not fixed0 total; 0 fixed; 0 not fixed0 total; 0 fixed; 0 not fixed

The four groups not marked fixed are:

  • High — Malicious Validator Frontrunning Attack; Sigma Prime, 2024-06. Audit status: acknowledged. An operator could set its own withdrawal credentials before Renzo’s stake. The team relied on permissioned validators with commercial agreements. Check: the default-branch snapshot Renzo-Protocol/contracts-public@fc8c8e0 may still contain the described code, with low confidence and deployment unconfirmed. The passage does not resolve downstream checks or operational protections.
  • Medium — Incorrect stakedButNotVerifiedEth Accounting Can Overinflate TVL; Sigma Prime, 2024-06. Audit status: acknowledged. A residual stake–verify–stake case remained in retesting. Check: the default-branch snapshot was rewritten and now delegates accounting to a library. The check is inconclusive; current deployment applicability is not established.
  • Medium — Double Use Of Validator Public Key Can Re-introduce Security Issues; Sigma Prime, 2024-06. Audit status: acknowledged. Reusing a key across delegators can reintroduce accounting discrepancies; the team accepted the operational risk. Check: the same default-branch snapshot may still contain the described code, with low confidence and deployment unconfirmed. Caller restrictions and off-chain administration remain unresolved by the passage.
  • Medium — Inaccurate queue filling with rebalancing assets; Halborn EVM Managed Vaults, engagement 2025-09-26–2025-10-02. Audit status: acknowledged; risk accepted 2025-10-08. Interest-bearing or rebalancing assets can cause excess queue funding. The team stated it would avoid rebasing underlying assets and added a warning. Check: unavailable because the audited repository was unavailable; this does not confirm a live issue. The report names Halborn internally, but its auditor-controlled publication is unverified. This concerns managed-vault code, not an established ezETH or pzETH deployment defect.

The contest-link entry has additional published findings outside these counts: eight high and fourteen medium. They remain undeduplicated, without assigned current statuses, and unconfirmed under the captured page’s classification (source reliability: unverified publisher). Missing extracted detail in the REZ-token and managed-vault reports also limits completeness; zero recorded groups does not prove zero vulnerabilities.

Current coverage remains bounded. Coverage is not established for pzETH’s MultiVault or Solana programs. Most deployed source was not compared with audit revisions because commits were not public. The measured RestakeManager and EzEthToken differences concern the compared 2023-11 revision; equivalence to later audits is not established. Sourcify reports runtime matches at listed addresses; those matches do not establish audit coverage.

The 2024-04-23 ezETH depeg is a material economic incident: historical risk research supports downstream liquidations and bad debt. The social assessment retains a major classification because incident-day TVL and final user losses are unknown; this does not establish insolvency or a contract exploit.

A published 2024-04-24 account reports roughly $56 million of liquidations, but that amount remains unconfirmed (source reliability: unverified publisher). A first-party post-mortem and completed comprehensive reimbursement remain unestablished. The published account must not be mistaken for verified final losses or a complete recovery record.

Recommendation

Assessment: Renzo is suitable for risk-tolerant users seeking staking and restaking exposure who can wait through product-specific exits. Users needing dependable short-notice access should avoid relying on receipt-token sales or instant withdrawals, especially alongside borrowing. Historical multi-firm ezETH review and delayed principal upgrades are useful safeguards, but immediate administrative paths, shared dependencies and incomplete current-version coverage remain material. Before increasing exposure, prioritize verified current controls, reconciled fees and backing, and executable exit liquidity; project-hosted audits retain their publication-provenance limits.

Report Sections

The detailed sections provide the evidence and product-specific limits behind this assessment.

  • Protocol Overview
  • Code Security
  • Economic Security
  • Social & Team
  • Governance
  • Cross-Risk Analysis

113 citations · click markers to view evidence

🏗

Protocol Overview

Protocol Overview: Renzo

What It Does

Renzo lets users earn rewards from staking and restaking while holding tokens they can use elsewhere in DeFi. Restaking means using staked assets to help secure additional services, which adds rewards and extra loss risks. Its documented offerings span the Staking Suite, Reserve Vaults and Enterprise Suite. These products require different assessments because their assets, exit terms and dependencies differ.

DeFiLlama classifies Renzo as Liquid Restaking. Its users include asset holders seeking automated rewards, DeFi users deploying receipt tokens as collateral, and institutions seeking managed strategies. The Enterprise Suite offers gated, customized deployments for institutional capital management.

The documented product and token set is:

ProductWhat the user holds and receives exposure to
EigenLayer ETH restakingezETH, backed by native ETH or stETH restaking.
Symbiotic restakingpzETH, minted from ETH or supported ETH staking tokens.
Other EigenLayer assetsezBTC for wrapped BTC; ezREZ for REZ; ezEIGEN for EIGEN.
Solana/Jito restakingezSOL for JitoSOL exposure, bzSOL for Binance staked SOL, and ezJTO for JTO.
Reserve VaultsezUSDC1 for stablecoin lending; ezUSCC1 for leveraged Superstate USCC exposure; ezCompETH1 for leveraged ezETH exposure on Compound.
Opolis bond distributionopusIAUSD, a fixed-term AUSD reinsurance bond built by Opolis. Its documented maturity was January 31, 2026; settlement is not confirmed.

REZ is the governance token. The docs also list ezSKATE, but explicitly identify Skate as its owner and manager, outside Renzo’s core restaking assets. It is another product and its controls are not attributed to Renzo here.

How It Works

Renzo pools assets, manages their deployment and issues receipt tokens; the core ezETH flow illustrates that model.

  1. Deposit and receive a receipt. The user sends native ETH or stETH through RestakeManager and receives ezETH. The official contract directory identifies the manager as the user entry point.
  2. Stake the native ETH. DepositQueue pools native ETH until the documented 32 ETH threshold. It then sends ETH to the Beacon Chain deposit contract. Withdrawal credentials point to an EigenLayer EigenPod, linking validator funds to restaking.
  3. Hold ezETH while rewards compound. Staking and restaking rewards increase the collateral represented by each token. The documented exchange rate is total ETH-denominated backing divided by ezETH supply; this accounting rate differs from a market trading price.
  4. Refill withdrawal liquidity before reinvesting. RestakeManager receives rewards, deposits and validator exits. It routes funds to WithdrawQueue when buffers need filling, and to DepositQueue when they are full.
  5. Request withdrawal, then claim. WithdrawQueue locks ezETH on request and burns it when the user claims underlying collateral. Instant withdrawals instead use available buffer liquidity, charge a variable fee and revert below the minimum buffer threshold.

The other products change the deployment step:

  • pzETH: deposited collateral is delegated to node operators associated with Renzo within Symbiotic. The FAQ describes Mellow vault queues when Symbiotic caps are full. A December 19, 2025 official blog listing also announces pzETH underwriting borrowing on Cap; current allocation amounts are unavailable.
  • Solana tokens: Jito restaking vaults secure Node Consensus Networks, services whose nodes validate information. ezSOL accepts JitoSOL in its underlying vault; the interface uses Jupiter swaps and a custom cross-program deposit call to accept SOL. bzSOL and ezJTO use the corresponding BNSOL and JTO vaults.
  • Reserve Vaults: ezUSDC1 allocates USDC to lending markets. ezUSCC1 converts USDC into USCC, supplies it to Aave Horizon and borrows USDC to repeat the investment. ezCompETH1 supplies ezETH to Compound, borrows wrapped ETH (WETH) and converts it into more ezETH. This repeated borrowing is called looping.

Exit terms are product-specific, and quoted times are expectations rather than guarantees:

ProductDocumented redemption terms
ezETHEthereum mainnet redemption; up to 15 days. The withdrawal guide separately gives seven days with a funded buffer and 10–15 days when empty.
pzETHReceive wstETH after processing stated as 14–21 days.
ezBTC, ezREZ, ezEIGENUp to 14 days, followed by manual claim of the underlying asset.
ezSOL, bzSOL, ezJTOTwo epochs, up to four days; each page states a 0.2% withdrawal fee, split equally between Renzo and Jito.

Reserve withdrawals also require available liquidity. ezUSDC1 has a three-day cooldown; ezUSCC1 has a three-day minimum and a pending fund-redemption period of up to one business day. ezCompETH1 estimates roughly three days. Each vault uses the lower share conversion price at request or claim, exposing withdrawing users to downside during the wait.

Yield Sources

Renzo combines underlying investment income with service rewards and token incentives; the balance varies by product.

  • ezETH and pzETH: Ethereum staking income plus rewards for securing external services. Non-ETH EigenLayer rewards are sold through ezAuctions and compounded into ezETH. The documented auction price falls linearly over five days until a buyer purchases the reward basket.
  • Other restaking assets: ezBTC seeks service rewards on BTC exposure. ezEIGEN receives EIGEN incentives. ezREZ combines service rewards, partner-token allocations and weekly compounding of Renzo instant-withdrawal revenue.
  • Solana: ezSOL combines staking, restaking and transaction-ordering income, called maximal extractable value. bzSOL combines BNSOL staking and Jito restaking; ezJTO earns Jito service rewards.
  • Reserve Vaults: lending income for ezUSDC1; USCC fund returns less financing costs for ezUSCC1; leveraged staking/restaking income less WETH borrowing costs for ezCompETH1. USCC’s underlying fund uses crypto cash-and-carry trades, staked Ether and US Treasury securities.
  • Opolis bonds: advertised income combines stablecoin returns and insurance-underwriting profits. The documented 40% annualized yield, approximately 20% return over six months, is an offering claim for the matured series, not verified realized performance.

Assessment: staking, lending interest and service fees provide economic yield sources, but returns can fall as borrowing costs rise or reward programs end. Incentive dependence is material for some strategies. Renzo’s rewards program distributes REZ through selected integration partners; the ezEIGEN page describes a first-year EIGEN emissions program beginning August 15, 2024. That historical program does not establish a current emissions rate.

Fees need care. The undated ezETH product page says both that Renzo passes 100% of EigenLayer rewards and charges 10% on restaking rewards, split equally between treasury and operators. Those statements are not reconciled. The measured state on Ethereum at block 26,345,115 on November 5, 2026 records DepositQueue.feeBasisPoints = 1500. Its deployed denominator and reward-base application are not established here; the raw value must not become an unconditional percentage.

All three Reserve product pages state a 20% performance fee, measured in USDC terms for the stablecoin vaults and ETH terms for the Compound vault. These are documented terms, not measured deployed fee settings.

No verified historical yield range for the core tokens is established by the available data. The app’s documented annual percentage yield calculation annualizes token-price changes over a default trailing 30-day window; it is backward-looking.

Dependencies & Integrations

Renzo relies on separate security, pricing and liquidity systems; receipt tokens do not remove those dependencies.

DependencyWhat Renzo relies on
Ethereum and EigenLayerValidator balances, EigenPods, restaking strategies and service rewards underpin ezETH. StrategyManager and DelegationManager are measured references from RestakeManager.
Symbiotic and MellowpzETH delegation and vault capacity. The FAQ says services are not functional; the December 2025 Cap announcement describes live underwriting. Current utilization is unconfirmed.
Jito, JitoSOL/BNSOL and JupiterSolana restaking, underlying staking exposure and deposit conversion.
Chainlink and Renzo pricing contractsDocs identify Chainlink as the primary oracle, a source of external price data, for stETH deposits and layer 2 network prices. RestakeManager references RenzoOracle; the USCC vault separately lists a USCC oracle without explaining its feed or fallback.
Lending and fund counterpartiesAave Core, Morpho, Aave Horizon and Spark for ezUSDC1; Superstate and Aave Horizon for ezUSCC1; Compound for ezCompETH1. Maximum allocations are limits, not observed portfolio weights.

The user-facing bridge is documented as powered by Hyperlane. Deposits from layer 2 networks (L2s) additionally depend on those networks’ canonical bridges: the docs describe a seven-day proof wait and batch transfers every 12 hours or at 32 ETH. The repository deployment notes name Arbitrum’s gateway/ArbSys and Base’s Lido/standard bridge paths.

Some chain contract pages still describe Connext transfers. The sources therefore describe multiple bridge generations and paths; they do not establish one universal route for every chain.

TVL & Adoption

The latest captured DeFiLlama snapshot records $102.09 million of total value locked, collected November 3, 2026 at 12:30 UTC. This is a captured valuation, not a live balance or a count of users.

ChainCaptured TVLShare of total
Ethereum$73,994,94272.482%
Linea$16,247,69315.916%
Arbitrum$3,884,4513.805%
Solana$3,310,5593.243%
Zircuit$1,987,0171.946%
Mode$1,278,5591.252%
Base$648,4790.635%
Optimism$453,9530.445%
Fraxtal$156,5040.153%
Binance$109,2610.107%
Monad$14,1200.014%
Sei$1,6400.002%

Ethereum accounts for about 72.5% and Linea 15.9% of the snapshot. Blast was deployed earlier but has zero captured TVL; that inactive deployment is not assessed.

The historical series shows $3.93 billion on June 1, 2024, $125.50 million on October 4, 2026, and $101.23 million on November 3, 2026. The daily history and headline snapshot differ on November 3; they are separate observations and are not forced to match. These dollar changes combine asset-price changes and changes in deposited quantities.

Notable documented integrations include the lending markets above, Cap/Symbiotic underwriting and Jito restaking. Product-level TVL, active-user counts and live integration balances are unavailable, so the protocol total does not establish adoption of each listed product.

Renzo TVL by chain in USD millions, DeFiLlama, 2026-11-03 (USD million)
Ethereum
73.995 USD million
Linea
16.248 USD million
Arbitrum
3.884 USD million
Solana
3.311 USD million
Zircuit
1.987 USD million

Technical Architecture

The ezETH core has a measured seven-day upgrade delay, while pzETH has a separate upgrade path with no imposed delay.

The control facts below were measured on Ethereum at block 26,345,115, timestamp November 5, 2026, 00:43:23 UTC. They describe that deployment and block.

Contract or componentRole and control
RestakeManager — 0x74a09653a083691711cf8215a6ab074bb4e99ef5User entry, backing calculation and allocation. The address context identifies it and reports a Sourcify runtime match; that is bytecode evidence, not audit coverage.
DepositQueue, WithdrawQueue and RenzoOraclePool staking deposits, service redemptions and configure price feeds. Together with the manager, their measured proxy-upgrade path runs through admin 0x91625601e2bbbeb7171c40c79fadbcfbff6a1982 and timelock 0x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7, with 604,800 seconds, or seven days.
OperatorDelegator accountingcalculateTVLs() reports backing by token and operator delegator. The measured manager reports two operator delegators; this is not a count of independent organizations or validators.
pzETH MultiVault — 0x8c9532a60e0e7c6bbd2b2c1303f63ace1c3e9811Its separate proxy admin is controlled by a Safe wallet 0x81698f87c6482bf1ce9bfcfc0f103c4a0adf0af0, requiring approvals from five of eight owners, with no imposed upgrade delay.
RoleManager — 0x4994efc62101a9e3f885d872514c2dc7b3235849Records administrator, oracle and pause roles. Its proxy upgrades also have the measured seven-day path.
ezREZ and ezEIGEN EzRVault instancesShare beacon 0xb9ebb6221c760c51dc43f1a4f2daf69f964ff86a, whose measured upgrade path uses the same seven-day timelock. The shared component creates common upgrade exposure.

The Reserve product pages identify separate entry contracts: 0x877bbA4238EF1DCA9BF574561da633049de3A334 for ezUSDC1, 0xc242DEEF9E4dBdF394fE425630F9aa5762dB2faC for ezUSCC1, and 0xd1a85beAED6b5D2e61D3C50C4bD56e4E46133686 for ezCompETH1. Their presence in documentation does not confirm their current administrator or strategy allocations.

The timelock’s recorded proposers include the protocol Safe 0xd1e6626310fd54eceb5b9a51da2ec329d6d4b68a, which is also its executor. That Safe requires three of five owners and has no measured enabled modules. This adds a delay to the upgrade paths above, rather than to every administrator action.

For ezREZ and ezEIGEN, this Safe can appoint a pauser and unpause without an imposed delay. The measured pauser address is zero for both; an appointed-pauser emergency path is therefore not currently established.

