Aura
Tier 4Security Analysis Report
TVL
$9.66M
Sections
7
report sections
Citations
301
evidence references
Version
v6
10/28/2026
Moderate Risk
Safety Score · out of 10
Executive Summary
Executive Summary: Aura
Aura is a closing protocol. Its pools accept withdrawals only, and AURA and auraBAL holders redeem in stages through 2027-05-10. The main risks for a holder are getting value out, not earning yield. No exploit of Aura is recorded, but no audit of the redemption path is in the sources, deployed code was never compared with any audited version, and the people running the admin keys are not named. Confidence is moderate.
Safety Rating
Overall Score: 5.7/10 Dimension scores: Code security 5.9 · Economic security 4.5 · Team transparency 6.2 · Governance 3.7 · Track record 8.6
- Code security
- 5.9 points
- Economic security
- 4.5 points
- Team transparency
- 6.2 points
- Governance
- 3.7 points
- Track record
- 8.6 points
Aura is winding down, so the rating reflects exit risk more than yield risk. Two score caps apply in principle: one because audit scope could not be linked to the deployed contracts, and one because core economic facts are unknown. Neither cap is binding on the displayed scores. The scores come from the worker reports and are not changed here.
Scores by Dimension
| Dimension | Key Finding |
|---|---|
| Code Security | Both high findings are fixed and 21 of 28 medium findings are fixed (two of the 28 are medium (estimated): an AI estimate, because that report states no rating), but audits were not linked to the deployed code. |
| Economic Security | Aura is in a governance-approved wind-down. auraBAL holders wait until 2027-05-10, and evidence on liquidity, oracles and loss cover is thin. |
| Team Transparency | A legal entity is named for the website only, and no individual team member is named. |
| Governance | A 4-of-7 Safe (multisig wallet) holds owner powers with a measured delay of zero, and the link between Snapshot votes and the Safe is unconfirmed. |
| Track Record | Aura ran from mid-2022 with no recorded exploit of its own contracts, but no complete outage or pause history exists. |
What This Protocol Does
Aura let holders of Balancer liquidity-provider tokens or BAL tokens hand them to Aura and earn boosted rewards without managing Balancer's long veBAL lock themselves. It is now ending: the official site says pools are withdrawal-only. DeFiLlama shows about USD 9.7M of remaining deposits, all on Ethereum.
Documented vs On-Chain
- The official sunset page says the final stage is "in 6 months". That wording is stale against the 2026-10-28 analysis date.
- An undated upgrade review by an individual reviewer (collected 2026-10-23) describes
BoosterOwnerSecondaryas proposed and not executed. On Ethereum at block 26289853 (2026-10-28), the owner chain shows it in place, and the booster owner contract is sealed. - A curated snapshot (2026-03-09) rated the product "Good" with "No death spiral concerns". The wind-down proposal (source reliability: unverified publisher) describes the opposite, and the economic report treats the snapshot as historical.
- The overview says it is not confirmed that the
AuraProxyOFTsudo address can callrescue(). The governance report reads an 8-of-13 Sudo Safe callingrescueon both bridge contracts, so the gap is partly closed. - The official sunset page says pools are in withdrawal-only mode. On Ethereum at block 26289853 (2026-10-28), the
BoosterreportsisShutdown: truewith 283 pools, andAuraLockeris also shut down. This agrees with the page.
Key Strengths
- Both high-severity audit findings for Aura Finance are fixed, and 21 of 28 medium findings are fixed. Two of those mediums are estimated mediums: the report states no rating, so this is an AI estimate.
- Review volume is large: three Halborn engagements, one Zellic assessment with no findings, and a public Code4rena contest (source reliability: unverified publisher). Several of these reports, including the fixed-finding counts above, are hosted on publications not verified as auditor-controlled. That lowers source confidence but not coverage credit.
- The Ethereum
Booster,AuraLockerand measured sidechain boosters report shutdown, which limits new actions.BoosterOwneris sealed with a 30-day forced-action delay that has not started. - The official site promises that LP, vault and locker positions stay withdrawable at any time.
- No exploit of Aura is recorded in the hacks dataset, and a 4-of-7 Safe can pause the bridge proxies at once.
Key Risks
- auraBAL is locked until 2027-05-10 and tracks Balancer's market. Likelihood: high that holders wait until Stage 3, medium that BAL price moves hurt them materially. Magnitude: the USD 9.7M TVL is context, not a loss ceiling, and the auraBAL share is not measured. Holders carry the BAL/WETH price move over the roughly six months to Stage 3, plus any early-exit cost. A loss of principal is not ruled out, because final proceeds are not compared with each holder's original contribution.
- LPs exit into Balancer pools, and a Balancer change dated 2026-10-30 is two days away. The date comes from a community-publication forum proposal, and the vote result is unconfirmed. Likelihood: medium for a worse exit or tooling problem, low for permanent loss. Magnitude: up to the Ethereum-held TVL for LPs who delay, though Balancer says withdrawals stay open.
- The redemption and admin path has no audit in the sources and no named operators. Likelihood: low to medium for a contract or key failure, and medium for a user losing funds to a fake site or a mistaken claim. Magnitude: the claims exposed are treasury assets for AURA redeemers and, for auraBAL, 90% of the BPT position. The treasury and BPT position sizes are not measured, so no dollar ceiling is given.
- Shared admin keys with no delay. One address sits on four Safes and owns the contract that owns each of the eight Ethereum-side bridge receivers. Compromise is low likelihood and moderate impact.
- Open audit findings remain. Seven medium findings are not fixed: 3 disputed, 3 acknowledged and 1 with no stated status. Two disputed voting-power findings (Code4rena M-03 and M-08) describe a pattern still present in the latest repository code.
Critical Findings
No critical findings identified. The ledger lists two high findings for Aura Finance, both fixed. A high finding for AuraSoulsV1 (a different product) is open and is excluded from Aura Finance counts.
Recommendation
Aura suits risk-tolerant holders who need to exit and can accept Balancer price risk until 2027-05-10. Holders who can leave now through LP, vault or locker withdrawal should consider doing so before Balancer's reported 2026-10-30 change. Use only the official site, because lookalike sites describe redemption limits. The worst case is a compromise of shared admin keys or a fault in redemption contracts that have no audit in the sources. New deposits make no sense for a closing protocol.
Report Sections
- Protocol Overview
- Code Security
- Economic Security
- Social & Team
- Governance
- Cross-Risk Analysis
31 citations · click markers to view evidence
Protocol Overview
Protocol Overview: Aura
Aura was a Balancer-based yield booster (it pooled users' Balancer LP tokens and BAL tokens to earn boosted rewards). It is now winding down. The official site says the Aura Protocol will terminate operations, pools are withdrawal-only, and AURA and auraBAL holders redeem in stages through 10 May 2027 . For a user, the question is no longer how to earn yield. It is how to withdraw, and how to redeem AURA and auraBAL for treasury assets. The evidence is strong on the sunset and the on-chain shutdown flags. The details of the governance vote are not established.
What It Does
Aura let people who hold Balancer liquidity-provider (LP) tokens, or BAL tokens, hand them to Aura and receive boosted rewards without managing Balancer's voting-escrow lock (veBAL, a long lock of BAL that raises reward rates) themselves. Its audit reports describe it as "a protocol built on top of Balancer to provide maximum incentives to Balancer liquidity providers through social aggregation of deposits and" its native token . The same report says "Aura can be seen as a spin-off of Convex Finance, which implements a similar model for Curve" .
- Target product: Aura Finance on Ethereum, the canonical deployment, plus its sidechain deployments listed by DeFiLlama.
- Category: yield aggregator and vote-locking layer on Balancer. DeFiLlama files it under "Yield" .
- Target users: Balancer LPs, BAL holders, and AURA holders who locked tokens to vote (vlAURA).
- Current status: the site states "it has been decided that the Aura Protocol will terminate operations" . The Aura app's Pools, Lock, Vault, Bridge and Redeem tabs remain.
Sunset timeline (official site)
| Stage | Date | What happens |
|---|---|---|
| 1 | 13 May 2026 | AURA holders redeem for a pro-rata share of treasury assets and receive rAURA redemption tokens |
| 2 | 10 May 2027 | Stage 1 claims close. rAURA redeems for unclaimed treasury assets plus 10% of the 80/20 BAL/WETH BPT |
| 3 | 10 May 2027 | auraBAL redeems for the remaining 90% of the 80/20 BAL/WETH BPT |
Stage 1 text: . Stage 2 text: . Stage 3 text: .
The page states that "LP, vault, and locker positions remain withdrawable at any time." . It also quotes an rAURA price range of $0.008705-$0.0215 "based on AURA redemption participation" . The page gives no method for that range.
The site links to a Snapshot proposal as the decision record. Its contents (dates, quorum, result) could not be read. Which chains the sunset covers is also not listed on the page.
How It Works
Aura sits between users and Balancer. The following flow describes the design, which the sources document only in part. The dated deployments show the Booster is now shut down, so steps 1-3 describe the original design and only withdrawal and redemption apply today.
- A user deposits a Balancer LP token into the
Booster, which its audit calls "the main deposit contract for liquidity pool tokens" . - The
Boosterpools liquidity providers' tokens: its job is "to allow liquidity providers from different pools to stake their tokens all together in a single place, collect the corresponding rewards, and track the accountancy properly to redistribute these among all the participants" . TheBoosterpoints at astakeraddress that matches theVoterProxyin the deployed-address list ; . - Rewards (BAL, AURA and extra reward tokens) are paid to depositors through reward-pool contracts. The deployed list names
RewardFactory,StashFactory,ExtraRewardStashV3,auraBAL RewardsandcrvDepositor.ExtraRewardStashV3wraps extra reward tokens in aStashTokenso governance decides whether they are processed . - BAL holders can convert BAL into auraBAL (Aura's liquid wrapper of locked BAL). AURA holders lock AURA in
AuraLockerand receive vlAURA (vote-locked AURA, used for governance voting). The Aura code repository warns that "auraBAL and vlAURA reward contracts are not on the 4626 standard" . - Today, a user withdraws LP, vault or locker positions, and AURA and auraBAL holders redeem on the schedule above.
Core mechanism
Convex-style booster: Aura's contracts hold BAL voting power in one place and share the boosted rewards with depositors. The Booster on Ethereum reports isShutdown: true with 283 pools, and AuraLocker is also shut down at the measured block ; .
Key smart contracts and roles (Ethereum)
| Contract | Role |
|---|---|
Booster (current: 0xA57b…B234) | Main deposit contract. An older Booster (0x7818…EE10) is also shut down |
VoterProxy | Holds the voting-power position on Balancer |
Aura token and AuraMinter | AURA token and its minter (owner measured as the zero address) |
AuraLocker | vlAURA vote-locking |
BoosterOwner / BoosterOwnerSecondary | Administrative owners of the Booster |
PoolManagerV4, PoolManagerSecondaryProxy | Add or shut down pools |
crvDepositor | Converts BAL into auraBAL |
L1Coordinator, AuraProxyOFT, AuraBalProxyOFT | Cross-chain messaging and token bridging hub |
Addresses come from the project's deployed-address list and .
Yield Sources
Yield came from Balancer's BAL emissions, boosted by Aura's pooled veBAL, plus AURA token emissions. Users also got extra reward tokens that Balancer pools offered. The Booster carries fee settings that split rewards between lockers, stakers and the caller who triggers harvests. On-chain, lockIncentive is 1800, stakerIncentive 400, earmarkIncentive 50 and MaxFees 4000 on a fee base of 10000 for the current Booster; the platformFee is 0 .
