Liquity V1
Tier 2Security Analysis Report
TVL
$129.37M
Sections
7
report sections
Citations
466
evidence references
Version
v2
11/16/2026
Good
Safety Score · out of 10
Executive Summary
Executive Summary: Liquity V1
The Liquity V1 core contracts have no owner and no admin key. Nobody can change them, pause them or take user funds through a key. That removes insider and upgrade risk. It also means nobody can fix a bug or a failed price feed. Collateral is far above debt today, but the audited code is not shown to match the deployed code, and the bug bounty has ended.
Safety Rating
Overall Score: 7.2/10 Dimension scores: Code security 6.0 · Economic security 7.4 · Team transparency 7.5 · Governance 9.5 · Track record 6.9
- Code security
- 6.0 points
- Economic security
- 7.4 points
- Team transparency
- 7.5 points
- Governance
- 9.5 points
- Track record
- 6.9 points
The rating weighs very strong governance and a sound economic design against thin audit-to-deployment evidence, an ended bug bounty and thin exit liquidity. Detailed reasoning is in the dimension table below.
Scores by Dimension
| Dimension | Key Finding |
|---|---|
| Code Security | Two firms audited the core in early 2021, but the audited versions are not linked to the deployed code and the bug bounty has ended. |
| Economic Security | Collateral is far above debt and oracle design is strong. Exit liquidity is thin and holdings are concentrated. |
| Team Transparency | The team and its Swiss company are named, and audits and addresses are published. |
| Governance | There is no governance. The core contracts have no owner, so no key can upgrade, pause or drain them. |
| Track Record | V1 has run since April 2021 with no exploit loss in the sources, but it had a disclosed oracle flaw in 2022. |
Track Record detail:
- The 2022 Tellor oracle flaw has an official post-mortem dated 22 November 2022. The team says it was exploited on the ETHW fork (a separate chain) and that mainnet never switched to Tellor. This is the team's own account.
- The February 2025 Stability Pool issue belongs to Liquity V2, another product of the same team. The pages checked give no root cause or loss accounting, and they do not name V1 contracts as affected.
- DefiLlama and DeFiHackLabs hold no records for this protocol. That does not mean no incidents occurred; the two events above are documented separately.
What This Protocol Does
Liquity V1 is a set of unchangeable Ethereum smart contracts. Users lock ETH and borrow the USD-pegged stablecoin LUSD with no interest charge. The products and tokens are LUSD, the LQTY token and the contracts for borrowing, liquidation, redemption and fee staking. DefiLlama lists Ethereum as the only chain. Liquity V2 (BOLD) and the Chicken Bonds contracts are other products and are not assessed here.
Key Strengths
- No admin key (all V1 core contracts). Core contracts have no admin key, and no other account was found that can move user funds, mint LUSD or pause the system. Limit: 115 of 275 on-chain privilege rows were unresolved and 96 unclear, so the absence of other powers rests on the docs and the audit, not on a full on-chain proof.
- Wide collateral buffer (Troves, LUSD). The system ratio is about 506% against a 110% liquidation line, based on the on-chain read at block 26411621 (2026-11-14) and the Chainlink price at that block.
- Two-source price feed (all products that rely on the ETH price). Chainlink is the primary feed and Tellor the fallback. Both have staleness checks, and Chainlink also has a 50% deviation check.
- Review depth (V1 core contracts only). Two independent firms examined V1, and Trail of Bits ran three engagements in early 2021. These audits do not cover the bridged LUSD copies on other chains.
- Stress record (Stability Pool, as the team describes it). On 19 May 2021 the pool absorbed about 93.5M LUSD of defaulted debt. This is the team's own account.
Key Risks
- 1. Unfixable code or oracle failure (all products: Troves, Stability Pool, LUSD, LQTY staking). A hidden bug in the 2021 contracts, or both Chainlink and Tellor failing together, hits a system with no pause and no upgrade path. Magnitude: up to the full collateral base, about 71,473 ETH or about $136M at the Chainlink price on 2026-11-14. Staked LQTY ($7.77M) is a separate figure not included in that amount. Likelihood: low. The fact that would change the band is a comparison of the deployed code with the audited commits.
- 2. A fast ETH drop drains the Stability Pool (depositors and leveraged Trove owners; LUSD). The pool holds about 24% of system debt, and the rest would be spread across remaining Troves. Magnitude: the pool's 6.44M LUSD is the first loss layer, and system-wide insolvency is remote. Only 64 Troves are open, so a few large positions drive system health. Likelihood: low for a net loss to depositors or any redistribution, and routine for leveraged Troves in a sharp drop. Per-depositor pool concentration is not measured and would change the band. No V1 reporting covers the 2024, 2025 and October 2026 market drops.
- 3. Exit squeeze for LUSD and LQTY holders (LUSD, bridged LUSD copies, LQTY). LUSD keeps its $1 floor through redemption, but open-market depth is thin. Redemption stops if the system ratio falls below 110%. Magnitude: LUSD supply on Ethereum is about $26.7M against $12.96M in the Curve pool and $28K in the largest Uniswap v3 pool (source reliability: unverified publisher). The Curve figure is pool liquidity, not depth at 2% price impact, and mainnet LUSD depth at 2% and 0.5% is not measured. Bridged copies on Optimism, Base and Arbitrum have supply of $79K, $8.9K and $70K with depth of $3,956, $447 and zero, and losses on the copies stay local to them. Likelihood: medium for a price dislocation, low for principal loss. LQTY has no measured depth on Ethereum, and a $100K probe moves its price 18.32%.