For pzETH, the measured pauseWithdrawals() holders are the single-key addresses 0x19d74871a530c97065d95278223e8b7a7cd5ba27 and 0x75936ce8899beadda3f8033838d5ca147807e6cf, with no imposed delay. A separate five-of-eight Safe, 0x9437b2a8cf3b69d782a61f9814baabc172f72003, can unpause withdrawals. A two-of-three Safe, 0x6e5cad73d00bc8340f38afb61fc5e34f7193f599, can change the deposit limit and rebalance without an imposed delay.

The pzETH read records limit = 0, with deposit and withdrawal pause flags false. These readings do not by themselves establish that new deposits can be accepted.

The core manager was not paused. Its RiskOracleMiddleware was deployed, with deposit, withdrawal-request, withdrawal-claim and instant-withdrawal pause flags all false. The instant-withdrawal docs describe anomaly safeguards as a future plan; the measured middleware establishes a deployed component, but does not establish that it is supplied by Chaos Labs.

Changeable settings include:

  • Core allocation and deposits: operator-delegator allocation weights, supported collateral and per-token deposit limits through RestakeManager administrator functions.
  • Pricing and fees: oracle addresses through RenzoOracle.setOracleAddress() and queue fee configuration through DepositQueue.setFeeConfig(). The exact callable holders and fastest delays remain unresolved for these settings and the allocation/deposit functions above.
  • Exits: WithdrawQueue buffer capacity and cooldown, plus instant-withdrawal minimum/maximum fees and minimum buffer. These are documented controls; exact current values are not established.

For ezETH, recorded pause-role holders include the Safe and timelock, but the permission analysis does not resolve the exact setPaused(bool) call path or its delay. Reserve Vault and Solana upgrade/pause controls are not established by these Ethereum measurements. Some pzETH strategy-setting holders also remain unresolved. For who can upgrade or pause core contracts, see Governance.

How ezETH deposits reach staking and withdrawal buffersOpen full view →
ACTORUserCONTRACTRestakeManagerCONTRACTDepositQueueEXTERNALBeacon Chain depositcontractCONTRACTWithdrawQueuedeposits through1stakes poolednative ETH in2refills withdrawalbuffer in4routes surplusfunds to4
  • (actor)
  • (contract)
  • (contract)
  • (external)
  • (contract)
  • step 1flow
  • step 2flow
  • step 4flow
  • step 4flow

Key Risks (Protocol-Level)

The main user tradeoff is liquid receipt tokens against delayed access to backing; leverage and shared dependencies can turn that gap into losses.

  • Slashing and strategy losses: native ETH sits in validators with EigenPod withdrawal credentials, while other products depend on their own restaking services or investment strategies. Punishment for validator or service failures can reduce backing. This risk is specific to the assets and services a product secures. For Opolis bonds, the docs explicitly allow principal to be slashed after yields are exhausted by insurance claims.
  • Liquidity and market-price risk: a redeemable token can still trade below backing when holders need immediate exits. Instant withdrawals depend on buffer availability, so they cannot guarantee liquidity in a rush. A published April 24, 2024 account reports an ezETH depeg and roughly $56 million of liquidations across DeFi. The account is available, but these event claims remain unconfirmed under its supplied publisher classification (source reliability: unverified publisher). It is not evidence of a verified protocol exploit.
  • Leveraged Reserve positions: ezCompETH1 assumes ezETH remains sufficiently close in price to WETH. ezUSCC1 relies on USCC value, borrowing costs and timely fund redemption. A widening discount or rising financing cost can force deleveraging or liquidation. Automated risk management reduces exposure only if pricing, execution and liquidity remain available.
  • Oracle, bridge and accounting dependencies: bad prices can misvalue deposits or leveraged collateral; interrupted bridges can delay assets or disconnect representations from backing. Integrators also depend on correct error handling: Renzo warns calculateTVLs() can revert and integrations must not continue with invalid data.
  • Administrator and contract complexity: funds span queues, restaking contracts, validators and external strategies. Upgrade delays protect the measured paths, but do not establish delayed parameter changes or equivalent controls for every product. Shared ezREZ/ezEIGEN beacon upgrades can affect both products. pzETH’s separate five-of-eight Safe can upgrade without a timelock. Managed-vault disclosures also identify vulnerabilities in Renzo and integrated protocols as risks.

Assessment: redemption delay is a routine planning constraint. Leveraged positions face greater liquidation exposure than simply holding the corresponding receipt token. Contract, slashing or bridge failures could cause severe losses, but the available data does not support a numerical failure probability.

Data Completeness

The evidence supports the product mechanics and captured scale more strongly than current product-by-product utilization, yields and permissions.

Available materials include official product documentation, contract-address context, audit reports, DeFiLlama snapshots/history and measured controls. No standalone articles directory or Blockworks quarterly report was supplied; the published depeg account resides among the documentation files.

The main limitations are:

  • Product-level TVL, active users, realized yield histories and current incentive rates are not established. The total TVL snapshot does not disclose which listed products contribute to it.
  • Documentation contains different bridge generations, conflicting ezETH reward-fee statements and differing withdrawal timing descriptions. The raw queue fee setting is measured, but its deployed percentage interpretation remains unverified.
  • Ethereum controls for ezETH, pzETH and ezREZ/ezEIGEN are measured at the stated block. Some exact configuration paths, Reserve Vault controls and Solana authorities remain incomplete; controls from one product are not transferred to another.
  • The USCC oracle’s upstream data and fallback are unspecified. The sources do not establish current Cap allocations, Opolis bond settlement or a universal present-day bridge route.
  • This overview does not establish equivalence between audited versions and current deployments. Audit-era descriptions and Sourcify bytecode matches do not establish current audit coverage for every product. The detailed assessment belongs in Code Security.

Insufficient data for this assessment: current realized returns for each product and a complete, verified set of operational permissions across all deployments.

206 citations · click markers to view evidence

🔒

Code Security

6/10

Code Security Analysis: Renzo

Renzo has substantial external review of its ezETH contracts, but the evidence does not establish equivalent coverage across every product. Most deployed source could not be compared with audit revisions, while two components have measured differences. Four issue groups are not marked fixed in the supplied audit ledger; their checks do not confirm a live deployment vulnerability. The details below separate audit conclusions, repository evidence and deployed-source comparisons.

Data Completeness

The evidence supports a detailed historical review, with incomplete links between audited revisions and current deployments.

The ledger covers findings from 11 audits. Findings in the ledger means 26 critical/high/medium issue groups, counted once after cross-audit deduplication at their latest audit status. Ungrouped not-fixed code-check rows would count as singleton groups; none adds a group here. These are audit issue groups, not individual audit entries or confirmed live vulnerabilities.

Read in full: the findings ledger, all 13 audit digests, the claims collection, audit metadata, the deployment coverage map and contract context. Selected audit sections were read to verify scope, dates, findings and methodology. This was not a general read-through of the audit collection. The repository reading was limited to the supplied finding-check passages.

The audit collection contains 13 included entries and one duplicate. One entry is only a Code4rena link. Its captured report page reports eight high and fourteen medium findings outside the supplied ledger; these have not been deduplicated or assigned current statuses here. This additional finding set is unconfirmed under the captured page’s publisher classification (source reliability: unverified publisher).

Other completeness limits:

  • The REZ-token and EVM Managed Vaults reports are flagged as lacking fully extracted finding text. Their pointed sections were checked for severe findings; the supplied ledger remains the status authority. HashEx has no extracted findings and its conclusion says no severity issues were found.
  • The Halborn November 2023 and Nethermind April 2025 digests were built from only the first 80,000 characters of their source material.
  • The address inventory is partial. The coverage map lacks attributed current addresses for DepositQueue, WithdrawQueue and OperatorDelegator. Later supplied on-chain measurements identify DepositQueue, as discussed below, without changing the supplied code-check evidence.
  • Code criticality remains unassessed for 6,680 uncovered lines, and four components were not measured. No files have unavailable line counts.
  • The deployment coverage map declares no additional collection gaps. Missing comparisons do not establish that published audits are missing.
  • Current Immunefi payout terms and the itemized asset scope remain unconfirmed because the primary program page could not be retrieved.

Audit Coverage

Review is broadest for ezETH and its bridges; vault, governance-token and other staking-product coverage must be assessed separately.

The documented surface includes staking and restaking tokens on Ethereum and Solana, Reserve Vaults, and an Enterprise interface. Named Ethereum assets include ezETH, pzETH, ezREZ and ezEIGEN, alongside REZ. Solana documentation names ezSOL, bzSOL and ezJTO, alongside REZ. Reserve Vault documentation lists USCC Leverage, USDC Yield and ezETH Compound Loop vaults. The reviewed Ethereum contracts do not establish review of the Solana programs or every Reserve strategy.

Blast had earlier deployments and no current TVL in the supplied snapshot; those historical deployments are not assessed as active exposure. Their source-verification gaps remain visible below.

The audit register below uses the supplied audit dates. Engagement dates and later report updates are distinguished where they differ. Tier 2 is the supplied auditor classification; unspecified tiers remain unspecified.

Auditor / tierAudit dateBounded subject and revision
HALBORN / Tier 22023-11ezETH core; 04bb57a (RewardHandler also 67baeb5).
Code4rena / Community contest2024-04Contest link only in the audit collection; captured page identifies the C4 Renzo repository.
HALBORN / Tier 22024-04-24REZ token; no revision established.
HALBORN / Tier 22024-05-31ezETH withdrawals and related core / bridge files; e70e1be.
Sigma Prime / Tier 22024-06ezETH restaking; bb1dc9e, retest 7dabe54.
HashEx / Tier unspecified2024-08-20REZ token only; preliminary report; product-code scope expressly excluded.
HALBORN / Tier 22024-09-19 (engagement September 12–16)EzRVaults; 58ae612, fix review PR 3.
Nethermind / Tier 22025-04-14ezETH core / EigenLayer slashing; eaceb5f → f4d9610.
Nethermind / Tier 22025-05-01ezETH L2 deposit and bridge contracts; c2bdea4 → cfe7249.
HALBORN / Tier 22025-06-05 (engagement June 2–11)Contracts PR 170/171 and EzRVaults PR 15.
Halborn (named inside report; register attribution absent) / Tier unspecified2025-09-26LiquidVaults / LEZyVault, strategies and queue; a0becb6.
HALBORN / Tier 22025-11-19 (engagement October 28)USCC strategy and withdrawal helper; 100491d; scope-to-deployment unresolved.
HALBORN / Tier 22025-12-09 (engagement November 26–27)LiquidVaults Aave leverage integration; 251721a, remediation 2e237ed.

The REZ review is not evidence of restaking-core review. HashEx explicitly excludes applications, operations and product code; its scope assessment disagrees with treating it as broader protocol coverage. The USCC strategy report’s deployment scope remains ambiguous.

Halborn’s managed-vault report names Halborn internally, although the audit register leaves its auditor field unset. Auditor-controlled copies are established for the original Halborn core review, HashEx, both Nethermind reviews and Sigma Prime. Other included reports are project-hosted or catalog-discovered copies whose auditor-controlled publication is unverified. That limits confidence in claims relying only on those copies; it does not itself reduce coverage credit.

Code4rena’s link is labelled April 2024 in the audit collection and June 2024 in protocol documentation. The captured report dates the review April 30–May 8, 2024 and the report June 7, 2024. Its finding links identify 519e518, but that is not an explicit audited-revision statement. The captured page is classified as an unverified publisher, so these supplemental details remain unconfirmed (source reliability: unverified publisher).

The counts below use only the supplied ledger and partition audit status. A code check does not move a group between these columns.

SeverityTotal issue groupsFixed per the auditorNot marked fixed by the auditor
Critical770
High1091
Medium963
Reviewed product / code surfaceIncluded audit entriesCritical: total / fixed / not fixedHigh: total / fixed / not fixedMedium: total / fixed / not fixed
ezETH core and bridges, including contest-link entry64 / 4 / 09 / 8 / 17 / 5 / 2
EzRVaults and LiquidVaults, including targeted integrations53 / 3 / 01 / 1 / 02 / 1 / 1
REZ-token reviews20 / 0 / 00 / 0 / 00 / 0 / 0

These categories describe the reviewed code, rather than assigning every deployment to a product. EzRVault code is shared by the documented ezREZ and ezEIGEN proxies. Coverage is not established for pzETH’s MultiVault, Solana programs, ezBTC, Opolis bonds, or every Reserve strategy. No included report supplies a formal-verification property set.

Deployed code vs audits

Measured line shares are 0.0% identical, 0.1% cosmetic, 14.4% changed, 83.1% unknown and 2.4% outside the compared audit scope. These cover physical lines of protocol-owned verified deployed files, counting each file once.

Comparison bucketFilesPhysical lines
IdenticalNot separately listedNot separately listed
Cosmetic110
Changed131,638
Unknown / not compared639,482
Outside the compared audit scope3276

The 0.0% identical share does not mean every deployed file differs. Most lines were not compared because audit commits were not public. Uncovered lines comprise 3,674 critical, 245 high, 797 peripheral and 6,680 with unassessed criticality. Reviewed roles include funds, access control, upgradeability, oracle inputs and utility/view code. Four components were not measured; zero files lack line counts.

The supplied denominators are: 43 deployments known; 39 with verified source; 6 compared with an audited revision or every searched commit; 5 bound to a public commit; 0 identical to an audited revision; 2 with unaudited changes; 0 still being searched; and 13 with an unresolved search. No additional collection gaps are declared. These denominators describe different checks and must not be added together.

The table preserves each listed chain and address. Chain identifiers use eip155:<chain ID>. Implementation and proxy addresses are both retained. A public commit identifies source lineage, not an audit.