Assessment: yield was mostly emission-driven. Aura earned BAL emissions and paid AURA emissions, and no source shows fee income independent of emissions. The basis is the reward-pool structure above. Pools are withdrawal-only, so new yield is not expected from the pools, though this report did not check every reward contract on every chain. On Ethereum at block 26289853, the auraBAL Rewards pool shows a periodFinish of 1778934311, which is 16 May 2026 (12:25 UTC) . A VirtualBalanceRewardPool on the same chain shows 1778998331, which is 17 May 2026 (06:12 UTC) . Both dates are in the past on the snapshot date.
Data gap: no historical APR series is in the available sources.
Dependencies & Integrations
- Balancer: Aura depends on Balancer's vault, gauges, veBAL and BAL emissions. The
BalancerSwapsHandlerpoints at the Balancer vault0xba12…bf2c8. A shutdown or change on the Balancer side would directly affect Aura's remaining position. - LayerZero: the cross-chain contracts use LayerZero.
AuraProxyOFTon Ethereum names a LayerZero endpoint . The June 2023 Zellic review says that inPausableProxyOFT(a shared base contract for the cross-chain token proxies) thesudoaddress, set at deploy and not changeable, can move any tokens out throughrescue().AuraProxyOFTon Ethereum reports asudoaddress (0xb370…Bf46) . Whether that address holds the samerescue()power on this contract is not confirmed in the sources. LayerZero is an external message layer. - Oracles: no oracle is named in the sources. Aura does not lend or liquidate, so price feeds are not part of the described design.
- Chains: DeFiLlama lists Ethereum, Base, Arbitrum, Optimism, Polygon, xDai (Gnosis), Polygon zkEVM, Avalanche and Fraxtal . The canonical deployment lists bridge receivers for Arbitrum, Optimism, Polygon PoS, Gnosis, Base, Polygon zkEVM, Avalanche and Fraxtal .
- Bridges: the canonical deployment lists a bridge-delegate receiver per sidechain, and AURA and auraBAL move over LayerZero. The sources do not state which bridge each chain uses.
TVL & Adoption
DeFiLlama's latest captured snapshot shows about $9.71 million of TVL (collected 28 Oct 2026): ; . The same snapshot gives a market cap of about $1.42 million . Because the protocol is withdrawal-only, this TVL is remaining deposits that have not yet withdrawn, not live capital being deployed.
All of the counted TVL sits on Ethereum; the other listed chains report zero.
- Ethereum
- 9.71 USD m
- Polygon zkEVM
- 0 USD m
Data gap: no dated TVL history is in the available sources, so the pre-sunset peak and the speed of the exit are not shown. No Blockworks (ex-Messari) reports or articles are in the workspace.
Technical Architecture
Conclusion: control sits with a 4-of-7 multisig (a wallet needing 4 of 7 signers to approve) that owns the key Ethereum contracts. No timelock delay is measured for its direct calls. The Booster, AuraLocker and the sidechain boosters are shut down, which limits what new actions are possible.
Who controls what (Ethereum, block 26289853, 28 Oct 2026)
- DAO multisig:
0x5fea…17a0is a Safe (a standard multisig wallet) with 4 of 7 signers . It ownsAuraLocker,L1Coordinator,AuraProxyOFTand the olderBoosterowner . AuraLocker: the multisig can callshutdown(),recoverERC20,modifyBlacklistandtransferOwnership, and the measured fastest delay is 0 seconds .- Current
Booster: its owner isBoosterOwner(0x228a…f04), andBoosterOwneris owned byBoosterOwnerSecondary(0xCe96…EaD), which is owned by the multisig .BoosterOwnerreportsisSealed: trueand a 30-day (FORCE_DELAY: 2592000) forced-action delay . The forced-action timer has not been started .BoosterOwneris theBoosterowner . The 2023 upgrade review states that once governance hands ownership toBoosterOwnerSecondaryit cannot be rolled back, and stash implementations can be sealed . - Pool management:
addPoolon theBoosteris only callable from thePoolManagerProxy, controlled at the end of the chain by Aura governance . The measuredPoolManagerSecondaryProxyreportsisShutdown: true. - Emergency pause:
AuraProxyOFTreportspaused: false, with the multisig asguardian. The sidechain README says its pausable token contracts let a guardian pause in an emergency. - Sidechains: On Polygon,
0xd86c…d68bis a 4-of-7 Safe that owns most sidechain contracts, and Arbitrum lists the same address as owner . Base lists a different owner address (0x9553…130e). The PolygonBoosterLitereportsisShutdown: truewith 32 pools . The Arbitrum and Base boosters also report shutdown at their measured blocks. The Gnosis chain was not read, Optimism access was refused by the provider, and Avalanche contracts came back without names, so their controls are not described here. - Upgradability: not established. The Ethereum on-chain read is partial, and the available reads do not test whether the core contracts sit behind upgradeable proxies. The measured owners can change settings through setter functions (see the parameter list below).
- Parameters owners can change: fees (
setFees), treasury, reward multiplier, bridge delegate, vote delegate, factories and pool shutdowns, per the measured privilege list .
Some owner holders on the older Booster and the owners of several helper contracts were not fully resolved (some are marked "not read"), so the control list above is partial. The Ethereum on-chain read itself is marked partial.
- (role)
- (contract)
- (contract)
- (contract)
- (contract)
- control
- control
- control
- control
Key Risks (Protocol-Level)
- Wind-down and redemption risk: funds in the redemption path depend on Aura's treasury and on the staged schedule to 10 May 2027. The rAURA price range of $0.008705-$0.0215 is the official page's figure, with no method given . Stage 2 is described as redemption of rAURA, the token minted in Stage 1. The available source does not state what an AURA holder who did not redeem in Stage 1 can claim afterwards, so such holders should confirm this on the official app. Stage 3 for auraBAL also falls on 10 May 2027.
- Where user funds sit: deposits sit in the
Booster,VoterProxyand reward pools, with locked AURA inAuraLocker. Control by a 4-of-7 multisig with 0-second measured delay on several functions means signers can act quickly. Users must trust those signers, whose identities are not documented in the sources. - Bridge and cross-chain risk: sidechain AURA and auraBAL depend on LayerZero messaging and native chain bridges. The Zellic review reported that in
PausableProxyOFTasudoaddress can rescue tokens . - Complexity: the audit coverage list names 92 components across the original Convex-style contracts, the 2023 upgrades, the
auraBALcompounder vault and the sidechain system. The list also includesAuraSoulsV1, which one audit describes as a contract "for buying and selling subject souls with protocol, subject, and LP bucket fees" . Its link to the Balancer booster is not established, so the totals are only an upper bound for the booster system. In the list, 47 components have no attributed current address. How many of those belong to the booster system was not separated, so audit coverage of today's deployed booster code is not assessed here. - Audit record: the 2022 Code4rena contest (source reliability: unverified publisher), Halborn audits (June 2022, March 2023, May 2023) and Zellic reviews (June 2023) cover Aura's contracts, but most are about 3 years old and several are hosted on unverified publication sources. No audit of the redemption or sunset contracts (rAURA and the Redeem flow) is in the sources.
- Dependency on Balancer: Aura's remaining value depends on Balancer's BPT (the 80/20 BAL/WETH pool token) and the BAL position held by
VoterProxy. Stage 3 pays out BPT, so its value tracks Balancer's market. - Phishing risk: web search returned lookalike non-official domains describing redemption limits. Use only the official site.
Data Completeness
- Available: the official sunset page; DeFiLlama snapshot (28 Oct 2026); deployed-address documents; on-chain control reads for Ethereum, Polygon, Arbitrum, Base and Avalanche (Ethereum marked partial); 11 audit documents (several from unverified publication sources); extracted claims.
- Missing: the Snapshot sunset vote contents (the proposal page did not load); the list of chains the sunset covers; the rAURA price method; the redemption contract addresses and any audit of them (a second web search found neither, and the project's repository README gives only a general security statement ); and a historical TVL or yield series.
- Not read on-chain: Gnosis (59 contract reads not made), Optimism (provider refused), and part of Ethereum (truncated seeds and time budget).
- Declared gap: the project's own security documentation page was declared in the source list but does not resolve, so it was not read.
- No oracle data: no oracle is documented; the claim that none is used is an inference from Aura's design as described in the sources.
- Web check: a web lookup confirmed the sunset page and found no further official sources ().
66 citations · click markers to view evidence
Code Security
Code Security Analysis: Aura
Aura Finance (a Balancer booster with about USD 9.7M TVL ) has several external reviews: a Code4rena contest (source reliability: unverified publisher), three Halborn engagements and a Zellic assessment. The two high findings for Aura Finance are fixed. Of 28 medium findings (two of them, from the Spagnuolo governance upgrades report, are estimated mediums: that report states no rating), 21 are fixed and 7 are not: 3 are disputed by the team, 3 are acknowledged and 1 has no stated status. Two of the disputed ones (Code4rena M-03 and M-08, both about voting power) describe a pattern that is still present in the latest repository code. The available material does not show which deployed revisions each audit covered. Deployed code was not compared with any audited revision, so audit coverage of the live contracts is unconfirmed rather than disproved.
Data Completeness
Findings ledger: 135 finding groups from 11 included audit files. Four of those groups (one high, two medium, one low) come from a review of AuraSoulsV1 (source reliability: unverified publisher), a bonding-curve "souls" contract that appears to be a different product from Aura Finance. They are counted separately below.
Read in full: the findings ledger, the audit index, the document map, the coverage map and the audit trust warnings. The audit digests, the claims file and the aggregator files were also read. Audit sections were opened only to quote specific findings (Code4rena sections 05, 12, 16, 26 and 31, the Hephyrius sidechain review, the Halborn engagement headers and the Zellic introduction).
Known gaps:
- The official security page
https://docs.aura.finance/aura/securitycould not be collected: the collection record says the host did not resolve in DNS. No bug bounty terms are therefore available, and a web check found no bounty program page, so the bounty status is unknown and no assurance level is assigned. - Sunset: the protocol documentation states that Aura is ending operations . The contracts remain deployed while pools are withdrawn. The audited revisions have not been linked to or compared with the deployed code, so the findings below describe the audited code, not confirmed live code.
- Two sources (the Code4rena report and the sidechain contracts API doc) were digested from their first 80000 characters only.
- Repository code was located for only six audited findings. Repositories for the remaining audits were not collected, so those findings are not code-checked.
- Several named contracts have no attributed current address:
AuraSoulsV1.buySouls,GenericVault,PausableProxyOFTandConvexMasterChef.add. - Component roles: criticality tags in the structured output cite a role description where the sponsor README, the documentation or an audit gives one. For
ArbitartorVault,cCrv,PoolFeeManagerProxy,DelegateRegistry,GenericVault,ChildGaugeVoteRewards,ForwarderHandler,GaugeMigratorandViewno source describes the security role, so they are tagged at the lowest level and that rating is not established. Where the coverage map already rates a component, the tag cannot go below that code-derived rating and cites a role description where one exists. ForAuraBalVault,AuraMerkleDrop,AuraPenaltyForwarder,AuraRelockers,BalancerSwapsHandler,ChildStashRewardDistro,CrvDepositorWrapper,ExtraRewardStashLiteModule,FeeScheduler,GaugeVoteRewards,OptimismBridgeSender,VeBalGrant,VirtualRewardFactoryandWardenQuestSchedulerthe sources give only a listing, an address or an audit mention, so their tags show the map rating without an independent role description. - No collection gaps are declared for the coverage map.