- 4. Audits not linked to the deployed code, and no active bounty (V1 core). The available material does not show which audit fixes are live, and the sources show no active paid disclosure channel. This comparison was not performed; it is a gap in the evidence, not a measured mismatch.
- 5. Unassessed control over the Chainlink-side aggregator (price feed). The
PriceFeedread names an aggregator owned by a separate Safe (multisig wallet) with an enabled module. This Safe sits outside the Liquity core contracts, and its powers over the aggregator are not assessed. Who can act through that module is not established. The cross-analysis rates this likelihood low, on a weak basis.
Critical Findings
Documented versus on-chain, for the points that matter most:
- The dev README (publication date not established) describes the Liquity admin address as a regular Ethereum address controlled by Liquity AG. On Ethereum at block 26411621 (2026-11-14),
LQTYTokenrecords a Safe with 8 owners and a threshold of 3. Because the README date is not established, the cross-analysis leaves this conflict unresolved. The key can only move the team's own LQTY. - The docs say fees cannot fall below 0.5%. On Ethereum at block 26411621 (2026-11-14),
BORROWING_FEE_FLOORreads 5000000000000000 with 1e18 precision, which gives 0.5%, and the decayed borrowing rate was about 0.54%. - The docs describe LQTY community issuance as a halving curve over 32 million LQTY. On-chain at block 26411621 (2026-11-14), 31,346,304 LQTY of that 32,000,000 is issued, so emissions are nearly exhausted.
Audit findings (V1 core contracts; four audits: three Trail of Bits and one Coinspect, all dated January to March 2021). Counts are issue groups: total / fixed per the auditor / not marked fixed.
| Severity | Total groups | Fixed per the auditor | Not marked fixed |
|---|---|---|---|
| Critical | 0 | 0 | 0 |
| High | 4 | 2 | 2 |
| Medium | 7 | 3 | 4 |
These counts sort audit statuses. They are not a count of live vulnerabilities.
Audit-to-deployment link: the deployed code is not compared with any audited version. The map gives the reason as an audit scope that could not be determined from the available material. That is an evidence gap, not proof of changed or unaudited code.
Groups not marked fixed:
- High, two groups: Trail of Bits, 2021-03-05, delegatecall issues in the
DSProxyproxy scripts. The audit says these could result in the proxy being self-destructed and losing funds. Audit status: not stated. Our check: unavailable, because the finding detail was not located. No deployed address is attributed toDSProxy. These concern peripheral scripts, not the core contracts. - Medium: ERC20 transfer restriction (Trail of Bits, 2021-01-13). Audit status: acknowledged, recorded as risk accepted. Our check: the described code is unchanged in the verified
LQTYToken. It is a deliberate choice the team accepted. - Medium: Stability Pool depositors front-running liquidations (Trail of Bits, 2021-01-13; listed as a known issue). Audit status: not stated. Our check: unavailable, because the audit text was not located. The medium severity is estimated by us because the audit states no rating.
- Medium: fee inflation when few LQTY are staked (Coinspect, March 2021). Audit status: open. Our check: the described code is unchanged in the verified
TroveManager, with moderate confidence. This is a limited code signal, not a confirmed vulnerability, and no exploit was tested. - Medium: liquidation incentives that ignore system health (Coinspect, March 2021). Audit status: open. Our check: the described code is unchanged in the verified
TroveManager, with moderate confidence.
The front-running group and the two proxy-script groups have no code check, so their live status is not confirmed. The map could not tell whether the deployed code contains the five fixes the auditors mark as made, because no fix review is available.
Other material issues from the reports:
- No component of the live V1 code can be patched, so any open finding stays open.
- Solidity 0.6.11 carries 13 listed compiler bugs, four rated medium. No code location meeting a bug's conditions was checked, so none is a finding.
Recommendation
Liquity V1 suits users who accept that a bug or oracle failure cannot be fixed and who want a no-admin, no-interest borrowing venue. Examples are ETH holders who monitor their collateral ratio and LUSD holders who understand the Stability Pool. The governance and key-compromise risk on the core contracts is very low , and the collateral buffer is wide today. Users with low risk tolerance should weigh the 2021 code, the missing link between audits and deployed code, the ended bounty and the thin exit markets for LUSD and LQTY. Leveraged borrowers face fast liquidation in a sharp ETH drop, and Stability Pool depositors can lose money in a flash crash. Do not treat the archived 2025 DefiSafety snapshot as current evidence.
Report Sections
- Protocol Overview
- Code Security
- Economic Security
- Social & Team
- Governance
- Cross-Risk Analysis
64 citations · click markers to view evidence
Protocol Overview
Protocol Overview: Liquity V1
Liquity V1 is an immutable (unchangeable) set of Ethereum smart contracts. Users lock ETH and borrow the USD-pegged stablecoin LUSD with no interest charge. The product set is LUSD, the LQTY token, and the contracts for borrowing, liquidation, redemption and fee staking. On-chain reads on 2026-11-14 show the core contracts have no owner and the docs say there is no admin key, so the main risks come from code, the ETH price oracle and market conditions rather than from a controlling party. Evidence is strong for mechanics and control state; it is weaker for current yields and for any audit of the live code.