ComponentDeploymentStatusAudit and revisionDeployed commitCriticalityNote
RestakeManagereip155:1:0xd5b3be349ed0b7c82dbd9271ce3739a381fc7aa0 (implementation of proxy eip155:1:0x74a09653a083691711cf8215a6ab074bb4e99ef5)Differs from the audited version in 15 filesHALBORN 2023-11; code-423n4/2024-04-renzo@04bb57a5d32fa2da93e2c2068522c78bcf776913Search incomplete: some repository trees or repositories could not be readAt least critical from assessed code; remainder unassessedMaterial changes; no claim about their timing.
EzEthTokeneip155:1:0x9bd289b14dd6e0782af82eeb3fcfeed4354cda2c (implementation of proxy eip155:1:0xbf5495efe5db9ce00f80364c8b423567e58d2110)Differs from the audited version in 5 filesHALBORN 2023-11; code-423n4/2024-04-renzo@04bb57a5d32fa2da93e2c2068522c78bcf776913Renzo-Protocol/contracts-public@edceda2 (master)At least high from assessed code; remainder unassessedMaterial changes; no claim about their timing.
MultiVaulteip155:1:0x0c5bc4c8406fe03214d18bbf2962ae2fa378c6f7 (implementation of proxy eip155:1:0x8c9532a60e0e7c6bbd2b2c1303f63ace1c3e9811)Version not comparedNot establishedSearch incomplete: some repository trees or repositories could not be readAt least critical from assessed code; remainder unassessedthe relevant repository was not read.
StrategyManagereip155:1:0xa7536c85898dc424c6bfe8592f983661f830656a (implementation of proxy eip155:1:0x858646372cc42e1a627fce94aa7a7033e7cf075a)Version not comparedNot establishedSearch incomplete: some repository trees or repositories could not be readAt least critical from assessed code; remainder unassessedthe relevant repository was not read.
EzRVaulteip155:1:0x3691daed64a285719128e85b8d87ae6b65259062 (implementation of proxy eip155:1:0xd4fcde9bb1d746dd7e5463b01dd819ee06af25db)Version not comparedHALBORN 2024-09-19; Renzo-Protocol/EzRVaults@58ae612Search incomplete: some repository trees or repositories could not be readAt least critical from assessed code; remainder unassessedthe audit repository was unavailable.
xRenzoBridgeeip155:1:0x34b861c07cb154a17611b589f77e18892b05725b (implementation of proxy eip155:1:0xc1036d6bba2fe24c65823110b348ee80d3386acd)Version not comparedSigma Prime 2024-06; repository unresolved@revision unresolvedRenzo-Protocol/contracts-public@57f6b3e (master)At least critical from assessed code; remainder unassessedthe audited revision could not be identified unambiguously.
xRenzoDepositeip155:56:0xfde1366f26cc3621d944afbc96177a90f2849d26 (implementation of proxy eip155:56:0xf25484650484de3d554fb0b7125e7696efa4ab99)Version not comparedSigma Prime 2024-06; repository unresolved@revision unresolvedSearch incomplete: some repository trees or repositories could not be readAt least critical from assessed code; remainder unassessedthe audited revision could not be identified unambiguously.
L2RateProvidereip155:42161:0x5eafb1d4b5bdfafe81715eebcc7713e418c80e78 (implementation of proxy eip155:42161:0xf25484650484de3d554fb0b7125e7696efa4ab99)<br>eip155:8453:0x87eae42e475b28b9b7ee8027a0f2b7623fda564c (implementation of proxy eip155:8453:0xf25484650484de3d554fb0b7125e7696efa4ab99)Version not comparedSigma Prime 2024-06; repository unresolved@revision unresolvedSearch incomplete: some repository trees or repositories could not be readhigh — access control, upgradeability, oracle and viewsthe audited revision could not be identified unambiguously.
Sei Mainneteip155:1329:0x6dcfbf4729890043dfd34a93a2694e5303ba2703Version not comparedNot establishedUnknown — unassessedverified source was unavailable.
Modeeip155:1:0x4d7572040b84b41a6aa2efe4a93efff182388f88Version not comparedNot establishedUnknown — unassessedverified source was unavailable.
0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7eip155:81457:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7Version not comparedNot establishedUnknown — unassessedverified source was unavailable.
Blasteip155:81457:0xe65843850106421eab49dd43d19fa2d75fc19a75Version not comparedNot establishedUnknown — unassessedverified source was unavailable.
EzAuctioneip155:1:0x5ab40d79d1c85670a10803853a307c862c1ba29f (implementation of proxy eip155:1:0x31e5b1750b872542af1848d62464e2936a387453)Version not comparedNot establishedSearch incomplete: some repository trees or repositories could not be readUnknown — unassessedthe relevant repository was not read.
XERC20eip155:1:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7 (implementation of proxy eip155:1:0x2416092f143378750bb29b79ed961ab195cceea5)<br>eip155:42161:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7 (implementation of proxy eip155:42161:0x2416092f143378750bb29b79ed961ab195cceea5)<br>eip155:56:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7 (implementation of proxy eip155:56:0x2416092f143378750bb29b79ed961ab195cceea5)Version not comparedHALBORN 2024-05-31; Renzo-Protocol/Contracts@e70e1beSearch incomplete: some repository trees or repositories could not be readUnknown — unassessedthe audit repository was unavailable.
XERC20eip155:59144:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7 (implementation of proxy eip155:59144:0x2416092f143378750bb29b79ed961ab195cceea5)<br>eip155:8453:0xb370a0bd64daf1c6b04eaf01c1dedf7c6bce30f7 (implementation of proxy eip155:8453:0xf757c9804cf2ee8d8ed64e0a8936293fe43a7252)Version not comparedHALBORN 2024-05-31; Renzo-Protocol/Contracts@e70e1beRenzo-Protocol/contracts-public@edceda2 (master)Unknown — unassessedthe audit repository was unavailable.
XERC20Lockboxeip155:1:0xeeada1f6061e30bb1501319bbffb44fdf7037ca8 (implementation of proxy eip155:1:0xc8140da31e6bca19b287cc35531c2212763c2059)Version not comparedHALBORN 2024-05-31; Renzo-Protocol/Contracts@e70e1beRenzo-Protocol/contracts-public@edceda2 (master)Unknown — unassessedthe audit repository was unavailable.
XERC20Factoryeip155:1:0x86e532b66872dc801d737276fc1d2ff425deb9cf (implementation of proxy eip155:1:0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3)Version not comparedHALBORN 2024-05-31; Renzo-Protocol/Contracts@e70e1beSearch incomplete: some repository trees or repositories could not be readUnknown — unassessedthe audit repository was unavailable.
L2RateProvidereip155:59144:0x6ba40a27b2231663a2f1b56de0e4de6bfdb571b2 (implementation of proxy eip155:59144:0x4d7572040b84b41a6aa2efe4a93efff182388f88)Version not comparedSigma Prime 2024-06; repository unresolved@revision unresolvedSearch incomplete: some repository trees or repositories could not be readUnknown — unassessedthe audited revision could not be identified unambiguously.
DepositQueueNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
WithdrawQueueNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
OperatorDelegatorNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
RoleManagerNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
LEZyVaultNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
LEZyVaultFactoryNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
MorphoDelegateStrategyNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
EulerSwapEzEthDelegateStrategyNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
EzRVaultsFactoryNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
RenzoOracleL2No current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
TimelockControllerNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.
xRenzoBridgeReceiverNo current address attributedNo attributed current addressNot establishedUnknown — unassessedthe audit names it but no current address is attributed.

The table shows 30 rows. The remaining 43 inventory rows also lack attributed current addresses and have unknown criticality: BalancerRateProvider, BalancerRateProviderStorage, BytesLib, CCIPReceiver, Connext, ConnextReceiver, DepositQueueStorage, EIP1271SignatureUtils, EzRVaultsFactoryStorage, HyperlaneReceiver, HyperlaneReceiverStorage, ILEZyVault, IMethStaking, IStakedTokenV2, IValueTransferBridge, IWithdrawQueue, IwstETH, L1, L2, LEZyVaultFactoryStorage, LEZyVaultStorage, LidoArbValueTransfer, LiquidVaults, LockboxAdapterBlast, METHShim, METHShimStorage, OperatorDelegatorStorage, OptimismMintableXERC20, OptimismMintableXERC20Factory, Renzo, RenzoOracleL2Storage, RenzoOracleStorage, RewardHandler, RewardHandlerStorage, RoleManagerStorage, StETHShim, StructuredLinkedList, WBETHShim, WBETHShimStorage, WETHUnwrapper, WithdrawQueueStorage, xRenzoBridgeReceiverStorage, xRenzoDepositNativeBridgeStorage. Unknown means unassessed, not peripheral risk.

The RestakeManager comparison includes eleven differing and four added files; the EzEthToken comparison includes four differing and one added file. They are materially different from the compared November 2023 revision. The supplied differences add a risk-oracle interface and role-check methods. Later audits exist, but their deployed-version equivalence has not been established. All supplied auditor-fix deployment checks are indeterminate and support no conclusion about deployment of reviewed fixes; this does not alter auditor-fixed statuses.

The provider context reports 37 runtime-matched addresses, two with Etherscan-listed source and four unverified, among 43 addresses. Sourcify runtime matches support the listed deployed-source identities. Etherscan-listed source is only an explorer verification claim. Neither is proof of audit coverage.

Additional deployment context

The later on-chain collection identifies contracts absent from the address table used by the coverage map. Its code-derived names add deployment context, not audit-version matches or new finding checks.

Chain and measurement blockNamed contracts absent from the original address table
Ethereum; 26345115 (2026-11-05T00:43:23+00:00) Outbox, GnosisSafeProxy, WETHPaymentReceiverEzETH, WETH9, DelegationManager, DepositQueue, RenzoOracle, RiskOracleMiddleware, RoleManager, BackingEigen, Eigen, RewardsCoordinator, StrategyBase, UpgradeableBeacon, WithdrawalBuffer, AllocationManager, PauserRegistry, StrategyFactory, WstETH, RatiosStrategy, SymbioticAdapter, ConnextDiamond, Router, LinkToken, BalancerRateProvider, EigenStrategy, Bridge, RollupAdminLogic, UpgradeExecutor
Base; 53525067 (2026-11-05T00:44:41+00:00) L2ERC20TokenBridge, xRenzoDepositNativeBridge, ERC20Bridged, EACAggregatorProxy, EzAdminL2, XERC20Factory, GnosisSafeProxy, L2CrossDomainMessenger, Recovery, SentryDelegation, HyperlaneReceiver, WETH9, AccessControlledOCR2Aggregator, ConfirmedTransactionModule
Arbitrum One; 521332195 (2026-11-05T00:46:13+00:00) ERC20Bridged, xRenzoDepositNativeBridge, EzAdminL2, XERC20Factory, GnosisSafeProxy, L2ERC20Gateway, L2ERC20TokenGateway, HyperlaneSender, aeWETH, UpgradeExecutor
BNB Smart Chain; 131723337 (2026-11-05T00:47:09+00:00) EzAdminL2, XERC20Factory, BridgeToken, ConnextDiamond, BEP20Ethereum, HyperlaneReceiver
Linea; 32520823 (2026-11-05T00:47:10+00:00) EzAdminL2, XERC20Factory, xRenzoDepositNativeBridge
Blast; 42493259 (2026-11-04T12:45:33+00:00) EzAdminL2, OptimismMintableXERC20Factory
Sei; 241552566 (2026-11-05T00:47:31+00:00) EzAdminL2, XERC20Factory

These entries are a gap in the original contract list. Their full measured records, including unnamed control addresses, appear in the on-chain context.

Audit-named components still lacking an address in that collection: UsccDepositStrategy, UsccDepositWithdrawalHelper, LEZyVault, EulerSwapEzEthDelegateStrategy, WithdrawQueue, LEZyVaultFactory, AaveV3DelegateStrategy, MorphoDelegateStrategy, WithdrawQueueStorage, Errors, LEZyVaultStorage, BaseDelegateStrategy, LEZyVaultFactoryStorage, RoleManagerStorage, EzRVaultsFactory, EzRVaultsFactoryStorage, EzRvaultStorage, EzEthTokenStorage, RestakeManagerStorage, IxRenzoDeposit, RenzoOracleL2Storage, RenzoOracleL2, xRenzoDepositStorage, CCIPReceiver, ConnextReceiver, IxRenzoBridge, xRenzoBridgeStorage, LockboxAdapterBlast, METHShim, METHShimStorage, RenzoOracleStorage, WBETHShim, WBETHShimStorage, RewardHandlerStorage, RewardHandler, BalancerRateProviderStorage, OperatorDelegator, OperatorDelegatorStorage, DepositQueueStorage, TimelockController, xRenzoDepositNativeBridgeStorage, HyperlaneReceiverStorage, LidoOPValueTransfer, EthOPValueTransfer, LidoArbValueTransfer, EthArbValueTransfer, xRenzoBridgeReceiver, IwstETH, xRenzoBridgeReceiverStorage, StETHShim, OperatorDelegatorAdminLib, WETHUnwrapper. No controls or deployments are inferred for them.

The coverage map reports DepositQueue without an attributed address; the later Ethereum measurement at block 26,345,115 identifies DepositQueue at 0xf2f305d14dcd8aaef887e0428b3c9534795d0d60. This does not supply a deployed-source comparison for the historical findings.

At Ethereum block 26,345,115, the history records RestakeManager’s implementation being replaced on December 11, 2025, after Nethermind’s April 14, 2025 audit. This is a version-coverage consideration, not a vulnerability finding. Blast upgrade history was not read for seven listed addresses; its retained measurement is block 42,493,259 on November 4, 2026.

Findings by Severity

Only groups not marked fixed are listed; fixed findings are trusted and appear solely in counts.

SeverityTotal issue groupsFixed per the auditorNot marked fixed by the auditor
Critical770
High1091
Medium963

Critical

Total groups: 7; fixed per the auditor: 7; not marked fixed: 0.

High

Total groups: 10; fixed per the auditor: 9; not marked fixed: 1.

  • Malicious Validator Frontrunning Attack — Sigma Prime, June 2024. Audit status: acknowledged. A validator operator could initialize its own withdrawal credentials before Renzo’s stake. The team cited permissioned validators with commercial agreements. Check: the default-branch snapshot (Renzo-Protocol/contracts-public@fc8c8e0) may still contain the code this finding describes; low confidence, deployment not confirmed (repository evidence). The check shows forwarding to the manager; downstream checks and operational protections are outside the supplied passage.

Medium

Total groups: 9; fixed per the auditor: 6; not marked fixed: 3.

  • Incorrect stakedButNotVerifiedEth Accounting Can Overinflate TVL — Sigma Prime, June 2024. Audit status: acknowledged. The retest found a residual stake–verify–stake accounting case despite an earlier implementation change. Check: the code was rewritten; the check could not determine whether this finding still applies (default-branch repository snapshot, Renzo-Protocol/contracts-public@fc8c8e0). Credential verification now delegates accounting to a library; the check is inconclusive.

  • Double Use Of Validator Public Key Can Re-introduce Security Issues — Sigma Prime, June 2024. Audit status: acknowledged. Reusing a key across delegators can reintroduce accounting discrepancies. The team accepted the operational risk. Check: the default-branch snapshot (Renzo-Protocol/contracts-public@fc8c8e0) may still contain the code this finding describes; low confidence, deployment not confirmed (repository evidence). Local key accounting is visible; caller restrictions and off-chain administration were not resolved by this passage.

  • Inaccurate queue filling with rebalancing assets — Halborn EVM Managed Vaults, engagement September 26–October 2, 2025. Audit status: acknowledged (risk accepted October 8, 2025). Interest-bearing or rebalancing underlying assets can cause excess queue funding. The team states it does not intend to use rebasing underlying assets and added a warning. Check: not marked fixed per the auditor (acknowledged); the code check is unavailable because the audited repository was unavailable. This is not confirmation that the issue remains live.

Recurring Patterns

No not-fixed critical/high/medium ledger group is reported by more than one audit.

Within Sigma Prime’s review, validator-key handling connects the high frontrunning issue and both medium accounting issues. This is a shared technical theme, not an additional merged finding. The separately captured contest findings require cross-audit deduplication before recurrence can be assessed across that additional source.

Code Quality Assessment

The sources show repeated remediation and multi-firm review of ezETH, with uneven evidence for tests, current versions and the wider product set.

The ledger records 22 of 26 critical/high/medium groups fixed per the auditor. The four remaining groups are acknowledged. Finding density is not meaningfully comparable across engagements because scopes and deduplication differ. Repo-only checks are evidence about Renzo-Protocol/contracts-public@fc8c8e0, not proof of a release or deployment.

Nethermind rated its April 2025 core test suite Medium and the May 2025 bridge test suite Low. The bridge tests detected 129 of 430 injected changes, a 30.00% mutation score. For xRenzoDepositNativeBridge, they detected 10 of 267 changes, or 3.75%. Mutation testing measures whether deliberate code changes are caught by tests. Sigma Prime also described its review as time-boxed and recommended further investigation of areas containing severe issues.

Renzo’s documentation states that a $250,000 Immunefi bounty is live and links public contract source. Sourcify matches are supplied for important Ethereum core contracts, but verification and publication are not established for the full product surface. No cited technical monitoring or automated-pause service establishes the additional assurance condition.

An unverified third-party assessment instead claims a $500,000 maximum for critical smart-contract reports, capped at 10% of funds at risk with a $100,000 minimum. That claim remains unconfirmed (source reliability: unverified publisher). The current payout ceiling and asset scope could not be confirmed from Immunefi because retrieval of its program page failed. The documented $250,000 amount is retained as the supported statement.

Review depth differs by product: ezETH has several independent firms, whereas the included managed-vault reviews identify one firm. Broader depth is not transferred to pzETH or Solana products. Deployment assurance is constrained by missing audited-version links and partial code-criticality assessment.

CertiK’s October 24, 2026 curated snapshot grades Renzo AA; Exponential’s April 10, 2026 archived snapshot labels its code quality good. These are dated external assessments, not proof of current deployed-code safety.

Known compiler bugs

The supplied compiler inventory covers Solidity 0.7.6, 0.8.10, 0.8.12, 0.8.15, 0.8.19, 0.8.20, 0.8.25, 0.8.27 and 0.8.30 across protocol and dependency contracts. Applicability has not been checked; these entries are not vulnerability findings or factor evidence.

Compiler bugSupplied severityConditions and triggerFixed in
KeccakCachingmediumoptimizer enabled; Keccak-256 caching, especially inline assembly. 0.8.3
UnsoundSpillInMutualRecursionmediumviaIR enabled; mutually recursive functions with spilled local variables. 0.8.36
SpillSlotCollisionAcrossMutualRecursionlow/mediumviaIR enabled; mutual recursion and stack-limit memory slots collide. 0.8.37
TransientStorageClearingHelperCollisionhighviaIR enabled and Cancun or later; both storage and transient-storage clearing. 0.8.34
InheritanceOrderReversalOnStorageEndWarningmediumno configuration filter supplied; warning about a storage base near the storage end. 0.8.36
MemoryByteArrayElementDeleteClearsWholeWordlow/mediumlegacy code generation; deleting one element of a memory bytes array. 0.8.37
AbiReencodingHeadOverflowWithStaticArrayCleanupmediumABIEncoderV2; static calldata array last in a tuple with dynamic components. 0.8.16
StorageWriteRemovalBeforeConditionalTerminationmedium/highYul optimizer enabled; assembly return/stop conditionally ends the external call. 0.8.17

The inventory lists 12 other lower-severity bug names. No bounded code observation establishes any compiler bug’s trigger here.