Audit Coverage
Eleven audit files are included. Two of them (Halborn 2022 and one Zellic copy) come from auditor-controlled repositories. The others come from the Aura repository via DeFiLlama, or from code4rena.com, and their publication host is not verified as auditor-controlled. These still count as evidence, with lower source confidence.
| Audit | Auditor (tier) | Date | Scope |
|---|---|---|---|
| Aura Finance contest | Code4rena (community contest) (source reliability: unverified publisher) | 2022-05; report dated 2022-07-26 | AuraLocker, ExtraRewardsDistributor, CrvDepositorWrapper, BaseRewardPool4626 and others; 1 high and 22 medium findings listed |
| Smart Contract Security Audit | Halborn (tier 2) | 2022-05-16 to 2022-06-28 | aura-contracts at commit b67d5b7d7fb87455533b5376e7c20157a6fc4e8c |
| AuraBAL Compounder audit | Halborn (tier 2) | 2023-03-06 to 2023-03-20 | compounder/*.sol |
| AuraBAL compounder report | Facundo Spagnuolo (individual reviewer) | 2023-03-24 | AuraBalStrategy, AuraBalVault, handlers |
| Sidechain audit | Halborn (tier 2) | 2023-05-09 to 2023-06-06 | AuraBalOFT, AuraBalProxyOFT, AuraOFT, AuraProxyOFT, Create2Factory, CrossChainConfig and others |
| Smart Contract Security Assessment | Zellic (tier 2) | 2023-06-15 (a second copy is dated 2023-06-16) | Sidechain, bridge and Convex-platform contracts; zero findings |
| Sidechain review | Hephyrius (individual reviewer, undated) | undated | VoterProxyLite, PoolManagerLite, BoosterOwnerLite, BoosterLite, BaseRewardPool4626 |
| Upgrades: Booster Owner, Pool Manager, Reward Stash | Hephyrius (individual reviewer, undated) | undated | PoolManagerV4, BoosterOwnerSecondary, StashToken, ExtraRewardStashV3 |
| Governance upgrades | Facundo Spagnuolo (individual reviewer, undated) | undated | scope not stated |
AuraSoulsV1 review | unnamed (different product) (source reliability: unverified publisher) | undated | src/AuraSoulsV1.sol; not counted as Aura Finance review depth |
Evidence for the main engagements: ; ; ; .
The three Halborn engagements, the Zellic assessment and the Code4rena contest (source reliability: unverified publisher) are independent of the team. The Hephyrius and Spagnuolo reviews are by individual reviewers whose independence from the team is not shown. A curated third-party snapshot lists code review by "PeckShield, code4rena, and Halborn" . No PeckShield report is in the workspace.
Unaudited or unconfirmed areas: the deployed system includes many contracts that no included audit names (see below). The Zellic report itself notes it does not guarantee the absence of later issues .
Deployed code vs audits
Deployed code was not compared with any audited revision: all 185 verified protocol files (35876 lines) are in the "unknown" bucket (100.0% unknown; identical, cosmetic, changed and unaudited are each 0.0%). The map lists 92 components: 23 have an audit that names them but whose audited revision could not be pinned, 67 have an audit scope that is not confirmed, and 2 have deployed source that is not verified. A further 47 contracts are named in audits but have no attributed current address. Lines not covered by a confirmed comparison, by the criticality rating of the coverage analysis's own code assessment (not a model review): 14759 critical, 303 high, 517 peripheral and 20297 unrated. The roles it rates include funds, access control, upgradeability, oracle and view/utility. Some code has not been assessed for criticality, and 11 components were not measured. No files lacked line counts. Most components are verified by Sourcify runtime match, which concerns deployed bytecode only and says nothing about audit scope.
The rows below are all gaps; critical rows come first. Thirty of 92 components are shown; 62 more are in the same states. The 47 named-only contracts include AuraSoulsV1, GenericVault, PausableProxyOFT, ConvexMasterChef, AuraMath and Interfaces.
| Component | Deployment (first listed) | Status | Audit | Criticality | Verification |
|---|---|---|---|---|---|
L2PoolManagerProxy | eip155:43114:0x2b6c227b26bc0ace74bb12da86571179c2c8bc54 | Deployed source not verified; not compared with any audit | none named | critical | not verified |
VoterProxyLite | eip155:10:0xc181edc719480bd089b94647c2dc504e2700a2b0 (+3 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
PoolManagerLite | eip155:10:0xf24074a1a6ad620adc14745f9cc1fb1e7ba6ca71 (+3 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
BoosterLite | eip155:10:0x98ef32edd24e2c92525e59afc4475c1242a30184 (+3 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
AuraBalOFT | eip155:137:0x223738a747383d6f9f827d95964e4d8e8ac754ce (+1 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
AuraLocker | eip155:1:0x3fa73f1e5d8a792c80f426fc8f84fbf7ce9bbcac (+1 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Code4rena 2022-05 | critical | Sourcify runtime match |
ExtraRewardStashV3 | eip155:1:0x37c3ebfd4b0cf66df19a413e92dd21e556915f98 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (internal upgrade review, undated) | critical | Sourcify runtime match |
BoosterOwnerSecondary | eip155:1:0xce96e48a2893c599fe2601cc1918882e1d001ead | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (internal upgrade review, undated) | critical | Sourcify runtime match |
PoolManagerV4 | eip155:1:0x8dd8cdb1f3d419ccdcbf4388bc05f4a7c8aebd64 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (internal upgrade review, undated) | critical | Sourcify runtime match |
CrvDepositorWrapper | eip155:1:0x68655ad9852a99c87c0934c7290bb62cfa5d4123 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Code4rena 2022-05 | critical | Sourcify runtime match |
ExtraRewardStashV3 | eip155:1:0x4a53301fe213eca70f904cd3766c07db3a621bf8 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (internal upgrade review, undated) | critical | Sourcify runtime match |
GnosisBridgeSender | eip155:100:0x908c9d41183ade493effc12bb9ad11a9333052bc | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
ExtraRewardsDistributor | eip155:1:0xa3739b206097317c72ef416f0e75bb8f58fbd308 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Code4rena 2022-05 | critical | Sourcify runtime match |
BoosterOwnerLite | eip155:10:0xf044ee152c7d731825280350d876cf760181d96f (+3 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (sidechain review, undated) | critical | Sourcify runtime match |
AuraOFT | eip155:100:0x1509706a6c66ca549ff0cb464de88231ddbe213b (+2 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
Create2Factory | eip155:100:0x53c09096b1dc52e2ef223b2969a714ee75da364f (+1 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
ExtraRewardStashV3 | eip155:1:0xf9c0f3431f859e773ed052758052e06b6d175742 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Hephyrius (internal upgrade review, undated) | critical | Sourcify runtime match |
ProxyFactory | eip155:137:0x731886426a3199b988194831031dfb993f25d961 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2022-06-28 | critical | Sourcify runtime match |
L2Coordinator | eip155:100:0x8b2970c237656d3895588b99a8bfe977d5618201 (+2 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
L2Coordinator | eip155:10:0xec1c780a275438916e7ceb174d80878f29580606 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2023-05-09 | critical | Sourcify runtime match |
KeeperMulticall3 | eip155:1:0x817f426b5a79599464488eccf82c3f54b9330e15 (+3 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
GnosisSafeL2 | eip155:10:0x3e5c63644e683549055b9be8653de26e0b4cd36e (+3 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
ChildStashRewardDistro | eip155:10:0xca85e2ce206b48ee28a87b0a06f9519abe627451 (+3 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
TokenFactory | eip155:10:0x87299312c820607f1e7e4d0c6715ceb594306fe9 (+2 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
L2PoolManagerProxy | eip155:100:0x2b6c227b26bc0ace74bb12da86571179c2c8bc54 (+2 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
StashFactoryV2 | eip155:10:0x8401b48760e70a39e6bbf861abd050c00362bae8 (+2 more) | Audit scope not confirmed for this component; deployed code not compared | none named | critical | Sourcify runtime match |
View | eip155:10:0xfa6b857cc17740a946c9eb85c1a6896f2e0be98e | Deployed source not verified; not compared with any audit | none named | peripheral | not verified |
RewardPoolDepositWrapper | eip155:100:0x0fec3d212bcc29ef3e505b555d7a7343df0b7f76 (+1 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Halborn 2022-06-28 | peripheral | Sourcify runtime match |
AuraClaimZap | eip155:1:0x623b83755a39b12161a63748f3f595a530917ab2 | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Code4rena 2022-05 | peripheral | Sourcify runtime match |
AuraVestedEscrow | eip155:1:0x24346652e0e2ae0ce05c781501fdf4fe4553fac6 (+4 more) | An audit names this component, but the audited revision could not be pinned; deployed code not compared | Code4rena 2022-05 | peripheral | Sourcify runtime match |
Inventory: the protocol publishes address lists in its documentation , but no first-party statement says the list is complete, so the inventory is treated as partial.
Findings by Severity
Target product (Aura Finance): 2 high (both fixed), 28 medium (21 fixed, 3 disputed, 3 acknowledged, 1 not stated). Other product (AuraSoulsV1): 1 high, 2 medium and 1 low, all open; these are excluded from the target counts and from the target's unresolved count above.
Critical
No critical findings in the ledger.
High
- User can forfeit other user rewards (Code4rena contest (source reliability: unverified publisher), H-01) — — Audit status: fixed.
- The first depositor may lose funds due to ERC4626 rounding issue (Halborn AuraBAL Compounder audit, HAL-01) — — Audit status: fixed.
- Insufficient Enforcement on Initial Soul Acquisition (review of
AuraSoulsV1(source reliability: unverified publisher), another product; not Aura Finance) — — Audit status: open — Check: not checked: the audit's repository code was not collected.
Medium
- Improperly Skewed Governance Mechanism (Code4rena M-03) — — Audit status: disputed — Check: disputed by the team; the described code is unchanged; not scored (latest repository code
aurafinance/aura-contracts-lite@1d4997e, an archived repository) . - Locking up AURA Token does not increase voting power of individual (Code4rena M-08) — — Audit status: disputed — Check: disputed by the team; the described code is unchanged; not scored (
aura-contracts-lite@1d4997e) . - Users may lose rewards if rewards are fee-on-transfer tokens (Code4rena M-11) — — Audit status: disputed — Check: disputed by the team; the described code is unchanged; not scored (
aura-contracts-lite@1d4997e) . - AuraBalRewardPool charges penalty to all users if AuraLocker is shut down (Code4rena M-18) — — Audit status: acknowledged. The team replied: — Check: the latest public repository code (
aurafinance/aura-contracts-lite@1d4997e) may still contain the code this finding describes; archived repository, live status not established . - Duplicate LP token could lead to incorrect reward distribution (Code4rena M-22) — — Audit status: acknowledged — Check: a fix is visible in the latest repository code (
aurafinance/convex-platform@e7c23cf); deployment not confirmed . - Limit the number of reward tokens in the Strategy (AuraBAL compounder report, Facundo Spagnuolo, 2023-03-24, M1) — — Audit status: acknowledged — Check: classed as a gas/efficiency finding by the code check, which gives it no security weight; the audit itself describes a possible harvest-availability limit when reward tokens can be added without bound .
- Malicious sudo/admin can mint infinite tokens via rescue()/sendFrom() loop (Hephyrius sidechain review) — — Audit status: not stated — Check: not checked: the audit's repository code was not collected. The finding describes a privileged-role power on sidechain tokens.