What It Does
Liquity V1 lets people borrow a stablecoin against ETH without paying interest. A borrower locks ETH in a position called a Trove and receives LUSD, a token designed to be worth $1. The docs describe it as "a decentralized borrowing protocol that allows you to draw interest-free loans against Ether used as collateral." Loans must keep a collateral ratio (the value of the ETH divided by the LUSD debt) of at least 110%.
- Category: CDP (collateralized debt position) stablecoin borrowing. DefiLlama lists the category as
CDPand the only chain as Ethereum. - Products and tokens in scope: the borrowing system, the Stability Pool, LUSD (stablecoin) and LQTY (fee-capturing token).
- Main uses, per the docs: borrow LUSD against ETH, deposit LUSD in the Stability Pool for rewards, stake LQTY for fee revenue, and redeem LUSD for ETH at face value.
- Target users: ETH holders who want liquidity without interest or governance exposure, LUSD holders who earn by backstopping liquidations, LQTY holders who want fee income, and arbitrageurs who redeem LUSD when it trades below $1.
- No official frontend: the core team does not run a web interface. Users reach the contracts through third-party frontends.
- Related products, not in this record: Liquity V2 (BOLD) is a separate DefiLlama family member and is not assessed here. The Chicken Bonds contracts appear in the address data as a separate team product that uses LUSD; they are context only.
How It Works
The system runs as a loop. Borrowers mint LUSD against ETH, the Stability Pool absorbs bad positions, and redeemers keep the price near $1. All logic runs on-chain: the dev README says "The system has no admin key or human governance."
User flow:
- Borrow. A borrower calls
BorrowerOperationsto open a Trove with ETH collateral and a requested LUSD amount. A one-off borrowing fee is added to the debt, and a 200 LUSD liquidation reserve is set aside. The reserve sits inGasPool, and the fee goes toLQTYStaking. - Back the system (optional). A LUSD holder deposits into the
StabilityPool. The depositor earns ETH from liquidations and LQTY rewards.CommunityIssuancereleases the LQTY rewards over time. - Liquidate. Anyone can liquidate a Trove once its ratio falls below 110%, through
TroveManager. The Stability Pool burns LUSD equal to the Trove's debt and receives all its collateral. The caller earns "200 LUSD + 0.5% of Trove's collateral" as gas compensation. - Redeem. Any LUSD holder can swap LUSD for ETH at $1 face value.
TroveManagerruns redemptions and sends the fee toLQTYStaking. Redemption is separate from repaying a loan; a borrower repays by adjusting the Trove. - Stake LQTY. An LQTY holder deposits into
LQTYStakingand earns a pro-rata share of borrowing and redemption fees in LUSD and ETH. There is no lock-up.
- (actor)
- (contract)
- (contract)
- (contract)
- (contract)
- (contract)
- (actor)
- (actor)
- (actor)
- (actor)
- step 1flow
- step 1flow
- step 1flow
- step 2flow
- step 3flow
- step 3flow
- step 4flow
- step 4flow
- step 5flow
Core mechanism
The core mechanism is overcollateralized borrowing with a fixed 110% minimum ratio. Three further rules shape it.
- Stability Pool first, redistribution second. If the pool is empty, a liquidated Trove's debt and collateral are spread over all other Troves in proportion to collateral.
- Redemptions form a price floor. Redeeming 1 LUSD for $1 of ETH, together with the 110% minimum ratio, creates a floor and a ceiling for the LUSD price.
- Recovery Mode. If the system-wide ratio falls below 150%, liquidation rules relax and the system blocks borrowing that would lower that ratio.
Key smart contracts and roles
| Contract | Role |
|---|---|
BorrowerOperations | Opens, tops up and closes Troves; issues and repays LUSD |
TroveManager | Liquidations, redemptions and each Trove's recorded collateral and debt; holds no value itself |
StabilityPool | Holds LUSD deposits and the ETH gains from liquidations |
ActivePool / DefaultPool | ETH and debt of active Troves; ETH and debt of liquidated Troves awaiting redistribution |
CollSurplusPool | Holds surplus ETH from fully redeemed Troves, returned to the borrower |
GasPool | Holds the LUSD liquidation reserves |
LUSDToken | The stablecoin; mints, burns and transfers LUSD |
SortedTroves | Linked list of Troves ordered by collateral ratio |
PriceFeed | Supplies the ETH:USD price |
HintHelpers | Read-only helper that computes hints for borrower operations and redemptions |
LQTYToken, LQTYStaking, CommunityIssuance | The LQTY token (100 million hard cap), fee staking, and issuance to Stability Pool depositors |
LockupContractFactory, Unipool | Vesting lockups and the LUSD:ETH Uniswap LP reward contract (peripheral) |
Addresses come from the official docs and were matched by Sourcify for several core contracts, for example BorrowerOperations at 0x24179CD81c9e782A4096035f7eC97fB8B783e007. Other core addresses (TroveManager, StabilityPool, ActivePool, LUSDToken, LQTYStaking) are in the official address table.
Yield Sources
Yield for Liquity participants comes from liquidation gains and protocol fees, plus LQTY emissions that are scheduled to taper. Borrowers pay no interest; they pay a one-off fee.
- Stability Pool depositors earn ETH from liquidated Troves and LQTY rewards. Liquidations usually occur just below a 110% ratio, so depositors expect to receive more collateral value than the debt they absorb. This is a loss risk if a Trove is liquidated below 100%, as in a flash crash or oracle failure.
- LQTY stakers earn LUSD from borrowing fees and ETH from redemption fees. All protocol revenue goes to stakers.