Score: 6.0/10

Rubric security-score-v2.6; data completeness 1.00

Provisional — data incomplete: Audit coverage: the protocol's current deployed contract addresses. This mark does not change the score.

Code security factors, points out of 10 (points)
Audit coverage
5.50 points
Audit depth
5.50 points
Unresolved findings
9.00 points
Assurance
5.50 points
Audit findings not marked fixed (findings)
High · acknowledged
1 findings
Medium · acknowledged
3 findings
FactorLevelEvidenceWeightPoints
Audit coverage (computed)L2supported (provisional)0.355.50
Audit depthL2supported0.205.50
Unresolved findings (computed)L4supported0.309.00
AssuranceL2supported0.155.50

157 citations · click markers to view evidence

💰

Economic Security

4.2/10

Economic Security Analysis: Renzo

Renzo users face principal risk from restaking losses, dependent protocols, and exits that can wait through withdrawal queues. Measured DEX routes do not establish enough capacity for a rapid $100,000 exit in canonical ezETH or pzETH. Related ezREZ claims depend on Renzo’s own REZ token and share upgrade control with ezETH. Incomplete backing, reward-split, and dependency measurements limit confidence in protection during stress.

The TVL record covers ezETH, its cross-chain representation, pzETH, ezSOL, bzSOL, and ezJTO. The captured TVL adapter counts Ethereum ezETH net of the lockbox, pzETH, cross-chain ezETH, and three Solana token accounts. The complete adapter contains no Reserve Vault, ezREZ, ezEIGEN, ezBTC, or Opolis position. ezREZ and ezEIGEN are additionally assessed for their evidenced shared controls, as explained below. Other products remain context unless a shared exposure is established. The adapter marks its TVL as double-counted across DeFi, so it is not a measure of unique system-wide capital.

At Ethereum block 26,345,115 on November 5, 2026, ezREZ and ezEIGEN share ezETH’s upgrade timelock, 0x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7. This connects their control exposure even though their balances are outside the captured TVL adapter. Each product’s collateral and losses remain distinct; shared control does not establish that REZ backs ezETH or pzETH.

The November 5 product captures discontinue ezSOL, bzSOL, and ezJTO minting; their withdrawal instructions remain available.

TVL & Liquidity Profile

The latest captured TVL is $102.09 million, concentrated on Ethereum and Linea.

The historical series fell $24.27 million over 30 days through November 3, 2026. Price changes explain $20.71 million; token-quantity change at current prices explains $3.56 million. Over 90 days, TVL rose $9.56 million despite a $9.30 million quantity decline, because price effects added $18.86 million. Coverage is 100% in both decompositions.

ChainTVL, USD millionShare of captured TVL
Ethereum73.994972.4821%
Linea16.247715.9155%
Arbitrum3.88453.8050%
Solana3.31063.2429%
Zircuit1.98701.9464%
Mode1.27861.2524%
Base0.64850.6352%
Optimism0.45400.4447%
Fraxtal0.15650.1533%
Binance0.10930.1070%
Monad0.01410.0138%
Sei0.00160.0016%

Shares are calculated from the captured chain values.

Blast was deployed earlier and has no current TVL; that deployment is not assessed.

The November 5 Ethereum exit probes target WETH through KyberSwap. For canonical ezETH, the $100,000 probe reported 15.85% impact; pzETH reported 97.87%. Both record $0 qualifying depth at 2% impact and no named venue. These coarse results mean no fetched probe qualified; they do not prove all smaller trades or other venues lack liquidity.

The 0.5% depth is unavailable for both assets. When available, that measure is a lower bound from coarse probes, not exact capacity; null is a gap. The xERC20 ezETH wrapper measurement errored and is not substituted for canonical ezETH. No successful USDC-target route or named venue is supplied. ezSOL, bzSOL, and ezJTO remain exit-depth gaps. No supplied market-cap or exit-depth measurement covers ezREZ or ezEIGEN; their 0.5% depth is also unknown.

Native exits provide an alternative: ezETH documentation states up to 15 days; pzETH states 14–21 days; ezSOL states two epochs, up to four days.

For the related ezREZ and ezEIGEN products, documentation states up to 14 days with a manual claim of REZ or EIGEN. These are product-specific normal exit descriptions, not proof of a bounded emergency pause.

The November 5 asset snapshot reports canonical ezETH market capitalization of $104.24 million and pzETH of $3.04 million. These market values do not measure immediately executable exit liquidity or verify backing.

Current beneficial-owner and depositor concentration is unavailable. The April 2, 2024 holder snapshot includes pooled contracts and cannot establish current whale ownership or justify counting those pools as single holders.

Oracle Risk

Chainlink supplies documented pricing inputs, but current checks and fallback behavior are not established for every product and chain.

Renzo names Chainlink as its primary oracle for stETH deposits and L2 price feeds.

Nethermind’s May 1, 2025 bridge review describes an Arbitrum rate oracle, Hyperlane transmission, sender validation, and updates to the destination mint rate. This establishes the reviewed architecture, not every current receiver’s configuration.

At Ethereum block 26,345,115 on November 5, 2026, RenzoOracle identifies stETH and secondary oracle 0x86392dc19c0b719886221c78ab11eb8cf5c52812. The read does not establish all primary-rate checks or a tested fallback.

Sigma Prime’s June 2024 review documented a 1% divergence check, including owner-forced rate updates. Its separate stale-L2-rate finding was resolved by adjusting the feed divergence threshold. Neither finding proves the same bounds are deployed today.

The 2024 Llama Risk assessment reported a centralized Binance wBETH input without liveness or anomaly checks. Current ezETH docs instead list ETH and stETH; the measured collateral-token count is one, but its identity was not read. The historical wBETH path is therefore an unresolved exposure question, not a confirmed current oracle flaw. A supplementary Ethereum archive read was refused by the provider, leaving that identity unconfirmed.

Assessment: stale cross-chain rates could misprice minting before collateral arrives. Impact can include excess receipt issuance; likelihood cannot be narrowed beyond possible without current per-path freshness, deviation, and owner-update checks. The available sources do not establish a current oracle-caused loss or live spot-DEX pricing path.

Flash Loan & MEV Exposure

The evidenced economic attack surface centers on accounting and transaction ordering; no current flash-loan exploit is demonstrated.

Sigma Prime’s June 2024 review describes operator-initiated undelegation excluding collateral from TVL, allowing excessive ezETH minting. The report explicitly says flash borrowing could amplify profits. It records remediation in commit 210aba1, while retesting found a circuit breaker blocking reward accrual. This historical review does not establish the vulnerability in the live implementation.

Halborn’s May 31, 2024 withdrawal assessment describes direct transfers inflating reported backing and affecting ezETH minting through sandwich attacks. Its summary does not state that item’s remediation status. The publication was supplied through the team’s repository and is not verified as auditor-controlled, so confidence in that audit-only claim is lower. It is a historical finding, not a confirmed live vulnerability.

Nethermind’s April 14, 2025 review reported a distinct slashing-accounting path that lowered TVL before a deposit, improving the attacker’s mint rate. That item is marked fixed, using withdrawable-share accounting.

Sigma Prime’s June 2024 review also documents a malicious validator operator front-running initialization with 1 ETH and attacker withdrawal credentials before Renzo’s 32 ETH deposit. The team acknowledged the risk and relied on permissioned validators under commercial agreements. That response is not a technical fix; the current deployment’s exposure was not established.

Reward auctions create competition over transaction ordering. Anyone can purchase the reward basket as its price declines; bidder controls include epoch, deadline, and maximum payment. These constrain stale bids but do not remove competition to execute first.

Flash loans can amplify capital-dependent accounting or pricing attacks. The cited historical defects warrant attention to rate and backing accounting, but do not establish such a defect in the live version. External DEX sales also expose users to slippage and ordering costs.

Tokenomics Risk: Yield, Leverage & Reflexivity

Staking and restaking provide named reward sources, but the share of paid yield funded by REZ incentives is unavailable.

  • ezETH: Ethereum staking and EigenLayer AVS rewards, auto-compounded into its exchange rate.
  • pzETH: ETH staking plus prospective AVS rewards; its FAQ also names Symbiotic and Mellow points. The FAQ still describes AVSs as not yet functional. These are documented sources and program statements, not measured current revenue allocations.
  • ezSOL: older documentation names JitoSOL staking, Jito restaking, and MEV rewards. The November 5 page discontinues minting; current paid rewards are not quantified. MEV means value extracted through transaction ordering; here the older mechanism docs name it as an income source.
  • bzSOL and ezJTO: older mechanism pages describe restaked BNSOL and JTO, respectively, with staking/restaking rewards. Their newer pages stop minting; continued reward amounts and remaining positions are not quantified.

Through November 3, 2026, captured protocol fees total $111,753 over 30 days and $9,634,545 over 365 days. Protocol revenue is $18,371 and $1,389,986 for those periods. These aggregate amounts do not allocate rewards among tokens or establish the revenue-versus-emission share of yield paid to users.

REZ is Renzo’s governance token. The Rewards program explicitly distributes REZ through integrations, but its current dollar value and share of paid yield are unmeasured.

Current-version product documentation describes deposits exchanged for restaking receipts, with withdrawals against underlying collateral. They do not describe a native borrowing position in these basic receipts. Users can borrow or loop through external DeFi venues, but current venue-specific maximum LTV, leverage, and unwind rules are unavailable. The Compound-loop and USCC leveraged Reserve Vaults are other products outside this adapter; their leverage cannot be assigned to the core restaking deposit.

The documented collateral is ETH/LSTs for ezETH and pzETH, JitoSOL for ezSOL, BNSOL for bzSOL, and JTO for ezJTO. These five core receipt mechanisms do not identify REZ as backing or a redemption asset. The separate ezREZ claim does redeem against REZ; its share of aggregate liabilities is unavailable.

ezREZ, a related product with shared upgrade control, mints claims against REZ. A fall in REZ’s market value therefore reduces the value of its backing even when claims remain covered in REZ units. This is product-specific own-token exposure, not evidence of a current spiral or REZ collateral in ezETH. Documentation names AVS and partner-token rewards, plus weekly compounding of instant-withdrawal fee revenue. The measured allocation among these sources and the own-token share of paid yield are unavailable.

ezEIGEN is the other related product with shared upgrade control. It represents restaked EIGEN and compounds EIGEN rewards; EIGEN is another issuer’s token, not REZ. Its document’s August 2024 first-year incentive allocation is historical and does not establish present paid yield.

The November 4 composition totals $78.97 million and labels 96.12% EZETH and 3.88% PZETH. Those are receipt-token accounting entries, not evidence that Renzo backs its claims with its governance token or a circular promise. The composition total also differs from the November 3 protocol TVL; neither can replace a liabilities reconciliation.

Fee documentation says both that all EigenLayer rewards pass through and that restaking rewards carry a 10% fee, split equally between treasury and operators. These statements require clarification before assuming a net yield allocation. They are documentation claims, not verified deployed fee calculations.

On Ethereum at block 26,345,115, the measured DepositQueue at 0xf2f305d14dcd8aaef887e0428b3c9534795d0d60 reports raw feeBasisPoints: 1500. The deployed denominator and exact fee scope were not verified, so this value is not converted to a percentage. It remains unreconciled with the documentation’s 10% reward-fee statement. The address is identified through RestakeManager.depositQueue(); it is absent from the supplied contract table.

Solvency, Liquidation & Bad Debt

Backing assets and redemption mechanisms are identifiable, but full backing against liabilities and uninterrupted exits are not independently established.

ezETH locks in WithdrawQueue when a withdrawal is requested and burns when underlying collateral is claimed. Buffers refill from rewards, deposits, and validator exits; routing new deposits to withdrawals is liquidity management, not evidence that new deposits fund investment returns.

The withdrawal summary states a three-day buffer delay, fourteen-day EigenLayer minimum, and nine-to-ten-day beacon exit component. Its expected wait is seven days with a funded buffer and ten-to-fifteen days otherwise. The fourteen-day upstream minimum and shorter expected completion figures are not reconciled in the documentation. They should not be added into a guaranteed end-to-end bound.

At Ethereum block 26,345,115, the risk middleware’s withdrawal-request, claim, and instant-exit pause flags were false; its own cooldown field was zero. This field does not replace the separately documented queue and upstream delays.

pzETH’s withdrawal pause was also false at that block. Its privileged pause function can execute immediately; unpausing is a separate privileged action. Available mechanism documents give no maximum suspension duration. This is an exit-availability risk despite withdrawals being unpaused at the measurement.

The November 5 asset read reports pzETH’s underlying as wstETH, total assets of 1,006.296666 wstETH, and a reported conversion rate of approximately 1.000157 whose unit is unconfirmed. This is a measured vault rate, not an independently established 1:1 ETH redemption promise or full protocol solvency proof.

Instant ezETH withdrawals charge a buffer-dependent fee and revert below the minimum buffer threshold. They are disabled when paused, so they do not guarantee stress liquidity.

The assessed core receipt deposits are not documented as debt positions requiring borrower liquidation. Liquidations, keeper permissions, incentives, and debt absorption matter when these tokens back external borrowing; current venue-specific records remain unavailable. The 2024 Llama Risk report records Morpho bad debt after the ezETH market depeg, which concerns downstream lending, not proof of a Renzo backing deficit. No reserve-deficit dataset is supplied for current lending exposures.

Incentive Alignment

Underlying staking rewards support the economic model, while adjustable REZ incentives can attract capital that leaves when rewards fall.

The ezPoints program ended with the March 31, 2025 snapshot. Its replacement, Renzo Rewards, distributes REZ through selected integrations, with periodic allocation changes. The fraction of present TVL tied to incentives is unknown.

No dedicated Renzo insurance, safety-module, or compensation fund with a usable size and payout mandate was identified in the documentation or supplementary research. The treasury receives fees, but no cited purpose commits it to covering user losses. WithdrawQueue is redemption liquidity, not insurance. Fund size versus TVL, payout triggers and priority, veto powers, and loss-cover cooldowns remain unavailable.

Opolis Reinsurance Bonds are another product: they underwrite health-insurance claims and can slash the bond investors’ principal. They are not evidence of insurance protecting ezETH or other Renzo depositors.

No completed Renzo reimbursement for the April 2024 depeg is established. The dated Llama Risk report says Gearbox bad debt would be covered by reserves; a promise does not prove payment. Available repayment reports do not establish a completed Renzo payment or coverage of all liquidated users.

The captured DeFiLlama hacks file contains no records. This is not a complete loss history or proof that no losses occurred. A dedicated loss-cover fund or completed compensation payment remains unestablished.

Economic Attack Vectors

Principal depends on multiple external systems, and a delayed exit can turn a market discount into a leveraged user’s realized loss.

Critical strategy layers are:

  • ezETH: Renzo deposit and operator contracts → Ethereum staking/EigenPods → EigenLayer operators and AVSs. stETH deposits additionally depend on Lido.
  • pzETH: Lido stETH/wstETH → Mellow vaults → Symbiotic collateral and operators/AVSs. The FAQ describes collateral waiting idle in Mellow before entering Symbiotic.
  • ezSOL: historical deposits converted SOL through Jupiter or Jito minting → JitoSOL → Jito restaking and NCNs. Current minting is discontinued. Each layer can impair principal or exit availability.
  • Cross-chain positions: Hyperlane messaging and the applicable canonical asset bridges, with Chainlink rate inputs. Nethermind’s bridge review describes optimistic issuance before settlement into Ethereum collateral.

bzSOL additionally relies on Binance BNSOL, and ezJTO on JTO and Jito restaking. Jupiter conversion flows for these tokens and ezSOL are historical deposit context because current pages discontinue minting. Both current withdrawal pages retain the two-epoch, up-to-four-day queue and total 0.2% withdrawal charge.

The related ezREZ and ezEIGEN deposits pass through their own receipt vaults → EigenLayer strategies → delegated operators and AVSs. Their shared upgrade timelock does not establish shared collateral pools.