- Lack of Fee Destination Update Functionality and Insufficient Enforcement on Final Soul Sale (review of
AuraSoulsV1, another product) — Audit status: open — Check: not checked: the audit's repository code was not collected. - Fixed medium groups (21): reported by Code4rena, Halborn and the two upgrade reviews; trusted as fixed and not re-checked. Two of them (the Spagnuolo upgrade findings on withdrawal freezing via
forceAddPooland on extra-rewards accountancy overflow) are medium (estimated): this is an AI estimate, because that report states no severity rating.
Low / Informational
Low: 20 groups (16 acknowledged, 3 fixed, 1 open; the open one belongs to AuraSoulsV1). Informational: 82 groups (41 with no stated status, 35 acknowledged, 6 fixed). Most are gas and style items. The 2022 Halborn engagement reported no critical, high or medium findings .
Recurring Patterns
- Three informational gas groups were reported by both Halborn (2022) and Code4rena (source reliability: unverified publisher): postfix operators in loops,
array.lengthin loop conditions, and!= 0versus> 0checks. All are acknowledged. - The two upgrade reviews report distinct findings, with one related theme. Hephyrius describes deposit freezing, where an overflow in the tokens received by
ExtraRewardStashV3can leave theBaseRewardPoolinoperable , and a separate deposit-theft scenario during the Booster shutdown that needs a privileged owner and operator role . Both are medium and fixed. Spagnuolo describes a governance griefing flow that freezes withdrawals throughforceAddPool, and a fix so that extra rewards' accountancy cannot overflow and freeze reward pools . Both of the Spagnuolo findings are medium (estimated) and fixed: the severity is an AI estimate because the report states no rating. The overlap is only that both reviews tie reward-token overflow in the extra-reward handling to frozen pools; the privileged-role deposit theft and theforceAddPoolflow appear in one review each. - One medium group,
lock()supporting onlymsg.sender, was reported by the Halborn sidechain audit and the Hephyrius sidechain review. It is fixed. - Reward accounting edge cases (locker rewards, fee-on-transfer tokens, duplicate pools) recur across the Code4rena findings.
Code Quality Assessment
Review volume is large for a protocol of this size: four reports from two independent firms (three Halborn, one Zellic) and a public contest (source reliability: unverified publisher), spread across the original Aura contracts (2022) and the AuraBAL compounder and sidechain contracts (2023). Both high findings are fixed, and 21 of 28 medium findings are fixed. The Zellic assessment found no issues. The reports cover different revisions, and the Zellic digest states that . No source names which reports cover the current deployed major version, so the depth of review of the live code is not established.
The main weakness is traceability. No included audit gives a pinned revision that matches the deployed contracts, and many deployed components (factories, bridge senders, pool managers, gauge vote rewards) are not named in any audit. The upgrade and sidechain reviews are by individual reviewers, and their reports are undated. Some Code4rena medium findings remain disputed or acknowledged. For two of them (M-03, M-08, both about voting power) the latest repository code still contains the described pattern and the team disputes the finding.
Of 157 addresses, 151 have a Sourcify runtime match, 4 have Etherscan-listed source (an explorer claim) and 2 are not verified: L2PoolManagerProxy on eip155:43114 and View on eip155:10. A runtime match concerns bytecode only and does not show which audit covered the code. The bug bounty status is unknown: the official security page was unavailable and a web check found no bounty program page, so no assurance level is assigned; the core contracts are mostly verified, but that alone does not establish the bounty condition. The protocol documentation says Aura is ending operations and pools are in withdrawal-only mode . A third-party snapshot reports no documented hacks since launch , and the hacks dataset holds no record for this protocol .
Score: 5.9/10
Rubric security-score-v2.5; data completeness 0.65
- Audit coverage
- 5.50 points
- Audit depth
- 3.50 points
- Unresolved findings
- 9.15 points
- Assurance
- 3.50 points
- High · open
- 1 findings
- Medium · open
- 2 findings
- Medium · not stated
- 1 findings
- Medium · acknowledged
- 3 findings
- Medium · disputed
- 3 findings
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Audit coverage (computed) | L2 | supported | 0.35 | 5.50 |
| Audit depth | — | not disclosed | 0.20 | 3.50 |
| Unresolved findings (computed) | L4 | supported | 0.30 | 9.15 |
| Assurance | — | not disclosed | 0.15 | 3.50 |
36 citations · click markers to view evidence
Economic Security
Economic Security Analysis: Aura
Conclusion. Aura is no longer an operating yield product. It is in a governance-approved wind-down: pools are in withdrawal-only mode, Stage 1 AURA redemption began on 13 May 2026, and AURA and auraBAL (the tokenised, locked-BAL receipt) holders are redeeming against treasury assets and a Balancer pool token. LP, vault and locker positions can still be withdrawn at any time. Withdrawals return Balancer pool tokens, so users must then exit through Balancer pools. Balancer's own wind-down (BIP-928) is reported only by community-tier and unverified sources, and its vote result and schedule are unconfirmed. auraBAL holders wait until 10 May 2027. Evidence on current yield, oracles, liquidity depth and loss cover is thin, so confidence is moderate to low.
TVL & Liquidity Profile
The latest DeFiLlama snapshot shows about USD 9.7M of TVL, all on Ethereum, taken on 2026-10-28.
- TVL: . Collection time: 2026-10-28T07:47:20Z.
- Chain split: Ethereum holds all of it; every other listed chain reports zero.
- Ethereum is the only chain with TVL, so it is the only material chain. The chain profile has no L2 stage because it is the base chain.
- AURA market cap is about USD 1.4M.
- Protocol contracts read on-chain: at Ethereum block 26289853 (2026-10-28 07:53 UTC) the v2
Booster(the contract that routes deposits into Balancer gauges) reportsisShutdown: true. This matches the sunset notice. The Gnosis chain (eip155:100) and Optimism were not read, so their state is not measured. - DEX depth: the measurement found no candidate token, so there is no depth figure at 2% or 0.5% impact. Depth is a gap, not zero. A web search for auraBAL and AURA depth on Ethereum captured no DEX or price data, so it remains unmeasured.
- Exit queue: LP, vault and locker positions can leave at any time. auraBAL has a hard delay: it is redeemable only at stage 3, dated 10 May 2027, for the remaining 90% of the 80/20 BAL/WETH pool token.
- Concentration: no holder or depositor distribution is in the sources, so top-holder concentration cannot be assessed.
Oracle Risk
Insufficient current data: no source describes the oracle setup of the live contracts. A targeted web search captured only the 2022 contest README, which says CrvDepositorWrapper adds single-sided BAL to the 80:20 BAL:ETH pool and converts the pool token to auraBAL, but not which price source sets the minimum output. (source reliability: unverified publisher; a May 2022 contest document that may not match deployed contracts). The only oracle evidence is a 2022 contest finding on that wrapper.
- Code4rena (May 2022, unverified publication host) found that
CrvDepositorWrappertakes a TWAP (time-weighted average price) from a low-volume 20/80 WETH/BAL Balancer pool, which updates infrequently, enabling sandwiching or failed transactions. The audit status is fixed, and the TWAP window is not stated. The contest report records Aura's reported resolution: use the wrapper price as a guideline and let thekeeperofAuraStakingProxyprovide the minimum output (minOut). That is a 2022 team statement ("for now") in an audit copy with unverified publication provenance. It does not show that the same arrangement is deployed now, and it makes the minimum output depend on a keeper key whose control is not described. - A Halborn review of the auraBAL vault flagged a hardcoded slippage setting, reported as risk accepted by Aura. The audit copy has unverified publication provenance (host not verified as auditor-controlled), so the finding and its status carry lower source confidence.
- No oracle-caused loss is documented. The sidechain contract documentation lists no price oracle.
Flash Loan & MEV Exposure
Flash-loan and sandwich exposure is documented only in 2022-2023 audits (the Code4rena and Halborn copies have unverified publication provenance); the sources show no live exploit of Aura.
- Sandwiching of harvest or deposit swaps was reported twice (the Code4rena TWAP finding and Halborn HAL-03 above). Halborn also reported an ERC-4626 first-depositor rounding bug in the vault, with status fixed as reported in the same unverified-provenance audit copy. Only Aura-specific audits count here.
- The Balancer v2 exploit of November 2025 used batch swaps with deferred settlement, which act like flash loans, on Composable Stable Pools. (source reliability: unverified publisher). That is a Balancer vector. Aura's exposure follows from the LP tokens it holds, and the sources do not quantify it.
- Booster shutdown and withdrawal-only status are observed (
isShutdown: trueat block 26289853). Remaining MEV exposure on other activity (withdrawals, redemptions, vault, locker, bridge or Balancer pool exits) is not assessed, and the sources contain no MEV analysis of it.
Tokenomics Risk: Yield, Leverage & Reflexivity
No workspace document gives a cited revenue-versus-emission split for current yield. A targeted web search also did not establish the AURA emissions share, the split of any bribe or voting-incentive income, or current documentation (docs.aura.finance was not captured), so the yield design cannot be graded from the sources.
- Historical design: Aura boosts Balancer LP and BAL stakers by pooling BAL into veBAL, and pays extra AURA incentives. (curated snapshot dated 2026-03-09, before the wind-down.)
- Sunset: the governance proposal (source reliability: unverified publisher) says yield for locked AURA holders was projected to fall below running costs and proposes "turning off AURA emissions". The official site confirms termination of operations.
- Reward mechanics, historical: the 2022 contest README says anyone can call
earmarkRewardsandearmarkFeeson the Booster to harvest BAL and extra gauge rewards, and the caller is paid an incentive. (source reliability: unverified publisher; pre-launch document that may not match the sunset-state contracts.) Rewards to auraBAL stakers use the same path. This describes mechanism, not the current revenue-versus-emission split. - Leverage and looping: none appears in the user position in the sources. No LTV or unwind rule applies.
- Own-token reflexivity: AURA is redeemed for treasury assets, and the redemption token
rAURAis quoted at a price range. The proposal quotes an approximate redemption value of USD 0.0226 per AURA (unverified publisher). The official site quotes this as therAURAprice range based on redemption participation, and the proposal gives its own estimate under then-current assumptions. Neither is yield. auraBAL is backed by BAL/WETH pool tokens, not by AURA.
Solvency, Liquidation & Bad Debt
Aura holds no debt positions, so liquidation and bad debt do not apply; the solvency question is whether staked LP tokens and locked BAL can be returned.
- Backing: LP deposits wrap Balancer pool tokens, and auraBAL is backed by the protocol's 80/20 BAL/WETH position locked in veBAL. The official page says 90% of that position goes to auraBAL holders at stage 3. The proposal (unverified publisher) says the lock unwinds in about 12 months, with "almost no counter-trade liquidity" for auraBAL in the meantime.
- Exit path: LPs, vault and locker positions withdraw at any time (official page). auraBAL exits at 10 May 2027. No exit is reported halted at the research date.
- Backing versus liabilities is not measured on-chain in the workspace: no token reads exist.
- Loss handling: no insurance fund, reserve or compensation fund appears in the sources, and a targeted web search identified none (absence in captured pages, not proof of none). Treasury assets are being distributed to AURA holders, not held for users.
- Bad-debt record: none found for the current version. No record is not proof of none.
Incentive Alignment
Incentives have turned from growth to closure: the official site confirms termination of operations and AURA redemption against treasury assets, while the proposal to turn off AURA emissions comes from an unverified publisher and its implementation is not confirmed.
- Mercenary capital: the proposal (unverified publisher) says depositors left after the Balancer exploit, BAL emissions were cut, and TVL fell.