- Fee size. Fees rise with redemption volume and decay over time. They have a 0.5% floor in normal mode, and the borrowing fee is capped at 5%. On-chain on Ethereum at block 26411621 (2026-11-14),
BORROWING_FEE_FLOORis 5000000000000000 andDECIMAL_PRECISIONis 1e18, which gives 0.5%. The decayed borrowing rate was 5419508501089961, about 0.54%. - Frontend operators earn a share of the LQTY that their users' Stability Pool deposits generate, set by a "kickback rate" from 0% to 100%.
- Measured fees. DefiLlama records 30-day fees of 264,128 and revenue of 144,497, and 365-day fees of 838,387 and revenue of 243,062. The currency unit is not stated in the file, so treat these as USD only if DefiLlama's page confirms it.
Sustainability. Fee and liquidation yield depends on borrowing, redemption and liquidation activity, not on emissions. LQTY emissions follow a halving curve, 32,000,000 * (1–0.5^year), from the 32 million LQTY community supply. Assessment: the LQTY token contract's deployment start time is 1617611590, about 5.6 years before the data date. On that curve about 98% of the 32 million was issued by now, so LQTY emissions are close to exhausted. The remaining yield is mainly fees and liquidation gains. No historical APY series is in the sources.
Dependencies & Integrations
Liquity V1 depends on one price oracle pair and on ETH itself.
- Oracles:
PriceFeeduses Chainlink ETH:USD as primary and Tellor ETH:USD as fallback. It switches to Tellor if Chainlink is stale for more than 4 hours, reverts, returns invalid data, or moves more than 50% between two updates. Tellor prices are at least 15 minutes old by design, to allow a dispute period. - Collateral: ETH only. DefiLlama's dependency data lists WETH as the backing asset, Chainlink as the oracle and Ethereum as the chain.
- Chains: DefiLlama lists a single chain, Ethereum, for TVL. The workspace data has no inactive chain list, so no earlier deployments on other chains are assessed.
- Frontends: independent third-party operators host the web interfaces. The docs list them but the core team does not run one.
- Other integrations seen in address data: the Chicken Bonds product (bLUSD) on LUSD and a Uniswap V2 LUSD:ETH pair used for LQTY rewards. These are not in the V1 DefiLlama record.
TVL & Adoption
Liquity V1 holds about $129 million on Ethereum in the latest DefiLlama snapshot, down sharply from its 2021 peak. DefiLlama (collected 2026-11-15) reports TVL of 129,366,537.33 USD, all on Ethereum, plus 7,768,314.81 USD of staked LQTY counted separately.
- Trend. Monthly samples show a peak of about $3.24 billion in May 2021, $332 million in October 2025, $232 million in January 2026, a low of $116 million in July 2026 and $157 million in November 2026. The 30-day change to 2026-11-15 is -50,320,632 USD, of which -49,784,106 USD is price effect. Most of the recent drop therefore comes from the ETH price, not from withdrawals.
- Current system size (on-chain, Ethereum block 26411621, 2026-11-14).
TroveManagerreports 64 Trove owners, collateral of 71472954192106610397506 (about 71,473 ETH) and debt of 26877642789150068487730665 (about 26.88 million LUSD).StabilityPoolholds 6439791742129871093942041 LUSD, about 6.44 million. - LQTY holders. The
LQTYStakingcontract holds 53.13% of LQTY supply.
- May 2021
- 3242 USD m
- Oct 2025
- 332 USD m
- Jan 2026
- 232 USD m
- Jul 2026
- 116 USD m
- Nov 2026
- 157 USD m
Whose funds the TVL counts: ETH locked by Trove owners, plus staked LQTY in the separate staking figure. Notable integrations are third-party frontends and the Chicken Bonds contracts; no partnership list appears in the sources.
Technical Architecture
The core contracts are immutable and have no administrator. For who can upgrade or pause core contracts, see Governance.
- Upgradability. The docs state "Liquity has no admin key, and nobody can alter the rules of the system in any way. The smart contract code is completely immutable." No standard proxy implementation or admin slot was recorded for the core contracts at the measured block, which is consistent with immutability.
- Ownership. On Ethereum at block 26411621 (2026-11-14), the
ownerofTroveManager,StabilityPool,ActivePool,DefaultPool,BorrowerOperations,PriceFeed,SortedTroves,CommunityIssuance,HintHelpers,CollSurplusPool,LockupContractFactoryandLQTYStakingis the zero address, meaning ownership is renounced.LUSDToken,GasPoolandMultiTroveGetterhave no readableowner, so their control was not measured that way. - Parameters. The 110% minimum ratio, 150% critical ratio, 200 LUSD liquidation reserve and fee floors are fixed at deployment. On-chain they read
MCR1100000000000000000 andCCR1500000000000000000 with 1e18 precision, i.e. 110% and 150%. - Emergency mechanisms. The available sources document no pause or shutdown function. Recovery Mode is the automatic response to a low system-wide ratio.
- Liquity admin address.
LQTYTokenrecords amultisigAddressthat is a Safe wallet with 8 owners and a threshold of 3 at block 26411621. The dev README says this address "has no extra privileges and does not retain any control over the Liquity protocol once deployed." - Oracle.