At least three external dependencies are evidenced: EigenLayer, Lido, Mellow, Symbiotic, and Jito, alongside bridge and pricing services. A complete current map of integration exposure caps, operator allocations, and remediation histories is unavailable. A single-verifier bridge failure is not established; the relevant verifier thresholds and transfer caps would decide that narrower risk.

Ethereum holds 72.48% of captured TVL and is classified as the base Ethereum chain; rollup stage and exit window do not apply. Linea holds 15.92% and is a stage-0 rollup. Its raw L2BEAT exit-window value is None, not a measured upgrade delay.

The remaining chain shares appear in the liquidity table. Arbitrum, Base, and Optimism are stage 1; Zircuit, Mode, and Fraxtal are stage 0, each with raw exit window None. Solana, Binance, Monad, and Sei are classified as other chains, with stage and exit window unknown. No chain-age claim is made without a launch-date source.

The available dependency map contains no scored protocol-dependency edges, no oracle-provider nodes, and no circular-backing or downstream-holder warnings. Its composition input is dated November 4 and TVL input November 3, 2026. Its only backing symbols are unresolved EZETH at 96.12% and PZETH at 3.87%; it does not look through these receipt tokens into underlying protocols. No dependency score, freshness deadline, or published-score comparison is available.

Historical evidence demonstrates the liquidation channel: the 2024 ezETH discount produced downstream liquidations and bad debt. Today, queue delays and thin measured secondary routes can again restrict arbitrage during an exit wave. That scenario has high impact for externally leveraged users; its present likelihood depends on unavailable leverage and pool exposures.

Nethermind’s April 14, 2025 review also acknowledged a checkpoint synchronization failure that can block new reward shares and validator-exit withdrawals. The team relied on off-chain watcher timing. This is an acknowledged historical issue; its status in the deployed implementation was not compared.

Aggregator Scores

Dated third-party ratings provide context but do not establish current economic controls.

CertiK’s October 24, 2026 captured security assessment is 87.21. Its freshness note explicitly disclaims a guaranteed current state.

Exponential’s April 10, 2026 archive snapshot gives a risk rating of Best and a reported value of 9.0. This is a historical external assessment, not this report’s computed result.

The ratings recognize the restaking design and reviewed code. They do not resolve the measured exit limits, pause duration, full backing reconciliation, or reward-emission split. No DeFiSafety assessment is supplied. DeFiLlama TVL remains the November 3 captured snapshot described above.

Data Completeness

Evidence is strongest for named staking mechanisms, captured TVL, native withdrawal descriptions, and sampled Ethereum controls; confidence is moderate overall.

The assessment uses mechanism documentation, audit sections, dated risk research, aggregator measurements, and published supplementary captures. No article collection or Blockworks quarterly report is supplied. Audit scopes were checked individually: ezETH and bridge reviews do not establish controls for pzETH, Solana tokens, or managed vaults. The team-hosted Halborn withdrawal report has unverified publication provenance; that lowers source confidence, not the protocol’s risk level by itself.

Material gaps remain:

  • Current revenue allocation and own-token emissions as a fraction of paid yield.
  • Current beneficial-owner concentration and campaign-linked TVL.
  • Full backing versus liabilities, including funds awaiting bridge settlement.
  • Current checks and fallbacks for every price and rate, including the historical wBETH path.
  • Complete dependency exposure, isolation caps, operator allocation, and verifier thresholds.
  • Executable exit capacity below $100,000, 0.5% depth, and non-Ethereum receipt-token depth.
  • A dedicated loss-cover mandate, fund size, and completed compensation record.
  • Current-version bad-debt and liquidation evidence where users borrow or loop.

The current revenue split and debt-position details remain unavailable after supplementary research. Loss-cover research did not establish a qualifying fund or completed payout. The RPC provider refused the supplementary archive read, leaving collateral identity unconfirmed. Insufficient data for this assessment of a complete current solvency ratio, paid-yield emission share, or guaranteed stress-exit capacity.

Score: 4.2/10

Rubric security-score-v2.6; data completeness 1.00

Provisional — data incomplete: Dependency risk: the protocol's current deployed contract addresses. This mark does not change the score.

Economic security factors, points out of 10 (points)
Solvency redemption
3.50 points
Oracle design
5.50 points
Dependency risk
3.50 points
Liquidity concentration
5.50 points
Yield design
3.50 points
FactorLevelEvidenceWeightPoints
Solvency redemptionL1supported0.353.50
Oracle designL2supported0.255.50
Dependency riskL1 (chain risk)supported (provisional)0.153.50
Liquidity concentrationL2supported0.105.50
Yield designL1supported0.153.50

101 citations · click markers to view evidence

👥

Team & Transparency

6.4/10

Social & Team Transparency: Renzo

Renzo has identifiable leadership and a named foundation, alongside public technical documentation and recent repository activity. These are useful accountability signals, but complete leadership histories and current community health remain unverified. The April 2024 ezETH depeg is a material negative in its history; available evidence does not establish the final user-loss outcome.

Team Identity

Renzo is publicly identifiable, although its full leadership and ownership structure remain only partly documented.

The official blog names James Poole as co-founder and credits him with building a Model Context Protocol server, announced February 18, 2026. This establishes a named leader and a specific first-party delivery claim, rather than an independently verified professional history.

The website terms name RestakeX Ventures Ltd. as the portal provider. Token-distribution terms, revised September 19, 2024, identify RestakeX Foundation as a Panama foundation. These establish organizational identity; they do not establish the foundation's current directors or beneficial owners.

The website terms select British Virgin Islands law and Cayman Islands arbitration. Governing law does not prove where RestakeX Ventures Ltd. is incorporated. Complete independently verified biographies, prior-project records and a current core-team roster were not established. Insufficient data for this assessment of individual professional track records.

Track Record

Renzo has documented operation since December 2023, but its history includes an ezETH depeg that harmed leveraged users.

A historical Prisma/Llama Risk assessment dates the beta mainnet launch to December 18, 2023. Family listing history and current contract creation dates are assessed separately; an old proxy's creation does not date every later implementation. The current official directory identifies RestakeManager and a withdrawal contract as core Ethereum components.

The supplied DefiLlama hacks and DeFiHackLabs snapshots contain no matched records. This means no qualifying exploit was found in those datasets, rather than proof of an incident-free history. The separately documented April 2024 depeg remains relevant.

On April 23, 2024, ezETH sharply depegged and subsequently traded at a discount. The historical Prisma/Llama Risk assessment describes liquidations and bad debt in external lending markets. This is treated as an economic-design incident affecting this deployment, rather than an established contract exploit.

Incident-date TVL, completed comprehensive reimbursement and net uncovered user losses are unknown. The incident is retained as major, the conservative rule-based default. This classification does not assert a proven loss-to-TVL threshold, protocol insolvency or a proven amount of uncovered user loss. Monthly TVL samples cannot supply the missing incident-day denominator.

Historical reporting described disabled withdrawals during that event. Current ezETH documentation says Ethereum mainnet withdrawals are enabled and can take up to 15 days. That closes one historical exit limitation, but does not establish full reimbursement or a verified incident fix.

Measured own tokens are Ethereum ezETH, its L1 bridge xERC20 representation, and pzETH. The pzETH directory's displayed address is truncated; its linked explorer URL contains the complete address used in the structured record.

The token measurements provide no usable daily depeg history: ezETH exchange rates were unavailable, and pzETH exchange-rate units were unconfirmed. Empty event lists therefore do not demonstrate price stability. No qualifying evidence establishes a 1:1 reference floor for these records. Complete recovery duration, outage history and migration outcomes are also unavailable, so the evidence is insufficient to assess resilience.

Social Presence

Public development continues, while current social reach and discussion quality remain unmeasured.

Across the three listed public repositories, commits occurred in three of the last twelve calendar months: December 2025, February 2026 and October 2026. The latest commit was October 15, 2026. The core contracts-public repository's activity was confined to December 2025 in that window; October activity came from operator-metadata. Contributor counts were not supplied.

The supplied official blog index lists product announcements and a month-in-review post, with its latest dated entry on February 18, 2026. It does not establish monthly communication during the last six months. An older blog index alone cannot establish that the team stopped communicating elsewhere.

The link inventory provides the application, GitHub organization and X account. It provides no Discord or Telegram URL. Attempts to refresh official community links and recent communications failed, leaving those facts open. The governance forum describes announcements in Discord and shows proposal and delegation categories, demonstrating community infrastructure rather than measured participation quality.

Current X followers, authentic engagement, Discord/Telegram membership and moderation quality are unavailable. Insufficient data for this assessment of community health.

Transparency Indicators

Technical disclosure is useful but does not establish complete financial or incident transparency.

Official documentation explains the deposit-token mechanism and publishes Ethereum contract addresses. Some current component bindings remain unresolved, preventing a complete inventory.

Another Renzo product, the USCC Leverage Vault, leaves its Aave Looping Strategy address unspecified in its contract list. This is a family-level disclosure gap; it does not establish ezETH or pzETH controls.

The audit index links named reports, including Halborn withdrawals from May 2024, Sigma Prime from June 2024 and Nethermind from April 2025. These are evidence of publication; they do not establish coverage of every current product or a regular review schedule.

Renzo's docs state that an Immunefi bounty is running. The source-code page advertises a maximum of $250,000, but its footer says it was last updated one year ago. Current payout and active status remain unconfirmed because the attempted web refresh failed.

Smart-contract source is publicly linked, and repository activity is measured. Full public-source completeness across current products is unverified. Consolidated periodic treasury, reserve and expense reports, and a first-party post-mortem for the depeg, were not established.

Red Flags

The strongest supported concern is the depeg and the incomplete record of its aftermath.

The historical risk assessment links a controversial token allocation to selling pressure and says the associated social post was deleted. This is a dated transparency concern, not evidence of a present policy of suppressing content.

A sudden current social-media silence, unrealistic return promises and unattributed code copying were not established. Missing measurements do not establish those allegations.

Positive Signals

Named accountability, technical publication and continuing commits provide concrete positive signals.

The foundation's stated jurisdiction and the named co-founder provide more accountability than an unidentified team. Published contract addresses, audit links and bounty information allow external scrutiny. Recent public commits show continuing work, although they do not establish development depth or roadmap adherence.

Governance documentation describes proposal discussion and voting. The supplied evidence does not measure voter breadth, substantive participation or execution consistency.

Data Completeness

Technical and legal sources are available; social quality and complete incident outcomes remain the main evidence gaps.

This analysis uses collected docs, document digests, repository measurements and targeted captured pages. It is not based on live social-media scraping. No article digests, article freshness index or Blockworks quarterly reports were supplied; historical risk research is used only for its historical claims.

The missing facts that could change this assessment are verified leadership histories, current community metrics, recent communication cadence, complete contract attribution, financial-reporting cadence and the depeg's final user-loss and recovery record. The absence of usable depeg measurements prevents an affirmative clean-history conclusion.

Score: 6.4/10

Rubric security-score-v2.6; data completeness 1.00

Provisional — data incomplete: Disclosure: the protocol's current deployed contract addresses. This mark does not change the score.

Team transparency factors, points out of 10 (points)
Team identity
7.50 points
Disclosure
5.50 points
Development activity
5.50 points
FactorLevelEvidenceWeightPoints
Team identityL3supported0.457.50
DisclosureL2supported (provisional)0.355.50
Development activityL2supported0.205.50

Track record: 6.1/10

Rubric security-score-v2.6; data completeness 0.80

Track record factors, points out of 10 (points)
Operating history
7.50 points
Incident record
6.25 points
Resilience
3.50 points
FactorLevelEvidenceWeightPoints
Operating history (computed)L3supported0.307.50
Incident record (computed)L2supported0.506.25
Resilience—not disclosed0.203.50

30 citations · click markers to view evidence

🗳

Governance

3.7/10

Governance Analysis: Renzo

Renzo combines Snapshot voting with substantial multisig control. Its main Ethereum upgrade path waits seven days, but other upgrades and fund-affecting settings can change immediately. Users should therefore treat the timelock as protection for particular paths, not every administrative action. Measured contract controls are clearer than the link between voting results and execution.

Governance Structure

Renzo operates a hybrid system: REZ holders vote, while named contracts and multisigs hold execution powers.

The voting page places voting on Snapshot at renzogovernance.eth. Votes are linear and token-weighted, with delegation to any address. Eligible balances include delegated REZ, legacy staked REZ, and ezREZ; the latter two self-delegate automatically.

The documented process has these conditions:

  • Initial proposal submission is restricted to whitelisted wallets; broader access is a later phase. The landing page calls this on-chain voting.
  • Discussion initially lasts five days and may change.
  • Voting initially lasts five days. The stated quorum is 100,000,000 REZ, with a 50% + 1 initial majority; both thresholds can change. Abstentions count toward quorum.

The landing page’s on-chain wording conflicts with the voting page’s explicit Snapshot venue. Neither page establishes automatic execution or a binding multisig charter. Insufficient data for this assessment of the decision-to-execution process. This does not establish that votes are advisory or that signers may disregard them.

Vote concentration is unmeasured: the supplied voting record contains no venues. Top-10 and top-1 shares, quorum as a share of supply, and top voters for the last five ended proposals are unavailable.

A documented decision is RP-4. Renzo’s official account of its passage says it sunset ezPoints in favor of Renzo Rewards; ezPoints concluded April 28, 2025. The latest five ended votes and their results could not be confirmed.

Admin Key Analysis

A 3-of-5 Safe controls major operational powers, and its timelock role does not delay its separate direct powers.

Unless stated otherwise, Ethereum facts below were measured at block 26,345,115, November 5, 2026, 00:43:23 UTC. The main Safe is 0xd1e6626310fd54eceb5b9a51da2ec329d6d4b68a. Its measured configuration is three signatures from five owners, with an empty module list and no guard reported. Modules are contracts that can execute through a Safe independently of its signature threshold.

Its five signer addresses are:

  • 0x0241e41c3a6e1c7e322b976db22a3eb2adef4bcc
  • 0xec8f3975cf04eba38bcf51f3297667abc992c40e
  • 0x7111434ab5018d4db90e8b1e214d44ad537e5870
  • 0xa724796812276dddf1dc65df91e11c1933a72371
  • 0xfaff23c643c89f6c8f2abe3d08e207591af93d1c

The sources do not identify these signers as people or establish organizational independence. Address diversity alone does not establish independent control.

The main timelock is 0x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7, with a measured minimum delay of 604,800 seconds, or seven days. Its roles are:

RoleCurrent holders
Propose and cancelMain 3-of-5 Safe; single key 0xadef586efb3287da4d7d1cbe15f12e0be69e0df0
ExecuteMain 3-of-5 Safe
Administer timelock rolesTimelock itself; main 3-of-5 Safe

The following table uses the measured privileged-function paths. A zero delay means immediate authority; an unknown delay does not mean zero. Emergency-labelled actions are not restricted to pause or cancellation.

Target and effectGuard and holderFastest measured delay
Core Ethereum upgradesEIP-1967 admin → main timelock7 days
Upgrade RiskOracleMiddlewareProxyAdmin owner → main 3-of-5 Safe0 seconds
Set ezREZ withdrawal unlock timeonlyOwner → main 3-of-5 Safe0 seconds
Sweep ERC20 tokens from ezREZ vaultonlyOwner → main 3-of-5 Safe0 seconds
Track emergency ezEIGEN withdrawalonlyOwner → main 3-of-5 Safe0 seconds
Change ezREZ cooldownonlyOwner → main 3-of-5 Safe0 seconds
Change ezREZ delegationonlyOwner → main 3-of-5 Safe0 seconds
Change ezREZ deposit limitonlyOwner → main 3-of-5 Safe0 seconds
Change ezREZ rewards destinationonlyOwner → main 3-of-5 Safe0 seconds
Upgrade pzETH MultiVaultProxyAdmin owner → 5-of-8 Safe0 seconds
Pause pzETH withdrawalsonlyRole → two separate single keys0 seconds
Pause/unpause RestakeManageronlyDepositWithdrawPauserAdmin; holder unresolvedUnknown
Mint ezETHonlyMinterBurner; holder unresolved in this function pathUnknown
Set DepositQueue fee configurationonlyRestakeManagerAdmin; holder unresolved in this function pathUnknown

The ezEIGEN vault has the same owner-controlled cooldown, delegation, limit, rewards-destination and ERC20-sweep functions as ezREZ. The two vaults’ withdrawal buffers also allow the main Safe to update withdrawal-buffer configurations immediately.