- Death spiral: the exposure is to Balancer-side reflexivity (BAL and AURA price falls and fewer incentives). Aura's wind-down was decided by votes (official site); Balancer's is reported only by weaker sources. The proposal describes deteriorating conditions: it says BAL emissions were cut, token prices fell, depositors withdrew and TVL dropped, "eroding the incentive flywheel" (unverified publisher). No source establishes a death spiral, whether any shortfall exists, or how the wind-down ends for non-redeeming holders. Exponential's earlier "no death spiral" call predates this (curated snapshot dated 2026-03-09).
- Loss cover: no fund found, including in a targeted web search. The DeFiLlama hacks record for this family is empty, so there is no payout history. A Binance Square post of August 2025 claims a USD 150M hack of Aura. (source reliability: unverified publisher). No non-weak source supports it, and the hacks dataset and Exponential ("No documented protocol hacks since launch") do not show it, so it is treated as unconfirmed.
- Chainalysis incident-response plan: the repository says Aura adopted one, which helps asset recovery but is not a funded cover.
Economic Attack Vectors
The main identified risk is dependency on Balancer, which is reported (by community-tier and unverified sources only) to be winding down; no price attack is documented, but other pool, contract and redemption risks are not excluded.
- Balancer dependency: Aura sits on Balancer's gauges, veBAL and pools. The Balancer v2 exploit hit Composable Stable Pools on many chains. (source reliability: unverified publisher). A Balancer post-mortem notes Aura assessed its exposure. (source reliability: community publication). Whether Aura users lost funds is not stated.
- Balancer wind-down (unconfirmed): the Balancer forum proposal BIP-928 sets 30 October 2026, two days after the research date, as the date when pools that can be paused move to withdrawals only. (source reliability: community publication). A post attributed to Balancer on X says BIP-928 passed. (unverified publisher). The captured forum thread shows only the proposal and a scheduled Snapshot vote, and a reply says the vote was reset; no official page captured shows the vote result or that the switch was executed. (community publication). Treat the vote result and the schedule as unconfirmed. Aura LPs who withdraw receive pool tokens and must then exit on Balancer; pools that cannot be paused continue, and the proposal says the frontend and routing treat the protocol as discontinued after these dates. Balancer states the contracts are non-custodial, so "withdrawing does not depend on Balancer or on anyone continuing to operate", and that recovery mode keeps withdrawals open where pausing requires it. (community publication). Balancer also says "withdrawals stay open the whole way through". (unverified publisher).
- Assessment: one identified exit risk is conditional. If the proposed schedule holds, an LP who delays exiting past Balancer's withdrawal-only date may have to rely on a discontinued interface, routing or bounty cover, which is the proposal's stated plan and not a confirmed outcome. The Balancer forum post says its bug bounty ends on 30 October 2026 (community publication). The open-withdrawal statements are proposed arrangements from community-tier and unverified sources, so they do not rule out other losses: pool-level, contract, bridge or redemption risks, and uncertainty about how each pool is treated (the proposal says treatment of each pool is published before 30 October).
- Bridge risk: sidechain deployments used an OFT (omnichain fungible token) bridge for auraBAL. DeFiLlama shows zero TVL on those chains, and the Gnosis state was not read, so residual bridge exposure is not measured.
- Liquidation cascades and bad-debt accumulation: not applicable; no lending leg.
Aggregator Scores
- DeFiLlama (captured 2026-10-28): TVL about USD 9.7M, category Yield, all on Ethereum.
- Exponential (curated snapshot, captured 2026-03-09 from the Wayback Machine): rating 7.0, "Good". It predates the wind-down and describes a live product with "Top 5% by total value locked", which no longer holds at USD 9.7M TVL. Treat it as historical.
- CertiK and DeFiSafety files are not present in the workspace.
- Comparison: the aggregator view of a mature, unhacked Balancer wrapper is outdated. The current picture is a managed shutdown with a hard dependency on Balancer.
Data Completeness
Available: DeFiLlama TVL, the official site sunset notice, 2022-2023 audits (Halborn auraBAL vault and Code4rena copies have unverified publication provenance), the Exponential snapshot, and Balancer and Aura governance pages (mostly weak publications).
- Web research run for this step (two tasks, both completed): one re-checked the Aura and Balancer wind-down and found the official Aura notice and the BIP-928 proposal text, but no captured official page for the BIP-928 vote result or for execution of the 30 October 2026 switch, and no page naming AIP-78 or an Aura withdrawal-only date. The other targeted the oracle, yield split, loss cover and DEX depth and captured only the Aura site and the 2022 contest README; it established none of those four. Their limits are the page count captured, not evidence that the features are absent.
- Missing: a revenue/emission split, holder concentration, DEX depth, token backing reads (asset checks and dependency graph return no candidates; the composition read was "too large" to assess fully), any current oracle description, a loss-cover fund, and the Aura security page (declared not published). A search for it found no relevant result.
- The Optimism and Gnosis state was not read on-chain.
- Confidence: moderate on Aura's wind-down status (official site plus on-chain
isShutdown), low on the Balancer wind-down vote and schedule, and low on oracle, yield and liquidity. The previous run's reports were not used as evidence.
Score: 4.5/10
Rubric security-score-v2.5; data completeness 0.50
- Solvency redemption
- 5.50 points
- Oracle design
- 3.50 points
- Dependency risk
- 5.50 points
- Liquidity concentration
- 3.50 points
- Yield design
- 3.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Solvency redemption | L2 | supported | 0.35 | 5.50 |
| Oracle design | — | not disclosed | 0.25 | 3.50 |
| Dependency risk | L2 | supported | 0.15 | 5.50 |
| Liquidity concentration | — | not disclosed | 0.10 | 3.50 |
| Yield design | — | not disclosed | 0.15 | 3.50 |
44 citations · click markers to view evidence
Team & Transparency
Social & Team Transparency: Aura
Aura announced and began a protocol wind-down in May 2026, so it is a closing protocol, not an active one. Pools are withdrawal-only and holders redeem on a published schedule through May 2027. The app Terms name a legal entity, Endpoint Ltd., a British Virgin Islands company, as the counterparty for the website and interface only. The sources name no individual team members, and the Terms do not make that company accountable for the protocol. The contract addresses are published and no exploit of Aura itself is recorded. Confidence is moderate: the people behind Aura appear only as pseudonymous handles, and several social and security pages could not be collected.
Team Identity
Aura's app Terms name a legal entity with a jurisdiction for the website and interface, but no individual person is named. The only people credited are pseudonymous handles.
- The wind-down proposal AIP-78 (source reliability: unverified publisher) lists its authors only as handles: .
- Recent code commits on the main contracts repository come from one handle,
0xahtle7. - The app Terms of Use are entered into between the user and
Endpoint Ltd., a British Virgin Islands company . They are governed by BVI law , with disputes going to binding arbitration at the London Court of International Arbitration . - The Terms limit that role. They call the company "a developer of open-source software" that "does not unilaterally offer, operate, or administer the Protocol", and they say it "has no oversight on or control over" the Protocol . They also say the company cannot change or stop transactions once submitted to the Protocol . They are dated 5 June 2024 , before the wind-down.
- The captured Terms give no registration number, registered address or directors. The sources do not confirm whether
Endpoint Ltd.still operates the live interface. - The sources give no founder names, backgrounds or prior projects. A web search for a team or about page found none.
- The Aura X account shows it joined in January 2022 (source reliability: unverified publisher) . That dates the account, not any individual's tenure.
- Third-party material describes Aura as a Balancer add-on that "can be seen as a spin-off of Convex Finance" . It names no operators.
Assessment: the named entity is accountable only for the website and interface under the Terms. The Terms do not establish who is accountable for the protocol itself, so protocol-level accountability rests on pseudonymous handles, governance votes and public code, not on identifiable people. The sources observe only the handles above, and they do not tie any handle to two or more years of verifiable work or show how long the same operators have run Aura.
Track Record
Aura has run since mid-2022 with no recorded exploit of its own contracts, and it is now ending by governance vote after a decline the proposal blames on Balancer-related changes.
- Family listing date is 2022-07-22 (code-derived). The current Ethereum
Boostercontract was created on 2022-12-13 . An earlier deprecatedBoosterdates from 2022-06-09. - The DefiLlama hacks dataset has no matching record for Aura (the
recordslist is empty). No incident row is therefore reported. A Binance Square post (source reliability: unverified publisher) claims Aura was hacked for $150 million . No other source corroborates it and the hacks dataset has no matching record, so it is treated as unconfirmed. - Balancer's November 2025 V2 exploit post-mortem (source reliability: community publication) says partners including Aura "responded quickly to assess exposure and protect users" . It states no Aura loss.
- Outage and pause history: the available sources give no complete record of Aura-specific pauses, outages or exit halts for 2022-2026, and a targeted web search found none. The sources therefore neither show nor rule out such events, and no resilience level is assigned.
- Delivery: the repository shows regular commits from July 2025 to April 2026, then the wind-down. Aura's own site says "it has been decided that the Aura Protocol will terminate operations" and "Pools are now in withdrawal-only mode" .
- The AIP-78 authors (their own account) say the reason was Balancer's decision to retire the veBAL system Aura builds on .
- The same proposal says that "since the recent Balancer exploit and subsequent ecosystem changes, the economic conditions that previously supported Aura and Balancer have materially deteriorated", and that "depositors have pulled back due to perceived security risk" . It gives no figures for the decline in TVL or token price.
- Stage 1 (AURA redemption for treasury assets) opened 13 May 2026. Stages 2 and 3 are scheduled for 10 May 2027, and the site says LP, vault and locker positions remain withdrawable at any time . The site quotes an rAURA redemption price of "$0.008705-$0.0214" .
- Depeg history of own tokens: the measured-asset check found no candidate tokens (
assets: [], reason "no_candidates"), so no depeg data exists for auraBAL or AURA . Theassetslist is empty.
Social Presence
Development was steady until the wind-down; community channels are only partly visible.
- GitHub: the
aura-contractsrepository shows 534 commits . The commit list runs through April 2026 and covers each month from October 2025 to April 2026, with biweekly gauge-weight and voting commits. The last visible commit is dated 10 April 2026, before the 13 May 2026 wind-down. Contributor count was not shown on the collected page. - X/Twitter (source reliability: unverified publisher): the account lists "19.1K Followers", "2,283 posts" and a pinned wind-down notice . Replies to the notice show holders regretting the closure; engagement quality otherwise cannot be assessed.
- The wind-down post of 14 May 2026 (source reliability: unverified publisher) shows about 10.6K views . This figure comes from that capture alone and is unconfirmed.
- Governance: the AIP-78 vote on Snapshot passed, and the capture lists 8 votes (source reliability: unverified publisher; the count is unconfirmed) .
- Discord and Telegram: no size or discussion data was collected. Blog and Medium updates: none from the team were collected. The security docs page
docs.aura.finance/aura/securityis a declared crawl item that could not be retrieved.
Transparency Indicators
Contract addresses are published first-party; audits are published; treasury reporting and a bug bounty were not found.
- Address list: the official docs publish mainnet addresses for
VoterProxy,Booster,AuraLockerand others . - Audits: the repository has an
auditsfolder withauraBalVault,sidechainandupgradessubfolders . Reports from Code4rena (May 2022 contest), Halborn and Zellic are in the audit set; their independence and coverage belong to the code-security analysis. - Incident readiness: the repository README says Aura partnered with Chainalysis on an Incident Response Plan . It lists
security@aura.financefor contact. - Treasury: the wind-down proposal describes treasury assets as "ETH + AAVE" to be distributed pro-rata . No periodic treasury, revenue or expense report was found.