PriceFeedfixes Chainlink and Tellor as sources at deployment. Their own upgrade controls are outside Liquity's contracts and were not assessed here. - Audits (of the original 2021 code, not a current-deployment comparison). Trail of Bits reviewed in January 2021 and March 2021, with a separate March 2021 review of proxy contracts, and Coinspect reviewed in March 2021. Whether the deployed bytecode matches the audited commits was not compared. The Coinspect scope lists
HintHelpers,ActivePool,PriceFeed,GasPool,StabilityPoolandBorrowerOperationsamong its items. The March 2021 proxy review covers the peripheralDSProxycontracts, not the core contracts. The V1 bug bounty was discontinued after four years.
- (oracle)
- (oracle)
- (contract)
- (contract)
- (contract)
- (actor)
- dependency
- dependency
- flow
- flow
Key Risks (Protocol-Level)
The main risks are an ETH crash that outruns liquidations, an oracle failure, and the age of the code. There is no admin key to compromise, but there is also nobody who can fix a bug.
- Where funds sit. Collateral ETH sits in
ActivePoolandDefaultPool, LUSD deposits and liquidation ETH inStabilityPool, and reserves inGasPool. Only the contract rules move it. - Immutability. A bug in deployed code cannot be patched. The docs themselves say "a hack or a bug that results in losses for the users can never be fully excluded." Assessment: since the code is 2021-era and the bug bounty has ended, the likelihood of a new finding is low but impact would be high, and no one holds the power to respond.
- Oracle risk. A failure of both Chainlink and Tellor, or a bad price, could trigger wrong liquidations. The docs note that a liquidation below 100% in a flash crash or oracle failure causes depositor losses.
- ETH price and Stability Pool depth. Collateral is ETH only. The Stability Pool held about 6.44 million LUSD against about 26.88 million LUSD debt at block 26411621, so a fast drop would rely on redistribution for the remainder. Redistribution spreads the loss across remaining Troves.
- Small borrower base. Only 64 Troves are open, so a few large positions drive system health.
- Peg risk. The docs state LUSD "isn't perfectly pegged to the USD, and can deviate slightly in both directions."
- Borrower redemption risk. A Trove can be partly or fully redeemed against, which reduces its debt and collateral.
- Incidents. No hack records for this protocol appear in the captured DefiLlama hacks or DeFiHackLabs data.
- Complexity. The core system has about 12 main contracts plus peripheral lockup, staking and proxy contracts. Third-party frontends add a trust surface outside the contracts.
Data Completeness
Mechanics, fees, oracle design and on-chain control state are well supported. Several items are open.
- Available: official V1 docs and dev README, four audit reports (three Trail of Bits, one Coinspect), a DefiLlama TVL and fee snapshot (collected 2026-11-15), and an on-chain read at Ethereum block 26411621 (2026-11-14).
- Missing: a historical APY series, a measurement of Stability Pool depositor returns, and a comparison of the deployed bytecode with the audited commits.
- Not assessed in depth: the
DSProxyfactory has no protocol-wide address in the sources.LQTYStakingandLUSDTokenappear in the official address table but lack an attributed address in the audit-coverage inventory. The Chicken Bonds contracts are a separate product. - Docs age: the V1 FAQ pages were marked "Last updated 2 years ago" when captured. The contract design is immutable, so parameters and mechanics are unlikely to have changed. Current fee and deposit numbers come from the on-chain read.
- No articles: the workspace has no Blockworks or other article sources.
- Currency unit of DefiLlama fee fields is not stated in the captured file.
126 citations · click markers to view evidence
Code Security
Code Security Analysis: Liquity V1
Liquity V1 core contracts on Ethereum are immutable, have no admin key, and were reviewed by two independent firms (Trail of Bits and Coinspect) in early 2021. Their source is verified and public. The bug bounty is discontinued, and the available material does not link the audited commits to the deployed bytecode. For users, review depth is good for the major version, but unresolved economic-design findings (fee inflation, liquidation incentives) and the missing bounty remain. Confidence is moderate: deployed-code comparison with audited revisions was not performed.
Data Completeness
- Read in full: the findings ledger, the audit index, the audit coverage map, the digests of the four audits, the bug-bounty documents, and the verified-source index.
- Audits in the ledger: four reports (three Trail of Bits, one Coinspect). Only specific sections were opened, to quote or verify findings; whole audits were not read.
- "Findings in ledger" means critical/high/medium issue groups, counted once per group after cross-audit deduplication, with the latest audit status. It is not a count of individual audit entries or confirmed live issues. The ledger holds 0 critical, 4 high and 7 medium groups.
- Gap: the March 5, 2021 Trail of Bits proxy review has summary-only finding entries and no extracted detail sections, so its high findings carry no code check.
- Gap: one file's line count is unavailable, and some deployed code has no criticality rating (see coverage summary).
- Gap: collection gaps were declared for four Ethereum addresses whose source is not verified at any provider (
0xe420cf281e567cb144838435bb5d6b482c102e63,0x5cca549ca706c39d68156e5e0a72ccbc95f563d0,0x884acfa4593a6fdba0a9373007e48ea9af881c42,0xf4a3fe99227f6060e4c1c62b557eee050b6483e4). They appear in the on-chain control data, not in the audited contract list. This uncollected scope is not evidence that audits are missing.