RoleManager records the timelock as default administrator. Its ezETH mint/burn role is assigned to RestakeManager and the withdrawal contract. Separate operational roles include the Safe, the timelock, and single keys. This role list does not resolve every function’s guard: no unlimited direct Safe minting claim follows from it.

Assessment: these powers support fast emergency intervention and routine restaking operations. They also leave users exposed to signer compromise, coordinated misuse and administrative mistakes. A timelock on proxy upgrades does not constrain an independently callable owner function.

Upgradability

Renzo’s measured contracts use several upgrade paths, including a protected Ethereum path and immediate admin-wrapper upgrades on other chains.

Renzo’s EigenLayer documentation explicitly states that its contracts are upgradeable through proxies and that a protocol multisig gates upgrades. Proxy upgrades can replace core execution logic, so users depend on upgrade controllers to preserve accounting and withdrawal behavior.

For Ethereum’s principal transparent-proxy path, each contract below points to admin 0x91625601e2bbbeb7171c40c79fadbcfbff6a1982. That admin’s owner is the main seven-day timelock, whose proposer, executor and canceller roles are listed above.

Ethereum proxyAddressUpgrade delay
XERC200x2416092f143378750bb29b79ed961ab195cceea57 days
RestakeManager0x74a09653a083691711cf8215a6ab074bb4e99ef57 days
XERC20Factory0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf37 days
EzEthToken0xbf5495efe5db9ce00f80364c8b423567e58d21107 days
xRenzoBridge0xc1036d6bba2fe24c65823110b348ee80d3386acd7 days
XERC20Lockbox0xc8140da31e6bca19b287cc35531c2212763c20597 days
DepositQueue0xf2f305d14dcd8aaef887e0428b3c9534795d0d607 days
RenzoOracle0x5a12796f7e7ebbbc8a402667d266d2e65a8140427 days
RoleManager0x4994efc62101a9e3f885d872514c2dc7b32358497 days
BalancerRateProvider0x387dbc0fb00b26fb085aa658527d5be98302c84c7 days
Withdrawal contract0x5efc9d10e42fb517456f4ac41eb5e2ebe42c89187 days

The ezREZ and ezEIGEN vaults instead use beacon 0xb9ebb6221c760c51dc43f1a4f2daf69f964ff86a. Both withdrawal buffers use beacon 0x18efef04f525a9bf0c6440b12ebb71bfa2ab5982. Both beacons are owned by the main timelock; each proxy’s measured beacon-upgrade path requires seven days.

RiskOracleMiddleware is the exception within the measured ezETH contract graph: admin 0x7a6bbd68457fea0f88766538c75cdde97336a173 is owned directly by the main Safe. Its upgrade path has zero delay.

The documented pzETH token is another restaking product, using Symbiotic. Its measured MultiVault path is proxy → admin 0x985e459801d37749c331bbd2673b665b9114fb01 → Safe 0x81698f87c6482bf1ce9bfcfc0f103c4a0adf0af0, requiring five of eight signatures with zero delay. The two Safes holding pzETH default-administration and upgrade ownership have the same eight owner addresses. Its 2-of-3 rebalance Safe can set limits and add subvaults; the classifier leaves the effect of rebalancing unresolved. These are separate controls from ezETH.

For the active non-Ethereum deployments measured below, the shared local ProxyAdmin address is 0xba8efd0b58b93e49e120db8cf5578f6c43c3614d. Its owner is an EzAdminL2 contract, itself upgradeable by another local ProxyAdmin owned by a 3-of-5 Safe. The second step permits immediate replacement of EzAdminL2. The recorded fastest delay for the first step remains unknown; it must not be silently converted to zero.

Chain and measurementLocal ProxyAdmin ownerAdmin of that owner → Safe
Base, block 53525067 (2026-11-05 00:44:41 UTC)0xa87c18c9865e47f507e0c739d16c336ad764fd950xdb33c25f8a2ce24e6b2543925fbef415c149d191 → 0x8410927c286a38883bc23721e640f31d3e3e79f8, 3-of-5, zero delay to upgrade EzAdminL2
Arbitrum One, block 521332195 (2026-11-05 00:46:13 UTC)0xe5219cf568d366ae4b96efb04d826e6f2e72daa00x5efc9d10e42fb517456f4ac41eb5e2ebe42c8918 → 0x0e60fd361ff5b90088e1782e6b21a7d177d462c5, 3-of-5, zero delay to upgrade EzAdminL2
BNB Smart Chain, block 131723337 (2026-11-05 00:47:09 UTC)0x1bd739c88cb90f88264488b914b6a1398840d4260x47b161ed66ab28876e44e284f5b112c332c0f103 → 0x0e60fd361ff5b90088e1782e6b21a7d177d462c5, 3-of-5, zero delay to upgrade EzAdminL2
Linea, block 32520823 (2026-11-05 00:47:10 UTC)0xbaacd5f849024dcc80520baa952f11adfc59f9d00x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7 → 0xb7092685571b49786f1248c6205b5ac3a691c65e, 3-of-5, zero delay to upgrade EzAdminL2
Sei, block 241552566 (2026-11-05 00:47:31 UTC)0x5247ecbf210f289c244813e89212bc3aed75aac10x2946399b2cf1ec41a1890d81969293de59e9c855 → 0x0e60fd361ff5b90088e1782e6b21a7d177d462c5, 3-of-5, zero delay to upgrade EzAdminL2

Each measured proxy using that first local admin is listed here. Address reuse across chains does not imply shared deployed state. Some older deposit addresses now identify as L2RateProvider; the table retains that measured name.

ChainProxies under the first local ProxyAdmin
BaseXERC20 0x2416092f143378750bb29b79ed961ab195cceea5 ; L2RateProvider 0xf25484650484de3d554fb0b7125e7696efa4ab99 ; XERC20 0xf757c9804cf2ee8d8ed64e0a8936293fe43a7252 ; xRenzoDepositNativeBridge 0x766a4df76cb091c41de22f75b6bd6733fb730357 ; XERC20Factory 0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3 ; HyperlaneReceiver 0xfde1366f26cc3621d944afbc96177a90f2849d26
Arbitrum OneXERC20 0x2416092f143378750bb29b79ed961ab195cceea5 ; L2RateProvider 0xf25484650484de3d554fb0b7125e7696efa4ab99 ; xRenzoDepositNativeBridge 0x2efa125b4fcd5b082a4233ba55b53feb2fb4ff33 ; XERC20Factory 0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3
BNB Smart ChainXERC20 0x2416092f143378750bb29b79ed961ab195cceea5 ; xRenzoDeposit 0xf25484650484de3d554fb0b7125e7696efa4ab99 ; XERC20Factory 0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3 ; HyperlaneReceiver 0x9fa6e50807fbb89695dd7c2d163fe23e93e17699
LineaXERC20 0x2416092f143378750bb29b79ed961ab195cceea5 ; L2RateProvider 0x4d7572040b84b41a6aa2efe4a93efff182388f88 ; XERC20Factory 0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3 ; xRenzoDepositNativeBridge 0x657eac5dc9e0dae2e949391377708acb896b2999
SeiXERC20 0x6dcfbf4729890043dfd34a93a2694e5303ba2703 ; XERC20Factory 0x9bce21489a7a85ed475c9e9d0dc94b4e6a1aebf3

The LlamaRisk assessment discussing April 2024 conditions reported a three-day timelock and no DAO. Current voting docs describe Snapshot; Ethereum state now confirms seven days on the main upgrade path. The timelock history records 259,200 seconds on March 28, 2024, then 604,800 seconds on November 1, 2024. The older statement that all upgrades share that protection does not describe the separate current paths above.

Seven days is not a guaranteed full exit window. Renzo’s ezETH withdrawal docs give expected waits of seven days with buffer liquidity and 10–15 days without it. They separately describe a three-day buffer delay, a 14-day EigenLayer minimum and a 9–10-day Beacon-chain exit time. These estimates are not a verified universal maximum.

The shared ezREZ and ezEIGEN vault controls illustrate the same limitation: both documents describe withdrawals taking up to 14 days, longer than the seven-day beacon delay.

Latest recorded upgrades and pending operations

Recorded upgrade history distinguishes implementation replacements from first assignments. The following are the latest events for the principal Ethereum controls; dates are event dates, not measurement dates.

ContractLatest eventRecorded interpretation
RestakeManager2025-12-11 17:19:59 UTC; block 23990887Implementation replacement; transaction includes main timelock
EzEthToken2025-04-11 19:24:11 UTC; block 22247770Implementation replacement; transaction includes main timelock
DepositQueue2025-07-18 17:03:11 UTC; block 22947455Implementation replacement; transaction includes main timelock
RoleManager2025-06-13 17:21:35 UTC; block 22697096Implementation replacement; transaction includes main timelock
Withdrawal contract2025-10-27 16:33:11 UTC; block 23669994Implementation replacement; no timelock identified in available history
RiskOracleMiddleware2026-07-17 18:15:59 UTC; block 25554143Implementation replacement; no timelock identified in available history
Vault beacon2026-02-06 16:33:35 UTC; block 24399005Implementation replacement; no timelock identified in available history
UpgradeableBeacon2026-02-17 16:41:23 UTC; block 24477810First recorded assignment; no timelock identified in available history
MultiVault2025-07-03 17:13:23 UTC; block 22840109Implementation replacement; no timelock identified in available history

The latest recorded EzAdminL2 replacements occurred on October 17, 2025, on every active measured non-Ethereum chain. Their event times and blocks are:

ChainLatest admin-wrapper upgrade, UTC
Base2025-10-17T15:43:49+00:00; block 36963241
Arbitrum One2025-10-17T15:42:56+00:00; block 390526544
BNB Smart Chain2025-10-17T15:44:57+00:00; block 64955131
Linea2025-10-17T15:48:54+00:00; block 24638591
Sei2025-10-17T15:38:00+00:00; block 173912315

An event without a matched timelock means none of the timelocks whose history was read was identified in that transaction. It does not by itself prove a bypass. The recorded histories contain at most 20 events per list; totals can exceed the displayed list.

The main timelock records 193 executed call entries and four pending call entries, including three entries of one batch. All listed ETAs precede the measurement date; none is recorded as waiting for a future ETA. Pending entries, ordered by ETA and then batch index, are:

ETAFunction selector and target
2025-11-12T08:52:59+00:000xe856218a → 0xdbd0609c918154c1fabda5ee74d8e0aef95eea9c, batch index 0
2025-12-10T15:01:11+00:000x3659cfe6 → 0xad99fb3242e3708c7071c76d4c519c4f76f78de4, batch index 0
2025-12-10T15:01:11+00:000x99a88ec4 → 0x7f38db25b26dd6725a6c7e819198ee8a9238bcf3, batch index 1
2025-12-10T15:01:11+00:000x99a88ec4 → 0x7f38db25b26dd6725a6c7e819198ee8a9238bcf3, batch index 2

The targets of these pending calls are not otherwise resolved in the control inventory. Their user impact remains unknown.

Signer and threshold changes

The main Safe’s latest changes reduced its measured threshold from four to three and replaced signer addresses. The retained history contains no module or guard change entries for that Safe; current modules are empty and its guard read is null.

Date, UTCMain Ethereum Safe change
2026-10-13 14:21:35removed owner 0xc66476c083812ef4af9567ec4b0b737230b5754d
2026-10-13 14:21:35changed threshold 3
2026-10-13 14:21:35removed owner 0x760791319c9f6e12173be39e0f476e68194d54b8
2026-10-06 08:20:35added owner 0x0241e41c3a6e1c7e322b976db22a3eb2adef4bcc
2025-08-16 03:48:11removed owner 0x565e8e4771c0ef8a64c8c7de224816ec1121e671
2025-08-14 00:37:11added owner 0xec8f3975cf04eba38bcf51f3297667abc992c40e
2024-12-13 19:21:11removed owner 0x9d359da6ca68ddbfb148d07b74aa3afd1f65039f
2024-12-12 20:42:59added owner 0x760791319c9f6e12173be39e0f476e68194d54b8
2024-11-01 22:04:59changed threshold 4
2024-11-01 22:04:59added owner 0x7111434ab5018d4db90e8b1e214d44ad537e5870
2024-02-21 18:24:35changed threshold 3
2024-02-21 15:40:59added owner 0xc66476c083812ef4af9567ec4b0b737230b5754d
2024-02-21 15:40:59added owner 0x9d359da6ca68ddbfb148d07b74aa3afd1f65039f
2023-12-05 22:02:11removed owner 0x9bb5ff8ac7400a6c1391b15c13403af30a207d7e
2023-12-05 21:58:59added owner 0xa724796812276dddf1dc65df91e11c1933a72371

The Base control Safe added 0x0241e41c3a6e1c7e322b976db22a3eb2adef4bcc and 0xec8f3975cf04eba38bcf51f3297667abc992c40e on October 6, 2026. It removed 0xc66476c083812ef4af9567ec4b0b737230b5754d and 0x760791319c9f6e12173be39e0f476e68194d54b8 on October 7. Arbitrum and BNB control Safes have the same current owner set as Ethereum; Linea and Sei still include 0x7607…54b8 and 0xc664…754d. All these measured Renzo control Safes have empty modules and no guard reported.

External control dependencies

External protocols have their own administrators. Their powers are dependency risks and do not establish the signer threshold of Renzo’s own administration.

The measured EigenLayer dependency path places StrategyManager, DelegationManager, EIGEN and related implementation upgrades behind admin 0x8b9566ada63b64d1e1dcf1418b43fd1433b72444. Its owner 0x369e6f597e22eab55ffb173c6d9cd234bd699111 is measured as a 1-of-2 Safe with empty modules and no guard reported. BackingEigen has another admin owned by another 1-of-2 Safe. These measurements are not evidence that Renzo owns either signer set.

Other measured proxy paths are summarized below. An unresolved contract holder has no assumed threshold or delay. Names are measured code identifiers, not proof of first-party attribution.