- Bug bounty: none is stated in the collected sources, and the security page could not be retrieved. Whether one existed is not established.
- Open source: the contracts, subgraph and docs repositories are public on GitHub, so the code is open.
Red Flags
- No individual team member is named; only pseudonymous handles appear, and the Terms entity
Endpoint Ltd.has no disclosed registration number, address or directors. - The protocol is winding down; the official site quotes an rAURA redemption price of "$0.008705-$0.0214" . No earlier price is cited for comparison.
- The wind-down proposal's capture lists only 8 votes. The count rests on one unverified capture, so the sources do not confirm it or show participation.
- No bug bounty or periodic treasury report found.
- An unverified Binance Square post claims a $150 million Aura hack. It has no corroboration and is treated as unconfirmed.
Positive Signals
- The wind-down is orderly: a governance vote, a published redemption schedule, and a standing promise that LP, vault and locker positions stay withdrawable.
- First-party mainnet address list and public source code.
- Steady commits for several months before the closure; no recorded exploit of Aura.
- A Chainalysis incident response plan is documented.
Data Completeness
- Available: official site, GitHub commit page and audits folder listing, deployed-address docs, Snapshot wind-down proposal, X pages and Balancer forum posts. No Blockworks reports or articles were collected.
- Missing: team names and biographies, entity registration details, Discord and Telegram data, blog/Medium updates, contributor count, the security docs page, any bug bounty record and treasury reports.
- This analysis rests on collected pages, not live social-media scraping. X, Snapshot and Etherscan captures come from unverified publishers and support only the facts they state.
Score: 6.2/10
Rubric security-score-v2.5; data completeness 1.00
- Team identity
- 7.50 points
- Disclosure
- 7.50 points
- Development activity
- 1.00 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Team identity | L3 | supported | 0.45 | 7.50 |
| Disclosure | L3 | supported | 0.35 | 7.50 |
| Development activity | L0 | supported | 0.20 | 1.00 |
Track record: 8.6/10
Rubric security-score-v2.5; data completeness 0.80
- Operating history
- 9.50 points
- Incident record
- 10.00 points
- Resilience
- 3.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Operating history (computed) | L4 | supported | 0.30 | 9.50 |
| Incident record (computed) | L4 | supported | 0.50 | 10.00 |
| Resilience | — | not disclosed | 0.20 | 3.50 |
31 citations · click markers to view evidence
Governance
Governance Analysis: Aura
The official notice says Aura's shutdown was decided by votes on Snapshot (an off-chain voting tool), while the on-chain read shows a 4-of-7 Safe multisig holding the owner powers on Ethereum. How, or whether, the Safe is bound to carry out vote outcomes is not established by any official or on-chain source. The read shows that Safe calling many bridge, locker and coordinator functions directly with a measured delay of zero, and no timelock was found among the addresses read. The official site says the DAO has voted to wind the protocol down. Pools are now withdrawal-only, so the practical user risk is whoever keeps these admin powers during the exit period. Signer identities, vote weighting and turnout, whether votes bind the Safe, and the sidechain controls could not be established, so confidence is moderate.
Governance Structure
Votes are held off-chain on Snapshot (per the official notice), and the owner powers sit with multisig wallets on-chain. The link between the two is unconfirmed: the only description of it is a 2023 community overview (see Voting mechanism). No on-chain Governor contract or timelock was found among the addresses read. The Ethereum read is partial: some seed addresses were truncated and others skipped under a time budget.
- Type: off-chain Snapshot voting alongside Safe-held on-chain powers; the execution relationship is unconfirmed. The official site says the shutdown was decided by votes:
- Voting mechanism: Snapshot, a gasless off-chain vote. Proposals are numbered AIP (Aura Improvement Proposal), and the wind-down is labelled AIP-78 on the Snapshot capture. A 2023 third-party overview, an unconfirmed historical account of the execution relationship, says Aura "uses Snapshot for all aspects of governance, with multisigs confirming the vote outcomes" and that voting power comes from AURA locked for 16 weeks (vlAURA) and delegated. (source reliability: community publication) No official source read describes the voting strategy, so the weighting rules are unconfirmed.
- Wind-down vote: the official sunset notice says the shutdown was decided by Aura votes (quoted above). The tally and turnout are unconfirmed: they come only from a Snapshot page capture of unknown publisher reliability, which shows AIP-78 as passed with For 17.2m, Against 0 and 8 votes. (source reliability: unverified publisher) No second source confirms the tally or the voter count: web searches for the result found none. I draw no participation or concentration conclusion from them.
- Proposal process and quorum: not documented. Targeted web searches found no proposal threshold, quorum, voting delay, voting period, snapshot-block rule or charter binding the Safe to votes. The only governance description found is the community overview above. Insufficient data for this assessment.
- Timelock: none found among the addresses read, and not proven absent for the unread ones. The privilege rows for the DAO Safe's direct calls show a
fastest_delay_secondsof 0 (see Parameter Controls). The one timer read is a 30-dayFORCE_DELAYon forced shutdown in the booster owner contract. Rows whose holder was not read carry a null delay, which means unknown, not zero. - Emergency bypass: on the Ethereum-side
ProxyOFTbridge contracts, the DAO Safe can callpause()andunpause()instantly, and an 8-of-13 "Sudo" Safe can callrescue. - Vote concentration: not measured. The only venue in the concentration data is the separate gauge-voting space
snapshot:gauges.aurafinance.eth, whose type is unsupported, so top-10 share and quorum share are null. That space votes on Balancer gauges, not on protocol changes, and the main Aura space is not in the measured set. No voting venue is named in the structured output for that reason, and the unconfirmed AIP-78 voter count is not used to estimate concentration.
Admin Key Analysis
Core powers on Ethereum sit with a 4-of-7 Safe. Instant rescue powers also sit with individual EOAs (externally owned accounts, single keys) and with 2-of-3 Safes.
Current Ethereum state is at block 26289853 (2026-10-28). The Safes below are:
| Safe | Threshold | Role |
|---|---|---|
DAO Safe (0x5fea4413…) | 4 of 7 | Resolved as dao() on the deployed contracts. Owns AuraLocker, L1Coordinator, both ProxyOFTs, the booster owner contracts, and others. |
Second DAO-labelled Safe (0xfc78f8e1…) | 4 of 7 | Holds AuraMerkleDrop rescueReward and withdrawExpired. |
Sudo Safe (0xb370ebd7…) | 8 of 13 | Calls rescue on the AuraProxyOFT and AuraBalProxyOFT bridge contracts. |
Admin Safe (0xab9ff9fb…) | 2 of 3 | Calls cancel on the vested escrow. |
Safe 0x21aed3a7… | 2 of 3 | Owns the Warden quest scheduler, with execute and emergencyWithdraw. |
- DAO Safe:
- Sudo Safe:
- Admin Safe:
- Signer overlap: one address (
0x30019eb1…) appears in the DAO, Sudo, Admin and0x21aed3a7…signer lists. The files read give no signer names, and web searches found no page naming the signers of the DAO or Sudo Safes (a Balancer documentation page lists Balancer's own multisig signers, not these), so whether signers sit outside the team cannot be shown. One address on the 2-of-3 Safe0x21aed3a7…(0x098cf33e…) also matches the author profile on the Snapshot capture of the wind-down proposal. That match is unconfirmed, because it rests only on the capture of unknown reliability, and it would not show whether that person is team or independent. (source reliability: unverified publisher)
What the keys can do, per the privilege data:
- Owner calls on the auraBAL vault, instantly: the DAO Safe holds
vaultExecuteonAuraBalProxyOFTwith no delay. The contract documentation says this function runs a call on the vault as the vault's owner. Which vault owner functions exist, and whether they reach user deposits, was not read. - Locker token recovery, limited in scope: the DAO Safe calls
recoverERC20onAuraLockerwith no delay. In the repository source, the function rejects the staking token (AURA) and any registered reward token, so it does not reach locked principal or registered rewards. It reaches other tokens held by the locker. The source guard line (AuraLocker.sol:257) matches the privilege row, but I did not compare the deployed bytecode with the repository. The locker also reads as shut down on-chain at block 26289853. - Single EOA with instant fund-moving power:
0x30019eb1…can callrescueTokenon theauraBalVaultreward handler. A second EOA (0x5452e6ab…) holds the same function on two other handlers. These handlers sit on the harvest path of the auraBAL vault. User principal custody there is not shown, so the exposure is likely limited to in-transit reward tokens. - Booster control: the on-chain owner chain on Ethereum (block 26289853) is Booster
0xa57b8d98…owned byBoosterOwner0x228a1420…, which is owned byBoosterOwnerSecondary0xce96e48a…, which is owned by the DAO Safe. An audit digest of the Hephyrius upgrade review (collected 2026-10-23; date of the review not stated; unverified host) describesBoosterOwnerSecondaryas proposed and not yet executed, pending the AIP-20 vote. The on-chain owner chain now shows the wrapper in place, so that digest's "proposed" status is out of date. Per the digest, the wrapper can proxy allBoosterOwnerfunctions, and a transfer of ownership to it cannot be rolled back. - Mint on sidechains: a sidechain review (published as the Hephyrius internal report, unverified host) flags that a malicious sudo or admin user could mint tokens by looping
rescue()andsendFrom(), rated Low-Medium. The audit's publication host is unverified, which lowers source confidence but not the protocol's trust. A web search found no source stating whether the audit's item was fixed, acknowledged or disputed, so its status is unknown. - AURA minting on Ethereum: the minter's owner is the zero address, so the owner path is renounced. Whether the minting right itself remains with the booster is not shown.
- Pause: the DAO Safe holds
pause()andunpause()on both bridge contracts. That is a benefit in an incident, since a 4-of-7 Safe can act at once.
Powers are reasonable for a protocol that bootstraps by operating Balancer deposits. But owner calls with no measured delay are a centralization cost, and the EOA rescue powers are broader than a pause-only role.
Upgradability
Most core contracts are Convex-style fixed contracts, and their behaviour changes through owner calls rather than proxy upgrades.
- Pattern: the Ethereum entries of the Aura contract table show no standard proxy slots set. The only proxies seen are third-party bridge and Polygon token contracts, not Aura's own. A missing proxy slot does not prove a contract cannot be upgraded, but no upgrade function appears in the privilege list either.
- Who changes behaviour: the DAO Safe calls owner functions directly on the locker, the bridge contracts and the coordinator. The current Booster is owned by the
BoosterOwnerchain, which the DAO Safe controls. Three Booster functions (setFees,setTreasury,setBridgeDelegate) are held by fee manager0xd0521c06…, whose type and signers were not read, so who can change booster fees today is not established. The Booster also reads as shut down. - Timelock on upgrades: none found among the Ethereum addresses read. A forced shutdown of a bugged pool uses a timer (30 days on-chain, see Governance Structure). That
BoosterOwnerLiteis the immutable variant used on some sidechains (the docs name zkEVM, Avalanche and Fraxtal), not Ethereum. - Immutable parts: some owner fields are the zero address (the pool manager proxy and a few others), meaning a renounced owner. The AURA token's minter has also lost its owner.
Parameter Controls
Where the holder was read, the DAO Safe changes parameters by direct owner calls with a measured delay of zero. Where the holder was not read, the delay is unknown.