Audit Coverage
Four audit reports from two independent firms cover the V1 protocol; the audited revisions are not shown to match the deployed contracts.
| Auditor | Date | Scope (from the reports) |
|---|---|---|
| Trail of Bits (tier 1) | January 13, 2021 | Core protocol: BorrowerOperations, TroveManager, pool contracts, LUSD and LQTY tokens |
| Trail of Bits (tier 1) | March 5, 2021 | Rapid review of the Liquity proxy (DSProxy-based scripts), commit b90440d |
| Trail of Bits (tier 1) | March 25, 2021 | Changes since the December 2020 review, plus a StabilityPool review and Unipool |
| Coinspect | March 2021 | Core contracts of liquity/dev up to commit dd7f59b9; front-end out of scope |
All four reports were retrieved from the auditors' own publication hosts, so provenance is trusted. The audits are dated January to March 2021, and the Ethereum core contracts were created on April 5, 2021 , so the audits predate that deployment by weeks to months. The later bridged-token deployments on other chains are not covered by these audits. The DeFiSafety snapshot (captured May 19, 2025, curated, not current) scores the protocol .
The documents name no other audit in this workspace. Liquity V2 audits belong to another family member and are not counted here.
| Severity | Total groups | Fixed per the auditor | Not marked fixed |
|---|---|---|---|
| Critical | 0 | 0 | 0 |
| High | 4 | 2 | 2 |
| Medium | 7 | 3 | 4 |
The counts are protocol-wide for V1 and cover all four audits above. They partition audit statuses; they are not a count of live vulnerabilities.
Deployed code vs audits
Deployed code on Ethereum is Sourcify-verified, but the deployed files were not compared with any audited revision: audit-to-deployment links are not shown, so none of the 19 verified components is shown audited at its deployed version.
Map counts (copied): 22 deployments known; 22 with verified source; 13 compared (with an audited revision or with every searched commit); 13 bound to a public commit; 0 identical to an audited revision; 0 with unaudited changes; 0 still being searched; 5 with an unresolved search.
Code share: deployed code was not compared with the audited version. Of 94 files and 19,602 lines, 100.0% are "unknown" (0.0% identical, cosmetic, changed or unaudited). The map gives the reason as: the audit scope could not be determined from the available material. Lines not covered, by assessed criticality: critical 10,143; high 2,643; peripheral 6,816. One file has no line count available, and code in MultiTroveGetter has no criticality rating. Zero components were left unmeasured. The map gives no complete first-party address list: three inventory gaps state no current address is attributed for DSProxy, LQTYStaking and LUSDToken.
Gap and critical rows first. All Ethereum rows are Sourcify runtime matches unless stated; most "bound" rows name liquity/dev@a1824dd (tag mainnet_coverage); StabilityPool is the exception, bound to liquity/dev@b9a16d0.
| Component | Deployment | Status | Audit | Criticality | Note |
|---|---|---|---|---|---|
| MultiTroveGetter | eip155:1 0xfc92d0e9fa35df17e3a6d9f40716ca2ce749922b | Audit scope not confirmed | none named | Unassessed | No commit search done |
| TroveManager | eip155:1 0xa39739ef8b0231dbfa0dcda07d7e29faabcf4bb2 | In an audit's scope; revision ambiguous, version not compared | Trail of Bits, Mar 2021 | At least critical from assessed code | Commit liquity/dev@a1824dd |
| BorrowerOperations | eip155:1 0x24179cd81c9e782a4096035f7ec97fb8b783e007 | Same as above | Trail of Bits, Mar 2021 | At least critical | liquity/dev@a1824dd |
| Unipool | eip155:1 0xd37a77e71ddf3373a79be2ebb76b6c4808bdf0d5 | Same as above | Trail of Bits, Mar 2021 | At least critical | liquity/dev@a1824dd |
| LockupContractFactory | eip155:1 0x2ebef24da09489218ba2becb01867f6daaedcd4b | Same as above | Trail of Bits, Mar 2021 | At least critical | liquity/dev@a1824dd |
| PriceFeed | eip155:1 0x4c517d4e2c851ca76d7ec94b805269df0f2201de | Same as above | Trail of Bits, Mar 2021 | At least high | liquity/dev@a1824dd |
| StabilityPool | eip155:1 0x66017d22b0f8556afdd19fc67041899eb65a21bb | Audit scope not confirmed | none matched | At least critical | liquity/dev@b9a16d0 (tag @liquity/lib-ethers-v3.1.0); the audits name a StabilityPool, but the map does not link them |
| ActivePool | eip155:1 0xdf9eb223bafbe5c5271415c75aecd68c21fe3d7f | Audit scope not confirmed | none matched | At least critical | liquity/dev@a1824dd |
| DefaultPool | eip155:1 0x896a3f03176f05cfbb4f006bfcd8723f2b0d741c | Audit scope not confirmed | none matched | At least critical | liquity/dev@a1824dd |
| CollSurplusPool | eip155:1 0x3d32e8b97ed5881324241cf03b2da5e2ebce5521 | Audit scope not confirmed | none matched | At least critical | liquity/dev@a1824dd |
| CommunityIssuance | eip155:1 0xd8c9d9071123a059c6e0a945cf0e0c82b508d816 | Audit scope not confirmed | none matched | At least critical | liquity/dev@a1824dd |
| LQTYToken | eip155:1 0x6dea81c8171d0ba574754ef6f8b412f2ed88c54d | Audit scope not confirmed | none matched | At least critical | liquity/dev@a1824dd |
| SortedTroves | eip155:1 0x8fdd3fbfeb32b28fb73555518f8b361bcea741a6 | Audit scope not confirmed | none matched | At least high | liquity/dev@a1824dd |
| HintHelpers | eip155:1 0xe84251b93d9524e0d2e621ba7dc7cb3579f997c0 | Audit scope not confirmed | none matched | At least high | liquity/dev@a1824dd |
| GasPool | eip155:1 0x9555b042f969e561855e5f28cb1230819149a8d9 | Audit scope not confirmed | none matched | At least peripheral | Search incomplete: some repository trees or repositories could not be read |
| LUSDStablecoin | eip155:5000 0xf93a85d53e4af0d62bdf3a83ccfc1ecf3eaf9f32 | Audit scope not confirmed | none | At least critical | Etherscan lists the source (explorer claim, no runtime match); search incomplete |
| OptimismMintableERC20 | eip155:8453 0x368181499736d0c0cc614dbb145e2ec1ac86b8c6 | Audit scope not confirmed | none | At least critical | Bridged LUSD token; search incomplete |
| L2StandardERC20 | eip155:10 0xc40f949f8a4e094d1b49a23ea9241d289b7b2819 | Audit scope not confirmed | none | At least critical | Bridged LUSD token; search incomplete |
| BridgedToken | eip155:59144 0xda8aefcf0f9b0b81915a2c124f913e58212d49df (implementation of proxy 0x63ba74893621d3d12f13cec1e86517ec3d329837) | Audit scope not confirmed | none | At least critical | Search incomplete |
The component criticality shows the floor from assessed code; the rest of each component is unassessed.