ChainProxy or proxy groupAdmin/beacon and controlling path
EthereumEzAuction (0x31e5b1750b872542af1848d62464e2936a387453), delay 604800 seconds 0xddff2552aefb3ce904d44757788edf25505da18e → 0x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7 → timelock 604800 seconds
EthereumStrategyManager (0x858646372cc42e1a627fce94aa7a7033e7cf075a), delay 0 seconds ; DelegationManager (0x39053d51b77dc0d36036fc1fcc8cb819df8ef37a), delay 0 seconds ; Eigen (0xec53bf9167f50cdeb3ae105f56099aaab9061f83), delay 0 seconds ; RewardsCoordinator (0x7750d328b314effa365a0402ccfd489b80b0adda), delay 0 seconds ; AllocationManager (0x948a420b8cc1d6bfd0b6087c2e7c344a2cd0bc39), delay 0 seconds ; StrategyFactory (0x5e4c39ad7a3e881585e383db9827eb4811f6f647), delay 0 seconds ; EigenStrategy (0xacb55c530acdb2849e6d4f36992cd8c9d50ed8f7), delay 0 seconds 0x8b9566ada63b64d1e1dcf1418b43fd1433b72444 → 0x369e6f597e22eab55ffb173c6d9cd234bd699111 → 1-of-2 Safe
EthereumOutbox (0x0b9857ae2d4a3dbe74ffe1d7df045bb7f96e4840), delay unknown ; Bridge (0x8315177ab297ba92a06054ce80a67ed4dbd7ed3a), delay unknown 0x554723262467f125ac9e1cdfa9ce15cc53822dbd → 0x3fffbadaf827559da092217e474760e2b2c3cedd → role-based controller; executor holders below
EthereumWETHPaymentReceiverEzETH (0x2b5b36597be376221a098361ceedaaf426c3fd01), delay 604800 seconds 0x783e71e3019ae02b127764100a270ac7bc940050 → 0x81f6e9914136da1a1d3b1efd14f7e0761c3d4cc7 → timelock 604800 seconds
EthereumBackingEigen (0x83e9115d334d248ce39a6f36144aeab5b3456e75), delay 0 seconds 0x3f5ab2d4418d38568705bfd6672630fcc3435cc9 → 0x942eaf324971440384e4ca0ffa39fc3bb369d67d → 1-of-2 Safe
EthereumStrategyBase (0xc9ad499c11677ea3815984295dd73c0858757b8e), delay 0 seconds 0x0ed6703c298d28ae0878d1b28e88ca87f9662fe9 → 0x369e6f597e22eab55ffb173c6d9cd234bd699111 → 1-of-2 Safe
EthereumConnextDiamond (0x8898b472c54c31894e3b9bb83cea802a5d0e63c6), delay unknown 0x4d50a469fc788a3c0cdc8fd67868877dcb246625 → 7-of-9 Safe
EthereumRollupAdminLogic (0x4dceb440657f21083db8add07665f8ddbe1dcfc0), delay unknown 0x3fffbadaf827559da092217e474760e2b2c3cedd → role-based controller; executor holders below
EthereumUpgradeExecutor (0x3fffbadaf827559da092217e474760e2b2c3cedd), delay unknown ; 0xe6841d92b0c345144506576ec13ecf5103ac7f49 (0xe6841d92b0c345144506576ec13ecf5103ac7f49), delay unknown 0x5613af0474eb9c528a34701a5b1662e3c8fa0678 → 0x3fffbadaf827559da092217e474760e2b2c3cedd → role-based controller; executor holders below
BaseL2ERC20TokenBridge (0xac9d11cd4d7ef6e54f14643a393f68ca014287ab), delay unknown ; ERC20Bridged (0xc1cba3fcea344f92d9239c08c0568f6f2f0ee452), delay unknown 0x0e37599436974a25ddeedf795c848d30af46eacf → authority unresolved
BaseL2StandardBridge (0x4200000000000000000000000000000000000010), delay 0 seconds ; L2CrossDomainMessenger (0x4200000000000000000000000000000000000007), delay 0 seconds ; 0x4200000000000000000000000000000000000018 (0x4200000000000000000000000000000000000018), delay 0 seconds 0x4200000000000000000000000000000000000018 → 0x8cc51c3008b3f03fe483b28b8db90e19cf076a6d → single key
BaseRecovery (0x3154cf16ccdb4c6d922629664174b904d80f2c35), delay unknown unresolved (type unresolved)
Arbitrum OneL2GatewayRouter (0x5288c571fd7ad117bea99bf60fe0846c4e84f933), delay unknown ; L2ERC20Gateway (0x09e9222e96e7b4ae2a407b98d48e330053351eee), delay unknown ; aeWETH (0x82af49447d8a07e3bd95bd0d56f35241523fbab1), delay unknown 0xd570ace65c43af47101fc6250fd6fc63d1c22a86 → 0xcf57572261c7c2bcf21ffd220ea7d1a27d40a827 → role-based controller; executor holders below
Arbitrum OneERC20Bridged (0x5979d7b546e38e414f7e9822514be443a4800529), delay unknown ; L2ERC20TokenGateway (0x07d4692291b9e30e326fd31706f686f83f331b82), delay unknown 0x1dca41859cd23b526cbe74da8f48ac96e14b1a29 → authority unresolved
Arbitrum OneUpgradeExecutor (0xcf57572261c7c2bcf21ffd220ea7d1a27d40a827), delay unknown 0xdb216562328215e010f819b5abe947bad4ca961e → 0xcf57572261c7c2bcf21ffd220ea7d1a27d40a827 → role-based controller; executor holders below
BNB Smart ChainConnextDiamond (0xcd401c10afa37d641d2f594852da94c700e4f2ce), delay unknown 0x9435ba7c661a0fd477deed640491de8c100325a7 → 7-of-9 Safe

The role-based external controllers above have the following measured executor holders. The table does not replace unknown fastest-delay values in their privileged-function records.

External controllerExecutor-role holders
Ethereum 0x3fffbadaf827559da092217e474760e2b2c3cedd0xe6841d92b0c345144506576ec13ecf5103ac7f49, timelock, 259200 seconds; schedule still applies to open execution; 0xf06e95ef589d9c38af242a8aaee8375f14023f85, 9-of-12 Safe; module path described below; 0x30668ee252a28986ee6dc21b26e4018ac0e339b9, contract; scope unresolved
Arbitrum One 0xcf57572261c7c2bcf21ffd220ea7d1a27d40a8270xf7951d92b0c345144506576ec13ecf5103ac905a, single key; 0x423552c0f05baccac5bfa91c6dcf1dc53a0a1641, 9-of-12 Safe; module path described below

Three external Safe paths include modules and therefore are not limited to their signature thresholds:

  • Ethereum Safe 0xf06e95ef589d9c38af242a8aaee8375f14023f85 is 9-of-12, with module 0x3fffbadaf827559da092217e474760e2b2c3cedd (UpgradeExecutor; no owner read).
  • Arbitrum Safe 0x423552c0f05baccac5bfa91c6dcf1dc53a0a1641 is 9-of-12, with module 0xcf57572261c7c2bcf21ffd220ea7d1a27d40a827 (UpgradeExecutor; no owner read).
  • Base external Safe 0xf0db7318a51a21c413cadd4abdc1e8a500fe5b1b is 4-of-9, with module 0xf3c72d97a5dcf0449e89bbce1a0581d8d15c0237; its type and owner are unknown in these reads.

No standard proxy slot is set for REZ in these measurements; that alone is not proof of immutable logic. The measured ConnextDiamond contracts expose facet lists and upgrade through diamondCut; its guard was not read. An immutable core cannot be established.

Parameter Controls

Administrators can alter withdrawal behavior, delegation and fees, and the available bounds differ across products.

The main Safe owns both ezREZ and ezEIGEN and can change their cooldown, deposit limit, delegation and rewards destination immediately. Both vaults read cooldownBlocks = 100800, vaultCooldownBlocks = 50400, and maxCooldownBlocksBasisPoints = 5000. These are raw block/configuration values, not a verified seconds-based maximum exit time.

Each vault reads raw vaultFee = 900, protocolFee = 100, with deployed BASIS_POINTS = 10000: 9% and 1% respectively. Each associated withdrawal buffer reads raw instantWithdrawalFeeBps = 1 and deployed BASIS_POINTS = 10000, or 0.01%. No fee cap is established by these reads. Buffer configuration is an immediate owner power.

For ezETH, DepositQueue reads raw feeBasisPoints = 1500 and exposes setFeeConfig(address,uint256) under onlyRestakeManagerAdmin. The function-path measurement does not resolve holders or fastest delay. RoleManager separately assigns RESTAKE_MANAGER_ADMIN to the main Safe and timelock; that authority evidence remains relevant despite the parser limitation.

Official ezETH docs state a 10% restaking-reward fee, split 50/50 between reserves and node operators. The raw deployed read is 1500; its denominator was not verified. A supplementary BASIS_POINTS() read at Ethereum block 26,345,115 was refused by the archive provider. The documentation discrepancy remains unresolved; 1500 is not converted to a percentage and is not a cap.

The native bridge deposit contracts on Base, Arbitrum and Linea read raw bridgeFeeShare = 5 and deployed FEE_BASIS = 10000, or 0.05%. BNB’s older xRenzoDeposit reads raw bridgeFeeShare = 0 and bridgeRouterFeeBps = 5, with deployed FEE_BASIS = 10000: zero and 0.05%, respectively. Linea also reads raw flatBridgeFee = 100000000000000; its unit remains unverified here. These are current values, not maximum fees.

updateBridgeFeeShare and token-time-discount changes on the newer native bridges are guarded by onlyOwner, pointing to each chain’s EzAdminL2; their fastest delays remain unresolved. The docs expressly allow Renzo to adjust discount/variable fees.

Assessment: cooldown changes can delay an exit, reward-destination changes can redirect subsequent rewards, and delegation changes alter operational exposure. The seven-day upgrade delay does not prevent these owner-controlled changes. Caps not established above remain unknown, rather than proven absent.

Decentralization Assessment

Renzo has community participation, but its measured execution layer remains concentrated in a small number of signer groups.

The token allocation does not prove current voting control. Official tokenomics state a total supply of 10,000,000,000 REZ and initial circulating supply of 1,150,000,000. The disclosed allocations are:

AllocationShare
Fundraising31.56%
Community32%
Core contributors20%
Foundation12.44%
Binance Launch Pool2.5%
Liquidity1.5%

The document specifies investor and contributor vesting, but no current entity-level holdings or delegated-vote concentration are measured. Separate allocations cannot be combined into one controlling voter. Whether governance power matches user stake is therefore unconfirmed.

The FAQ plans to move AVS selection from Snapshot to full on-chain governance once processes are defined, tested and proven. It gives no completion date.

Sigma Prime’s June 2024 restaking review described sensitive operator and admin powers, then recorded centralization as resolved through a governance protocol. The team’s stated sequence was moving default administration to a timelock and later a DAO. Current RoleManager confirms the timelock holds default administration; separate direct Safe powers remain. This audit resolution does not establish complete decentralization.

The current native-restaking role lists three single keys: 0x64e968003c934f8d7ad9a4e30f48ee8e2409bae6, 0x3b8c27038848592a51384334d8090dd869a816cb, and 0x3f773dc3ccc70b3d2a549713ac8d556af949d4e8. Their precise current OperatorDelegator powers are not resolved because that contract’s deployed address is missing. The historical audit description is not treated as proof of a current arbitrary-fund-transfer path.

Assessment: key-person risk is material because signer independence is undisclosed and single-key operational roles exist. The observed core Safe needs three signatures; the evidence does not prove that one person controls those signatures or can alone drain core funds.

Governance Risks

The strongest demonstrated risk is immediate administrative change outside the main timelock; the strongest demonstrated safeguard is delayed replacement of principal Ethereum implementations.

  • Signer compromise or collusion: compromising the main Safe can enable an immediate middleware upgrade and the documented direct vault actions. Assessment: this is a low-to-moderate likelihood, high-impact scenario. Evidence of signer independence and custody arrangements could change that likelihood band.
  • Exit-window mismatch: seven days may expire before a normal withdrawal finishes. A delayed upgrade therefore does not guarantee a user can fully exit first.
  • Single-key intervention: pzETH withdrawal pausers can stop withdrawals immediately. This supports emergency containment but creates an availability dependency.
  • Voting capture, bribery and apathy: linear delegated voting permits concentration in principle. Actual concentration and turnout remain unknown, and no cited passed attack or proposal against users establishes a current governance-attack condition.
  • Control changes over time: the October 2026 threshold reduction shows that safeguards can change. It is a documented configuration change, not evidence of malicious intent.
  • Operational error: sensitive delegation and withdrawal controls can harm users even without key theft. Sigma Prime explicitly warned that trusted relationships can cause significant harm through malicious or accidental use.

Assessment: centralized response powers are useful during slashing, accounting or bridge incidents. Their benefit depends on prompt, correct action by the same controllers whose compromise or mistake creates the risk.

Data Completeness

Measured Ethereum and five active non-Ethereum control graphs support a firm conclusion about concentrated execution authority, but governance participation and full product coverage remain incomplete.

The assessed evidence covers ezETH and its measured bridge deployments, REZ governance, and the documented shared Safe/timelock exposure of ezREZ and ezEIGEN. The separate pzETH controller is reported separately. Renzo’s broader Reserve Vaults and Enterprise products are described in the docs, but their deployed control paths are not fully bound to addresses. Their evidence cannot be replaced by ezETH’s controls. Managed-vault documentation explicitly allows configuration and integration changes without prior notice.

The Staking Suite spans Ethereum and Solana. Solana product controls, including ezSOL and ezJTO, remain unmeasured; the Ethereum multisig threshold and delay are not transferred to them.

Blast was deployed earlier and has no current TVL in the supplied snapshot; its deployment is not assessed. Its partial measurements also lack upgrade history for seven listed contracts. Current controls on Solana, Mode, Zircuit, Fraxtal, Monad and Optimism were not measured in this control file; no controls are inferred for them.

Important limitations are:

  • No measured Snapshot venue, top voters, concentration shares or last-five-proposal result set; no demonstrated rule binding voting to execution.
  • Signer addresses are available for measured Safes, but personal identities and independence are not.
  • OperatorDelegator, managed-vault strategies and several audit-named components have no resolved current address. DepositQueue and RoleManager were discovered through RestakeManager calls despite their omission from the original address table.
  • Some privileged-function guards and delays remain unresolved. Missing slots do not prove immutability, and missing histories do not prove no changes occurred.
  • DepositQueue’s deployed fee denominator and cap remain unverified after the archive provider refused the supplementary call. Numeric uncertainty does not remove independently evidenced admin authority.
  • Sigma Prime’s cited restaking review comes from a verified auditor-controlled publication. Other audit copies include unverified publication provenance; they are not treated as independently confirmed controls or as a protocol penalty merely for that provenance.

Insufficient data for this assessment of unmeasured deployments, signer independence and binding vote execution. None of these gaps proves that a safeguard is absent.

Score: 3.7/10

Rubric security-score-v2.6; data completeness 0.75

Provisional — data incomplete: Admin control: the protocol's current deployed contract addresses. This mark does not change the score.

Governance factors, points out of 10 (points)
Admin control
5.50 points
Change delay
1.00 points
Decision process
3.50 points
FactorLevelEvidenceWeightPoints
Admin controlL2supported (provisional)0.455.50
Change delayL0supported0.301.00
Decision process—not disclosed0.253.50

370 citations · click markers to view evidence

🔗

Cross-Analysis

Cross-Risk Analysis: Renzo

Renzo’s clearest combined risk is a rush to exit: delayed redemption and limited measured trading liquidity can turn a discount into liquidation losses. A second risk is that upgrade notices can expire before holders finish withdrawing. Shared administrators and external restaking systems can spread disruption across products, although the evidence does not establish a current exploit.

The assessed total value locked (TVL) scope comprises ezETH, its cross-chain representations, pzETH, ezSOL, bzSOL and ezJTO. REZ matters through governance and incentives. Related ezREZ and ezEIGEN are included for shared upgrade-control exposure. Reserve Vaults, ezBTC and Opolis bonds remain other-product context where shared exposure is not established; their balances are not added to protocol TVL.

Enterprise is another documented product whose deployed control paths are not fully established; no Enterprise balance is attributed to this TVL. TVL measures deposited asset value. The captured adapter counts capital also used elsewhere in decentralized finance (DeFi), so the total is not unique system-wide capital.

Top Risks for a Holder

The three risks below rank the strongest evidenced holder-loss channels, with liquidity stress first because it has a documented precedent. Likelihood bands are qualitative assessments, not annual probabilities.

  1. A discounted exit becomes a liquidation loss — ezETH, cross-chain ezETH and pzETH; REZ/ezREZ face related confidence and token-price exposure. Selling pressure can outrun redemption and trading liquidity; users borrowing against receipts can lose collateral before backing becomes withdrawable. The April 2024 ezETH depeg caused downstream liquidations and bad debt, rather than establishing a Renzo contract exploit.

    Magnitude: November 5 Ethereum-to-WETH probes reported 15.85% impact for a $100,000 canonical ezETH trade and 97.87% for pzETH. These are route-specific quotes, not executed losses or general loss forecasts. They do not establish whether smaller trades can execute. Native exits are documented as up to 15 days for ezETH and 14–21 days for pzETH; forced sellers can realize a discount, and leveraged users can lose their posted equity. REZ price declines also reduce ezREZ’s dollar backing; core receipts are not documented as REZ-backed.

    Likelihood: medium. Holders can trigger an exit wave without privileged access; queues slow arbitrage, and the historical event demonstrates the transmission channel. Current leverage, funded buffers and executable liquidity below $100,000 would change this band.

  2. Administrative compromise or error outlasts the exit window — ezETH, pzETH, cross-chain ezETH, and shared-control ezREZ/ezEIGEN. Malicious upgrades or harmful settings can change withdrawal behavior while holders are still waiting to exit. The core seven-day upgrade delay does not cover every administrative path, and pzETH has separate immediate upgrade authority.

    Magnitude: a malicious implementation can put the affected product’s principal at risk; a pause can freeze exits, with no maximum suspension duration established. The $102.09 million November 3 TVL is the assessed footprint, not a measured amount drainable through any one function. pzETH’s November 5 market capitalization was $3.04 million, a valuation rather than guaranteed recoverable backing.

    Likelihood: low, with limited confidence. Multisig protection across every privileged path is unconfirmed; legitimate mistakes can also cause disruption. Ethereum measurements at block 26,345,115 show a main 3-of-5 Safe, a wallet requiring three of five signatures. A separate pzETH upgrade Safe requires five of eight signatures, but pzETH withdrawal pausing requires only a single authorized key. Signer independence, custody arrangements and independent monitoring remain unknown.