- What changes: reward and fee settings (
feesrows), pool additions, gauge votes, harvester allowlists, and rescue token routing. - Who: the DAO Safe holds 46
configrows and 9feesrows at a delay of 0 in the privilege list. These cover the locker, both bridge contracts, the coordinator and an older Booster listed asboosterV1(0x7818a1da…). They do not cover the current Booster'ssetFees, whose holder was not read (above). An example issetQueueDelayonAuraProxyOFT, a bridge parameter. - Bridge receivers (Ethereum side): the Ethereum chain entry lists eight
BridgeDelegateReceivercontracts labelled for Arbitrum, Optimism, Gnosis, Polygon, Base, ZkEvm, Avalanche and Fraxtal. The sidechain documentation links the Arbitrum-labelled one to an Ethereum address, so these are Ethereum-side contracts associated with those sidechains, not contracts on the sidechains themselves. Each of the eight is owned by contract0x817f426b…, aKeeperMulticall3, whose owner is the single EOA0x30019eb1…. So owner calls on those eight Ethereum-side receivers follow one key. What the receivers' owner functions can change was not read, and the controls of the contracts on the sidechains themselves (including Gnosis and Optimism, which were not read) are not covered by this row. - Bounds: the current Booster exposes a
MaxFeesvalue of 4000 against aFEE_DENOMINATORof 10000. I did not read the code that enforces it, and bounds on the bridge and locker parameters were not extracted. Insufficient data for this assessment. - User harm: possible through fee or rescue changes. The protocol is now in withdrawal-only mode, and positions stay withdrawable.
Decentralization Assessment
On the facts read, a 4-of-7 Safe plus a few EOAs control the contracts, and the role of the votes in directing them is unconfirmed.
- Today: off-chain voting (per the official notice), on-chain powers held by Safes and EOAs, no timelock found. Whether the Safe is bound to vote outcomes is unconfirmed. This sits toward the centralized end of the spectrum, with a small, partly overlapping signer group.
- Roadmap: none beyond the wind-down. The official notice confirms withdrawal-only pools and redemption stages. The Snapshot capture lists further proposal items, such as sunsetting the Aura Foundation and subsidiaries and turning off AURA emissions, but that is unconfirmed. (source reliability: unverified publisher) No plan to renounce or timelock admin keys after the wind-down was found. The Balancer DAO says it plans its own permission cleanup, but that concerns Balancer, which is another product.
- Key person risk: the shared signer
0x30019eb1…is on four Safes and also holds EOA rescue powers. No single signer holds the DAO Safe's powers, but 4 of 7 colluding or compromised signers could make the instant owner calls listed above. How far those calls reach user deposits is limited for the locker (AURA and registered rewards are excluded) and unmeasured for the vault owner calls. - Token distribution: not measured. The turnout shown on the Snapshot capture is unconfirmed, so I draw no conclusion about concentration or participation.
- Status: the protocol is winding down. The notice gives Stage 1 as 13 May 2026 and Stage 3 as 10 May 2027. The notice does not say which accounts administer the treasury and redemption contracts in that period, so future authority is unknown.
Governance Risks
- Single point of failure: the DAO Safe (4-of-7) makes the locker, bridge and coordinator owner calls listed above with no measured delay, and sits at the top of the current Booster's owner chain. A compromise of four signers could act at once. The reach into user deposits is limited for the locker and unmeasured for the vault owner calls.
- Booster shutdown deposit-theft and deposit-freezing issues, current exposure not established: the Hephyrius review digest lists two privileged-actor issues, both acknowledged and rated collectively "low to medium risk". Theft needs a contrived sequence by an attacker holding owner and operator rights. Freezing needs a malicious token whose received amount overflows, plus a governance hijack. The review says the proposed upgrades address both:
PoolManagerV4removesforceAddPooland seals the pool manager,BoosterOwnerSecondaryseals ownership, andStashTokenwraps extra rewards with a 1e38 supply cap. On-chain at block 26289853, the ownership-sealing parts are in place: the booster owner contract reads as sealed and owned byBoosterOwnerSecondary, and the secondary pool manager reads as owned byPoolManagerV4and shut down. I could not confirm that the stash-side fix (StashTokenand the modifiedExtraRewardStashV3) is deployed. Because of the extra steps, the reviewed mitigations and the shutdown state, I do not conclude that either path is executable today, and I do not treat the Safe's key set as sufficient for them. - EOA keys:
rescueTokenon reward handlers can be called by single keys with no delay. - Sidechains: the Sudo Safe (8-of-13) controls bridge
rescueand, per that sidechain review, could mint on sidechains. Sidechain Safes on Polygon, Arbitrum, Base and Avalanche show thresholds of 4-of-7, 5-of-9 and 3-of-6. The Gnosis chain was not read and Optimism was refused by the provider, so those controls are not measured. - Governance attack and apathy: resistance to borrowed or flash-loaned voting power is not established. No source found describes the Snapshot space's voting strategy, voting delay, voting period or snapshot-block rule. The community overview says power comes from AURA locked for 16 weeks and delegated, which would make short-term borrowing harder, but no official source confirms it. Turnout on the wind-down vote is unconfirmed (see Governance Structure). Whether the Safe is bound to act on vote outcomes is not established, and no charter was found in a partial search.
- Future authority: the redemption schedule runs to 10 May 2027. The files read do not show which accounts administer the treasury and redemption contracts during that time, or whether roles will be revoked or kept. Today the DAO Safe owns the locker, the coordinator and the bridge contracts read, but future authority is unknown.
- Benefit: a small Safe can pause the bridges and respond to an exploit within minutes, which matters during an orderly exit.
Data Completeness
Available: on-chain control state for Ethereum, Polygon, Arbitrum, Base and Avalanche (Ethereum read at block 26289853), the sunset notice, a Snapshot capture of AIP-78 (unknown reliability), audit digests and claims, and a community overview of Aura governance. The Ethereum read is partial (some seed addresses were truncated by a time budget).
Missing. Signer identities, proposal rules and the sudo-mint audit status were searched on the web without a usable result:
- Signer identities for every Safe, so outside-team signers cannot be shown. A search found no page naming the DAO or Sudo Safe signers.
- Measured vote concentration and quorum for the main Snapshot space, and independent confirmation of the AIP-78 tally and turnout.
- Proposal rules, voting strategy, voting delay and period, and any charter binding the Safe to votes. A search found only the community overview.
- Whether the Safe executes Snapshot outcomes at all today. The official notice shows a vote and the on-chain read shows Safe powers, but no official or on-chain source links the two; the link rests on a 2023 community overview.
- Fee and parameter bounds beyond the Booster
MaxFeesvalue. - The holder of the current Booster's
setFeesand related functions (0xd0521c06…). - Gnosis and Optimism control state.
- Audit status of the sidechain sudo-mint item. A search found no source stating whether it was fixed, acknowledged or disputed, and none on whether the upgrade fixes were deployed beyond the on-chain ownership state above. One further search on the Snapshot space settings failed and was not repeated.
Score: 3.7/10
Rubric security-score-v2.5; data completeness 0.75
- Admin control
- 5.50 points
- Change delay
- 1.00 points
- Decision process
- 3.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Admin control | L2 | supported | 0.45 | 5.50 |
| Change delay | L0 | supported | 0.30 | 1.00 |
| Decision process | — | not disclosed | 0.25 | 3.50 |
42 citations · click markers to view evidence
Cross-Analysis
Cross-Risk Analysis: Aura
Aura is a closing protocol, so the main risks for a holder are about getting value out, not about yield. Pools are withdrawal-only, and AURA and auraBAL holders redeem in stages through 10 May 2027. The three biggest risks combine several domains: the long auraBAL lock-up tied to Balancer's market, a Balancer change dated 30 October (two days away) that only community-tier sources report, which could remove interface and routing support for LP exits, and a redemption and admin path with no audit in the sources and no named operators. The Balancer wind-down and its dates come only from community-tier and unverified publications, so they are unconfirmed. No exploit of Aura is recorded. Confidence is moderate, because deployed code was never compared with any audited version and key on-chain and vote facts are missing.
Top Risks for a Holder
1. auraBAL is locked until 10 May 2027, and its value depends on Balancer's market. (economic + overview + social)
- What can go wrong: auraBAL redeems only at Stage 3, for the remaining 90% of the 80/20 BAL/WETH pool token (BPT) . The wind-down proposal (source reliability: unverified publisher) describes auraBAL as having "almost no counter-trade liquidity" , so the sources neither show nor rule out an early market exit; DEX depth was not measured. The BPT paid out tracks Balancer's market . The same proposal says that without veBAL "Aura's core value proposition ceases to function" .
- Magnitude: the snapshot TVL of about USD 9.7M, all on Ethereum, is context, not a loss ceiling . The sources do not show that it includes every auraBAL claim, and the auraBAL share of it is not measured. The official page publishes the payout: auraBAL redeems for the remaining 90% of the 80/20 BAL/WETH BPT . Assessment: holders carry the BAL/WETH price move over the roughly six months to Stage 3, plus any early-exit cost. The sources do not measure how final proceeds compare with each holder's original contribution, so a loss of principal is not ruled out. The proposal (unverified publisher) calls the Stage 3 payout "a significant recovery for holders" . A search for an insurance, reserve or compensation fund found none ; that is an absence in captured pages, not proof that none exists.
- Likelihood: high that auraBAL holders wait until the Stage 3 date (the official schedule fixes it); medium that BAL price moves hurt them materially. Basis: a volatile asset held through a lock of about six months (a general market fact), plus the proposal's own account of BAL emission cuts and falling token prices, which comes from an unverified publisher alone . The missing fact is the auraBAL share of TVL and any DEX depth (the depth measurement found no candidate token).
2. Aura LPs exit into Balancer pools, and a community-reported Balancer change on 30 October is two days away. (economic + code + social)
- What can go wrong: Aura LP withdrawals return Balancer pool tokens, and users must exit through Balancer pools . A Balancer forum proposal, BIP-928 (source reliability: community publication), sets 30 October 2026 as the date pools that can be paused move to withdrawals only. No official page confirms the vote result or execution. The proposal's bug bounty ends the same day, and its interface and routing are treated as discontinued afterwards . Aura's own bug-bounty status is unknown, because its security page did not resolve . If the Balancer schedule holds, no bounty is confirmed on either side after 30 October.
- Magnitude: up to the Ethereum-held TVL of USD 9.7M for LPs who delay. Balancer's own text says withdrawals stay open, so the likelier loss is stranded or worse-priced exits, not frozen funds .
- Likelihood: medium for a worse exit or tooling problem, low for permanent loss. Basis: the vote result and schedule rest on community-tier and unverified sources, and per-pool treatment was still unpublished. An account of a November 2025 Balancer exploit on Composable Stable Pools comes from an unverified publisher . It is unconfirmed here, and it is not used as a basis for this band. Aura's loss from it is not stated. A holder who does not want to rely on Balancer's interface and routing support after that date can exit before it, since the proposal also says withdrawals stay open.
3. The redemption and admin path has no audit in the sources and no named operators. (code + governance + social + overview)
- What can go wrong: no audit covers the redemption or sunset contracts . The sources do not say which accounts administer the treasury and redemption contracts through May 2027 . The DAO Safe (4 of 7 signers) makes many owner calls with a measured delay of 0 seconds . No individual team member is named . A holder who redeems relies on contracts with no audit in the sources, run by accounts the sources do not identify. Lookalike phishing domains also describe redemption limits .