Inventory rows (audit-named contracts with no attributed current address): 18 names, all with unknown criticality: BorrowerOperationsScript, BorrowerWrappersScript, DSProxy, ERC20Mock, ETHTransferScript, ITellor, LockupContract, LQTYStaking, LQTYStakingScript, "LUSD and LQTY tokens", LUSDToken, Migrations, "Pool contracts", "StabilityPool contract", StabilityPoolScript, TellorCaller, TokenScript, TroveManagerScript. Some of these (LQTYStaking, LUSDToken) are live contracts present in the on-chain control data, but the contract list does not attribute them an address.
Fix checks: the map could not determine whether the deployed code contains the auditors' fixes for the five findings the auditors mark fixed, because no fix review is available. This supports no claim either way.
On-chain control, Ethereum block 26411621 (2026-11-14): the StabilityPool owner reads as the zero address, i.e. renounced . Protocol docs state:
Findings by Severity
Critical
Total groups: 0; fixed per the auditor: 0; not marked fixed: 0. No critical findings are present in the ledger.
High
Total groups: 4; fixed per the auditor: 2; not marked fixed: 2.
- Delegatecall in DSProxy can self-destruct proxy (Trail of Bits, March 5, 2021) — — Audit status: not stated — Check: not marked fixed per the auditor (not stated); our code check is unavailable: the audit's finding detail was not located in the extracted text. The affected DSProxy has no attributed deployed address.
- Second delegatecall issue in DSProxy can lead to fund loss (same audit, second high finding) — same citation — Audit status: not stated — Check: not marked fixed per the auditor (not stated); our code check is unavailable for the same reason.
Medium
Total groups: 7; fixed per the auditor: 3; not marked fixed: 4.
- ERC20 transfer restriction can disturb 3rd-party interactions (Trail of Bits, January 13, 2021) — — Audit status: acknowledged (the fix log records "Risk Accepted" ) — Check: the code this finding describes is unchanged in the source verified for eip155:1:0x6dea81c8171d0ba574754ef6f8b412f2ed88c54d (Sourcify runtime match; identical to public commit liquity/dev@a1824dd) . This is a deliberate design choice the team accepted.
- Loss evasion by Stability Pool depositors front-running liquidations (Trail of Bits, January 13, 2021; listed there as a known issue; medium severity estimated) — — Audit status: not stated — Check: not marked fixed per the auditor (not stated); our code check is unavailable: the audit text was not located. The severity is an AI estimate because the audit states no rating.
- Inflating fees facilitated in low LQTY pool participation scenario (Coinspect, March 2021) — — Audit status: open — Check: the code this finding describes is unchanged in the source verified for eip155:1:0xa39739ef8b0231dbfa0dcda07d7e29faabcf4bb2 (Sourcify runtime match; identical to public commit liquity/dev@a1824dd) . Confidence of the check is moderate (7 of 10); no faster fee decay is visible.
- Liquidations incentives misaligned with system overall health (TCR) (Coinspect, March 2021) — — Audit status: open — Check: the code this finding describes is unchanged in the source verified for eip155:1:0xa39739ef8b0231dbfa0dcda07d7e29faabcf4bb2 (Sourcify runtime match; identical to public commit liquity/dev@a1824dd) . Confidence moderate (7 of 10).
Recurring Patterns
- No not-fixed critical, high or medium group is reported by more than one audit. Two groups come from Coinspect alone and relate to economic incentives rather than code defects. Two high groups come from a single Trail of Bits proxy review.
- Assessment: the unresolved findings concern economic design (fee inflation, liquidation incentives, stability-pool front-running), where an immutable protocol cannot patch the deployed contracts. The deployed contracts cannot be patched. For the two Coinspect economic findings, the code check found the described code unchanged in the verified source; the accepted ERC20 restriction is also unchanged there, and the team's stated position on it is risk acceptance. The front-running group and the two proxy-script groups are not marked fixed per the auditor, with no code check, so their live status is not confirmed.