  3. A backing-system, slashing or settlement failure reduces principal — ezETH and bridged ezETH, pzETH, ezSOL, bzSOL and ezJTO. Each receipt inherits failures in its own underlying staking, restaking and custody chain. Cross-chain ezETH additionally depends on asset settlement and rate delivery; separate token symbols do not create independent backing security.

    Magnitude: permanent loss of backing or extended redemption interruption in the affected allocation. Linea represents $16.2477 million, or 15.9155% of captured TVL; Solana represents $3.3106 million, or 3.2429%. These locate exposure, not a forecast that those entire balances will be lost. Product allocations and loss caps do not support a reliable protocol-wide dollar-loss estimate.

    Likelihood: medium. Operators, external administrators or infrastructure failures can trigger disruption across several required layers. Loss severity depends on allocations and isolation caps; complete current measurements of these controls would change the band. Current Solana minting is discontinued, but withdrawal exposure remains.

Combined Attack Vectors

The strongest combined vectors connect market exits, administrative timing and dependencies; none below asserts a demonstrated live exploit.

Vector 1: Incentive disappointment starts an exit-and-liquidation loop

  1. A reward change or confidence shock prompts receipt holders to sell. Historical reporting links the April 2024 allocation announcement to selling; current incentives remain adjustable.
  2. Prices on decentralized exchanges (DEXs) fall as selling exceeds available liquidity. Arbitrageurs cannot instantly turn every discounted receipt into backing because native redemption takes days. The measured routes and product exit terms are detailed in holder risk 1.
  3. External borrowers face liquidations; additional sales can deepen the discount. An unleveraged holder can wait, while a borrower may lose equity before redemption completes. The historical ezETH event establishes this channel, but not today’s leverage distribution.

Domains: incentives and token economics + market liquidity + social confidence + downstream lending. Likelihood: medium: no attacker permission is required to sell; actual borrower exposure is missing. Potential impact: severe: discounted exits, liquidation losses and bad debt in exposed lending markets. A confidence shock could also hurt REZ and ezREZ prices, without making REZ collateral for ezETH.

Vector 2: An upgrade notice arrives too late for an orderly exit

  1. Compromised controllers schedule a harmful core or shared-vault upgrade. Principal Ethereum implementations wait seven days; ezREZ and ezEIGEN share the upgrade timelock.
  2. Holders request redemption, but ezETH can require 10–15 days without a funded buffer; ezREZ and ezEIGEN document up to 14 days. A seven-day notice therefore cannot assure completed withdrawal before execution.
  3. Holders seeking faster escape meet the measured liquidity constraint. Immediate middleware upgrades and direct vault settings add separate intervention paths; this does not establish unlimited immediate Safe minting or a demonstrated core-drain function.

Domains: governance execution + redemption liquidity + shared contract controls. Likelihood: low, with limited confidence: the malicious case requires privileged compromise or collusion; independent monitoring and cancellation protection are unconfirmed. Potential impact: catastrophic: a harmful replacement could impair the affected product’s principal and exits. Signer custody, independent cancellation arrangements and available exit liquidity would change the likelihood. pzETH is a separate, faster variant because its 5-of-8 upgrade path has no delay.

Vector 3: Rate delivery and asset settlement fail together

  1. An outage or compromised rate-delivery path affects cross-chain minting while deposited assets await settlement. Nethermind’s May 1, 2025 review describes immediate xezETH issuance, a cross-chain ezETH representation, followed by batched transfers to Ethereum.
  2. If current freshness and deviation controls permit an inappropriate rate, issuance can exceed backing. If they reject it, minting can stop while settlement or pricing recovers. Current path-by-path checks are incomplete; a stale-rate exploit is not established.
  3. Recovery depends on chain-specific bridges, receivers and controllers. Governance reports immediate replacement of local EzAdminL2 wrappers, but leaves the underlying proxy path’s fastest delay unknown. The bridge review’s architecture cannot prove every current deployment has identical controls.

Domains: oracle delivery + optimistic issuance + bridging + chain-specific administration. Likelihood: low: the loss-producing case needs a rate or settlement failure that survives validation; sender validation is described in the reviewed architecture. Current verifier thresholds, settlement exposure and per-chain rate checks could raise or lower the band. Potential impact: severe: unavailable exits or excess cross-chain claims; exposure beyond the affected route depends on unmeasured isolation and issuance caps.

Vector 4: An upstream administrator defeats the protection expected from Renzo’s timelock

  1. An external administrator is compromised or makes a harmful implementation change. At Ethereum block 26,345,115, the governance report measures a 1-of-2 Safe controlling the EigenLayer dependency path, separately from Renzo’s main Safe.
  2. The affected external contracts alter the behavior of assets already deployed through Renzo. The documented ezETH route includes Ethereum validators, EigenPods, EigenLayer operators and services. Related ezREZ/ezEIGEN also use EigenLayer, but hold distinct collateral.
  3. Renzo’s delayed upgrades cannot undo a change in an external system. Users remain dependent on withdrawal and recovery there; broader multi-firm ezETH audits do not establish review of all dependency code.

Domains: external governance + strategy dependencies + redemption timing + audit-scope boundaries. Likelihood: low: privileged compromise or serious administrative error is required; current custody and external exposure caps are missing. Potential impact: catastrophic: loss or immobilization across affected EigenLayer allocations. This is a common-dependency scenario, not a claim that Renzo controls the external signers.

Systemic Risk Factors

Product and chain diversity leave substantial common dependencies; the supplied dependency graph does not measure their underlying protocol weights.

The graph records unresolved EZETH at 96.12% and PZETH at 3.87%. Its TVL input is November 3, 2026, and composition input November 4. No underlying protocol score, score-validity deadline or oracle-provider node is supplied. The empty warnings list contains no circular-backing or downstream-holder warnings; this does not prove either is economically absent. These receipt-token entries are not measured weights in EigenLayer, Lido or Symbiotic and do not establish circular backing.

The following routes require reassessment within economic dependency risk. Yield and leverage effects remain separate from borrower liquidation and bad-debt effects.

PositionLayers carrying the positionCross-domain implication
ezETHRenzo → Ethereum staking/EigenPods → EigenLayer operators and services; stETH adds Lido.External code, operators and administrators can affect backing despite Renzo’s own upgrade delay.
pzETHLido stETH/wstETH → Mellow vaults → Symbiotic collateral and operators/services.Vault capacity and withdrawal availability combine with external slashing exposure. The official December 19, 2025 Cap underwriting announcement adds a possible borrowing exposure whose current allocation is unknown.
ezSOL, bzSOL, ezJTOJitoSOL, BNSOL or JTO respectively → Jito restaking vaults → Node Consensus Networks, the services they secure.Remaining holders inherit underlying-asset, program and exit risks. Minting is discontinued; Jupiter conversion is historical deposit context.
Cross-chain ezETHRate inputs → messaging → local issuance → canonical asset bridges → Ethereum backing.Price and settlement can fail at different times; current universal routing and isolation limits are not established.
Shared-control ezREZ/ezEIGENProduct-specific receipt vault → EigenLayer strategy → delegated operator/services.Common upgrade control and dependency exposure coexist with separate collateral.

Other-product context also needs complete routing before any exposure is attributed to this record. ezUSDC1 uses Aave Core, Morpho, Aave Horizon and Spark. ezUSCC1 uses Superstate USCC and Aave Horizon; ezCompETH1 loops ezETH collateral and WETH borrowing through Compound. Their leverage affects their own holders; an unmeasured downstream position does not establish a core ezETH backing liability.

  • Chain and bridge interruption: Ethereum holds 72.48% of captured TVL; Linea holds 15.92% and is a stage-0 rollup. An L2 sequencer or bridge interruption can constrain transfer and exit even when Ethereum backing remains intact. The raw exit-window value does not establish a delay guarantee. Blast was deployed earlier and has no current TVL; that deployment is not assessed.
  • Correlated market stress: ETH/Lido exposure can connect ezETH and pzETH outcomes despite different restaking systems. Falling liquidity across DeFi can simultaneously weaken discounted-token arbitrage and collateral liquidation. The common Lido paths are evidenced above; this correlation is an assessment, not a measured loss correlation.
  • Stablecoin and fund stress: other-product ezUSDC1 and ezUSCC1 inherit stablecoin and lending/fund risks; ezUSCC1 also depends on timely fund redemption. No such reserve-vault exposure is assigned to core TVL without a demonstrated shared position.

Risk Amplifiers

The main amplifiers are exit timing, shared control and uncertainty about which historical protections apply to current versions.

  • Notice shorter than withdrawal: a seven-day upgrade wait can coexist with a 10–15-day exit; market selling becomes the remaining fast option. This couples governance risk directly to slippage, as vector 2 explains.
  • Shared signers across chains: governance reports the same current owner set on Ethereum, Arbitrum and BNB. Compromising overlapping signers can correlate failures across local administrations, although each chain’s state and permissions remain distinct.
  • Operational trust beside code controls: Sigma Prime’s June 2024 high-severity validator-frontrunning finding was acknowledged; the team relied on permissioned operators with commercial agreements. Current deployment exposure remains unconfirmed. Combined with undisclosed signer independence, this concentrates reliance on correct human operations; it is not proof of a current validator-theft path.
  • Review does not guarantee current-version behavior: the published comparison reports 83.1% of lines unknown and 14.4% changed, with two components showing measured differences. Most code was not compared because audit commits were unavailable; this is neither a measured all-code mismatch nor evidence of a new vulnerability.
  • Historical bridge-test limits: Nethermind’s May 2025 bridge review rated the test suite Low. Tests detected 129 of 430 deliberate code changes (30.00%); xRenzoDepositNativeBridge tests detected 10 of 267 (3.75%). These historical results limit assurance that tests would catch regressions; they do not demonstrate a current defect.
  • Losses lack a measured backstop: the economic report establishes neither a dedicated loss-cover fund nor completed comprehensive compensation. Queue liquidity and the other-product Opolis reinsurance bonds do not supply depositor insurance.

Risk Mitigators

The strongest mitigators protect specific paths and products; none guarantees prompt redemption or recovery from losses.

  • Independent ezETH review: Sigma Prime’s June 2024 scope covers ezETH restaking contracts, and Nethermind’s April 14, 2025 review covers core contracts and EigenLayer slashing integration. Nethermind’s May 1, 2025 review separately covers bridge smart contracts. These scopes support ezETH review depth, not pzETH or Solana coverage. Auditor-controlled publication is established for these reviews; project-hosted audit copies retain their separate provenance caveat without an automatic risk penalty.
  • Multisigs and delayed principal upgrades: measured Ethereum core upgrades wait seven days. The main Safe requires three signatures; pzETH’s separate upgrade Safe requires five. These constrain single-key replacement of core logic, while immediate ancillary paths remain. The measured block and thresholds are given in holder risk 2.
  • Native redemption and emergency controls: queues provide an alternative to selling at a discount. At Ethereum block 26,345,115, measured ezETH middleware withdrawal flags and pzETH’s withdrawal-pause flag were false. This is a point-in-time state, not a guaranteed emergency exit.
  • Rate-message validation: the reviewed bridge architecture validates senders before updating the mint rate. It reduces one forgery channel, while current per-path freshness and fallback guarantees remain incomplete.
  • Accountability and external scrutiny: named leadership, a foundation, public addresses and audit links offer accountability. A documented $250,000 Immunefi bounty adds a reporting incentive, but current payout terms and asset scope remain unconfirmed; it is not insurance.

Contradictions & Inconsistencies

Several differences reflect source age or scope, while reward, timing and deployment questions remain unresolved.

  • Broad product catalog versus measured protocol scope: the overview lists Reserve Vaults, ezBTC and Opolis bonds. The economic report’s captured adapter excludes them and separately establishes shared-control relevance for ezREZ/ezEIGEN. The narrower measured perimeter governs exposure totals here; catalog inclusion is not evidence that a product backs protocol TVL.
  • Solana deposits versus current discontinuation: overview deposit descriptions are historical mechanics. November 5 product captures discontinue ezSOL, bzSOL and ezJTO minting and retain withdrawal instructions. This report assesses remaining holders rather than assuming new deposits remain open.
  • pzETH service readiness: the economic report cites a FAQ describing services as not yet functional; the overview also records an official December 19, 2025 announcement of live Cap underwriting. Publication of that announcement rules out a blanket assertion that no such deployment was announced. Current allocation and realized underwriting income remain unconfirmed.
  • Withdrawal timing and upgrade protection: seven-day buffered and 10–15-day unbuffered estimates coexist with a stated 14-day EigenLayer minimum. These are not reconciled into a guaranteed end-to-end bound. The old three-day, all-upgrades timelock description is superseded for measured Ethereum principal paths by seven days, with separate immediate paths.
  • Reward-program milestones: economic security calls the March 31, 2025 snapshot the ezPoints program’s end; governance states it concluded April 28, 2025. Primary documentation establishes frozen balances at the March 31 snapshot and separately states an April 28 conclusion. The reason for the interval is not explained; neither date establishes current emissions.
  • Rewards and deployed fee units: ezETH documentation claims both complete EigenLayer reward pass-through and a 10% reward fee. The Ethereum block 26,345,115 raw queue setting is 1500, with its denominator unverified. The conflict remains open; 1500 is not asserted as 15%, and the uncertainty does not remove administrator authority.
  • Incident publication versus confirmation: a published April 24, 2024 account exists and reports roughly $56 million of liquidations. Its event claims remain unconfirmed under an unverified-publisher classification (source reliability: unverified publisher). The social report’s missing first-party post-mortem is a narrower gap, not absence of every published account. The stronger historical risk assessment independently supports depeg, liquidations and bad debt; completed comprehensive reimbursement and net uncovered losses remain unknown.
  • Audit fixes versus deployment uncertainty: the code ledger marks all seven critical groups fixed. Historical audit summaries and incomplete deployed-version comparisons do not turn them into current open critical vulnerabilities. Supplemental Code4rena findings are separately unconfirmed and undeduplicated; they cannot silently replace ledger counts.
  • Governance wording versus execution evidence: documentation’s on-chain wording conflicts with Snapshot voting. Neither establishes automatic execution or proves signers can disregard votes.

Overall Risk Profile

Assessment: Renzo’s most likely damaging failure mode is a liquidity discount becoming a realized loss for forced sellers or leveraged users. Its worst case is a privileged or upstream implementation failure that destroys backing or immobilizes multiple affected allocations. Compared with liquid-restaking peers, Renzo has concrete multi-firm ezETH review and a measured principal upgrade delay, but product variety does not eliminate common administrators or underlying systems. No like-for-like peer measurements support a numerical safety ranking. Holders should assess each receipt’s collateral, exit route and control path separately; neither REZ incentives nor a headline audit count demonstrates dependable stress liquidity.

Data Gaps

  • Some source material could not be collected; the report may be incomplete.
  • Some source material is unavailable because it was not found publicly.
  • Some source material is unavailable because the publisher restricts access.
  • Current deployment addresses could not be confirmed for some audited contracts.
  • Source summaries cover only the first 80000 characters of 4 document(s); later content was not assessed.

Missing current exposure and control details limit loss sizing more than identification of the principal combined risks.

  • Loss sizing: complete backing versus liabilities, in-flight settlement balances, product allocations, operator/service concentration and dependency isolation caps. The graph has no underlying protocol scores or validity deadlines; receipt composition is not a solvency reconciliation.
  • Exit-and-liquidation loop: live buffer balances, executable smaller trades, Solana exit depth and current venue-specific leverage/liquidation rules. These would determine how quickly a discount becomes a realized loss.
  • Control compromise: signer identities, independence and custody; unresolved privileged-function delays; unmeasured Solana controls and several chain deployments. A measured Safe threshold does not answer these questions.
  • Current applicability of old findings: later audited revisions were not fully compared with deployments. Validator-key, accounting and checkpoint concerns remain historical or repository-bounded unless a current match is established. This report adds no new code vulnerability claim.
  • Program and incident outcomes: paid-yield incentive share, current Cap allocation, completed depeg compensation and a first-party post-mortem. The already published third-party account remains available with its stated verification caveat. Related-product Opolis settlement also remains unconfirmed; it does not establish a loss-cover resource for these receipts.

130 citations · click markers to view evidence

Preview3 · Default / Rows