- Magnitude: the claims exposed are those paid through the redemption path: treasury assets for AURA redeemers and, for auraBAL, 90% of the BPT position . The size of the treasury and of the BPT position is not measured, so no dollar ceiling is given. The AURA market cap of about USD 1.4M and the official rAURA range of USD 0.008705-0.0215 per token, which comes with no method , are snapshot context only and do not cap what the claims are worth.
- Likelihood: low to medium for a redemption-contract or key failure. Assessment: Stage 1 redemption opened on 13 May 2026, about five and a half months before the analysis date, and the hacks dataset (collected 23 October 2026) holds no record for Aura , which argues for low. The redemption contracts are new, have no audit in the sources and have no identified owners, which argues against going lower; the dataset's coverage of such contracts is not shown. The missing facts are the redemption contract addresses, their owners and any audit. Medium for a user losing funds to a fake site or a mistaken claim. Basis: lookalike domains already exist, but how many users they reach and the frequency of losses are not measured.
Combined Attack Vectors
Vector 1: Shutdown locker plus the reward-pool penalty path
- The 2022 Code4rena contest finding M-18 (source reliability: unverified publisher; the report still counts as evidence) says that if
AuraLockeris shut down,lock()always reverts, so claimants ofAuraBalRewardPoolmust take a penalty path. The team replied that shutdown was "highly unlikely" . The repository code still shows a 20% penalty (reward * 2 / 10), though the repository is archived and deployment is not confirmed . - On-chain,
AuraLockeris now shut down at Ethereum block 26289853 . - The
auraBAL Rewardspool's reward period ended in May 2026 . Whether holders still have unclaimed AURA rewards in that pool is not shown, so the exposure is possible but unmeasured.
- Domains: code audit finding + on-chain shutdown + wind-down.
- Likelihood: medium (the trigger condition is met; the live code behaviour is unverified). Impact: minor (a penalty on unclaimed reward amounts, not principal). No other worker report links these two facts.
Vector 2: Key compromise across shared signers and bridge rescue
- One address (
0x30019eb1…) sits on the DAO, Sudo, Admin and Warden-scheduler Safes and also holds an instantrescueTokenrole . The same EOA owns theKeeperMulticall3that owns each of the eight Ethereum-side bridge receivers . What those receivers' owner functions can change was not read. - An attacker who takes that key gets instant
rescueTokencalls on reward handlers and owner calls on the receivers. Four DAO Safe keys would add instantpause(),unpause()and other owner calls on the bridge contracts. The 8-of-13 Sudo Safe can callrescueon both bridge proxies, and a sidechain review by an individual reviewer (Hephyrius, undated; source reliability: unverified publisher) rates the ability of a malicious sudo to "mint infinite tokens" as Low-Medium, with its fix status unknown . - The audit set cannot show whether the live bridge code matches any audited version: all 185 verified files are in the not-compared bucket . Signers are unnamed, and no timelock was found.
- Domains: governance (keys, no delay) + code (unlinked audits, sudo-mint finding) + social (pseudonymous team).
- Limits: all counted TVL is on Ethereum, the Booster and Locker are shut down, and the locker's
recoverERC20excludes AURA and registered rewards. The reach into the auraBAL vault viavaultExecuteis not read. - Likelihood: low (needs 4 of 7 or 8 of 13 signers, or one specific key). Impact: moderate (bridge-held tokens and in-transit rewards; severe only if
vaultExecutereaches vault deposits).
Vector 3: Balancer-side deterioration hits a position with no confirmed market exit
- Aura's wind-down proposal (source reliability: unverified publisher) says BAL emissions were cut, token prices fell and depositors pulled back . An unverified publisher reports a November 2025 Balancer exploit on Composable Stable Pools, which is unconfirmed here , and a Balancer post-mortem (source reliability: community publication) notes Aura assessed its exposure but states no Aura loss .
- auraBAL cannot be redeemed before Stage 3, and the proposal describes almost no counter-trade liquidity for it. LP exits return BPT that still has to be withdrawn or sold through Balancer pools .
- If the community-reported Balancer schedule holds, Balancer's bounty and interface support end on 30 October. Aura's own bounty status is unknown . Neither a proposed end to Balancer's bounty and interface support nor an unknown Aura bounty shows that other controls disappear; the sources do not give a full inventory of remaining controls.
- Domains: economic (Balancer dependency, thin liquidity) + governance (the Aura Safe is the named control in the sources; the Balancer vote result is unconfirmed) + code (no confirmed bounty).
- Likelihood: medium. Basis: the price-decline and emission-cut accounts come from the unverified proposal alone, and the Balancer exploit is documented in community-tier and unverified sources, so none of them carries the band by itself. The band rests on the exposure length for a volatile asset (a general market fact). The Balancer schedule is unconfirmed, and no Aura loss from the exploit is stated.
- Impact: moderate, severe only for large auraBAL positions. Basis: Balancer says withdrawals stay open, so the likelier outcome is price loss and awkward exits, not frozen funds.
Systemic Risk Factors
- Balancer is the load-bearing dependency but is absent from the structured graph. The dependency graph shows only an Ethereum chain edge at share 1.0 . It has no Balancer edge and no warnings, and its composition data is too large to assess fully from the available data . The overview shows Aura depends on Balancer's vault, gauges, veBAL and BPT . This belongs to the dependency-risk factor of the economic owner for reassessment (the economic report grades it L2). Circular backing and downstream holders cannot be checked, because the graph holds no protocol-level backing data.
- Chain and bridge: all TVL is on Ethereum, so L2 sequencer risk is small. LayerZero and native bridges link sidechains that report zero TVL; residual bridge exposure is not measured, and Gnosis and Optimism were not read.
- Macro/correlation: BAL price, Balancer pool health and AURA price move together. The treasury being paid out is mostly ETH and AAVE (per the proposal), so AURA redeemers carry ETH and AAVE price risk and auraBAL holders carry BAL/WETH price risk.
- Single-operator drift: the same few signers sit on several Safes, so one operational failure can hit several controls.
Risk Amplifiers
- Wind-down + direct owner calls: at the measured block the DAO Safe's direct owner calls have a delay of 0 seconds, and no plan to renounce or timelock them after the wind-down was found . Future authority through May 2027 is unknown. Only forced actions on the Booster owner carry a 30-day delay. The last visible commit on the main contracts repository is dated 10 April 2026 ; later development activity and team size are not shown in the sources.
- Unlinked audits + shared keys: no audit version is matched to deployed code, and the keys that can act on the deployed code are partly shared.
- No loss cover found + thin liquidity described: no compensation fund was found, and the proposal describes almost no counter-trade liquidity for auraBAL; DEX depth was not measured.
- Pseudonymous team + limited legal wrapper: the named entity covers only the website, and the proposal lists sunsetting the Aura Foundation and subsidiaries (unverified publisher), so recourse after May 2027 is unclear.
- Reliance on weak captures: the AIP-78 tally (8 votes, unverified publisher) and the Balancer vote both rest on weak sources, so decisions are hard to verify.
Risk Mitigators
- Shutdown flags on-chain: the Booster,
AuraLockerand measured sidechain boosters reportisShutdown: true, which limits new actions.BoosterOwneris sealed with a 30-day forced-action delay that has not been started . - Review depth for the original system: three Halborn engagements, one Zellic assessment (no findings) and a public contest; both high findings are fixed, and 21 of 28 medium findings are fixed. Two of the 28 are medium (estimated): an AI estimate, because the Spagnuolo governance upgrades report states no rating. Several of these reports are hosted on publications not verified as auditor-controlled, which lowers source confidence but not coverage credit.
- Pause capacity: a 4-of-7 Safe can pause the bridge proxies at once, which helps in an incident.
- No recorded exploit: the hacks dataset is empty and the only hack claim is an uncorroborated post.
- Stated withdrawal promise: official site says LP, vault and locker positions stay withdrawable at any time .
- Incident plan: a Chainalysis response plan is documented, but it is not funded cover.
Contradictions & Inconsistencies
- Status of the Hephyrius booster-shutdown findings (undated individual-reviewer reports; source reliability: unverified publisher): the code report says both findings are "medium and fixed" . The governance report says the digest lists them as "both acknowledged" and rated low to medium . The reports do not reconcile this, and the governance report adds that it could not confirm the stash-side fix is deployed.
sudoaddress on the bridge: the overview says it is not confirmed that theAuraProxyOFTsudo address holdsrescue()power . The governance report reads an 8-of-13 Sudo Safe callingrescueon both bridge contracts. The governance reading is the more specific, on-chain-based one; the overview's gap is partly closed.- Incident record vs resilience: the social report grades the incident record at the top level (L4) from an empty hacks dataset, while its resilience factor is "not disclosed" and it states there is no complete outage or pause record. An empty dataset does not show the absence of incidents; the Balancer-exploit exposure of Aura is also not quantified.
- Team identity credit vs governance opacity: the social report scores team identity L3 because a legal entity is named, but it also says no individual is named and the entity covers only the website. The governance report says signers cannot be shown to be outside the team. The scores were not changed here; a holder should treat accountability as thin.
- rAURA price range: the upper bound differs between captures of the official page: $0.0215 in the overview and economic reports, $0.0214 in the social report, and other captures show $0.0217 and $0.0223 . The figure moves with redemption participation, so no single number is a stable value.
- Sunset timing wording: the governance and overview reports quote the schedule as it appeared on the page ("in 6 months"); that phrase is stale relative to the 28 October 2026 analysis date.
- Yield sources: the Exponential snapshot (captured 9 March 2026) rates the product 7.0 "Good" and "No death spiral concerns" ; the wind-down proposal describes the opposite. The economic report treats the snapshot as historical.
Overall Risk Profile
Aura's composite risk is moderate for a closing protocol with a hard dependency on Balancer, whose own wind-down is reported by community-tier sources only. The scores of the four domains (code 5.9, economic 4.5, governance 3.7, social 6.2 with track record 8.6) are not recomputed here. The most likely failure mode is value impairment, not an exploit: auraBAL holders wait until 10 May 2027 with no confirmed market exit, and LPs and redeemers exit into Balancer pools, where the community-reported 30 October change could reduce interface and routing support. The worst case is a compromise of the shared admin keys or a fault in redemption contracts that have no audit in the sources, during the wind-down, with 0-second owner calls at the measured block, no loss fund found and no confirmed bounty. No peer data is in the sources, so no comparison with Convex-style booster peers is made.
- TVL
- 9.71 USD m
- AURA market cap
- 1.42 USD m
Data Gaps
- A protocol source page is unavailable because it is not published.
- Collection of GitHub repositories stopped at the collection limit; some material may be missing.
- Current deployment addresses could not be confirmed for some audited contracts.
- Source summaries cover only the first 80000 characters of 2 document(s); later content was not assessed.
- Redemption contract addresses, their owners and any audit are missing, so Top Risk 3 cannot be sized or ruled out.
- The auraBAL, vault and LP split of the USD 9.7M TVL is missing; asset checks and the DEX depth measurement found no candidate token.
- The Balancer BIP-928 vote result and the 30 October execution are not confirmed by an official page, and the per-pool treatment is unpublished.
- Signer identities, the link between Snapshot votes and the Safe, and the owner of the current Booster's fee functions (
0xd0521c06…) are unknown. - Gnosis and Optimism state was not read, and part of the Ethereum read is partial, so Vector 2's bridge reach is not fully assessed.
- Deployed code was not compared with any audited revision, so every audit-based mitigation is unconfirmed for the live contracts. The status of the sidechain sudo-mint item is unknown.
- The dependency graph has no Balancer edge and no composition data, so circular backing and downstream-holder warnings could not be assessed.
51 citations · click markers to view evidence