Code Quality Assessment
- Review depth: two independent firms examined the V1 major version, and the tier-1 firm ran three engagements in early 2021. A Trail of Bits report records fuzzing with Echidna . Coinspect called the code well written and well tested . No public contest or formal-verification report is in the workspace.
- Fix rate: of 11 critical/high/medium groups, 5 are fixed per the auditors. The 6 not marked fixed are 2 proxy-script findings with no code check, 1 accepted design choice, 1 unchecked front-running item, and 2 unresolved economic findings confirmed in deployed code.
- Assurance: the V1 bug bounty is discontinued . It had a maximum pool of USD 250,000 . Core contracts have public, Sourcify-verified source, compiled with Solidity 0.6.11.
- Age: the audited code dates from 2021 and the contracts cannot be upgraded, so any later-discovered issue cannot be fixed in place. The audits do not cover later changes; the map found no commit-level comparison with audited revisions.
- Known compiler bugs: Solidity 0.6.11 (15 contracts, including the Ethereum core) carries 13 listed bugs. Four are rated medium: KeccakCaching, EmptyByteArrayCopy, DynamicArrayCleanup and AbiReencodingHeadOverflowWithStaticArrayCleanup. Another, MemoryByteArrayElementDeleteClearsWholeWord, is rated low/medium. The other 8 are rated low or very low; none is rated high. For example, KeccakCaching applies with the optimizer on (https://blog.soliditylang.org/2021/03/23/keccak-optimizer-bug/). No cited code location meeting any bug's conditions was checked, so none is a finding. Other versions (0.7.6, 0.8.9, 0.8.15, 0.8.18, 0.8.19) appear on bridged-token and Chicken Bonds contracts.
Score: 6.0/10
Rubric security-score-v2.6; data completeness 1.00
- Audit coverage
- 5.50 points
- Audit depth
- 7.50 points
- Unresolved findings
- 8.29 points
- Assurance
- 5.50 points
- High · not stated
- 2 findings
- Medium · open
- 2 findings
- Medium · not stated
- 1 findings
- Medium · acknowledged
- 1 findings
| Severity | Groups | Fix visible (deployed code) | Fix visible (repository only) | Described code still there (deployed code) | Described code still there (repository only) | Not checked or inconclusive |
|---|---|---|---|---|---|---|
| Critical | 0 | 0 | 0 | 0 | 0 | 0 |
| High | 2 | 0 | 0 | 0 | 0 | 2 |
| Medium | 4 | 0 | 0 | 3 | 0 | 1 |
Not scored: disputed groups and groups excluded for a later audit fix or privileged role.
Low confidence or archived: the check was not confident enough, or the repository is archived; the described code may still be there.
Deployed code: the source verified for the live contract. Repository only: the protocol's public repository; deployment not confirmed. “Described code still there” is a limited code signal, not a confirmed vulnerability: no exploit was tested; other contracts, configuration or operations may mitigate it. Findings the auditor marked fixed are trusted, not re-checked. Ask the team whether these are fixed or mitigated. Raw penalties: deployed code 1.20, repository only 0.00, audit status alone 0.51; Unresolved findings loses min(1.20 + 0.00 + 0.51, 10) of 10 points.
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Audit coverage (computed) | L2 | supported | 0.35 | 5.50 |
| Audit depth | L3 | supported | 0.20 | 7.50 |
| Unresolved findings (computed) | L3 | supported | 0.30 | 8.29 |
| Assurance | L2 | supported | 0.15 | 5.50 |
27 citations · click markers to view evidence
Economic Security
Economic Security Analysis: Liquity V1
Liquity V1 is an immutable, governance-free borrowing protocol on Ethereum with no admin keys. Assessment: its system collateral ratio is about 506% against a 110% liquidation line, and the Stability Pool covers about 24% of system debt. The main weaknesses are thin exit liquidity and high holder concentration for LUSD, a shrinking $129M base, and two medium audit findings marked not fixed. No bad-debt event is on record, but the freshest official statement dates from May 2025.
TVL & Liquidity Profile
Liquity V1 holds about $129.4M of TVL, all on Ethereum, and that figure fell about 17% in seven days while mostly reflecting the ETH price.
- The 7-day change of -$32.0M splits into a price effect of -$34.3M and a small net inflow of tokens (+$2.3M at current prices).
- The 30-day change of -$50.3M is almost all price effect (-$49.8M).
- Assessment: the declines come from the ETH price, since the single collateral is ETH and token quantities moved little. This is not a run on deposits.
- DefiLlama also lists $7.77M of LQTY staking value, which sits outside TVL.
- Ethereum is the only chain, so no chain profile or cross-chain split applies.
On-chain state at Ethereum block 26411621 (2026-11-14T07:07:11Z):
- System collateral is 71,472.95 ETH.
- System debt is 26,877,642.79 LUSD.
- The Stability Pool holds 6,439,791.74 LUSD.
- There are 64 Trove owners (borrowers with open positions).
- Assessment: at the Chainlink ETH price of $1,904.25 (answer 190425300022 with 8 decimals, same block), collateral is about $136M. That gives a system ratio near 506%.
- Assessment: DefiLlama's lower $129.4M figure likely reflects a different price time. Both are far above debt.
Exit liquidity for LUSD, the stablecoin users borrow and hold:
- The LUSD supply on Ethereum is $26.7M by Etherscan's on-chain market cap (source reliability: unverified publisher).
- DefiLlama reports a market cap of $26.65M across 8 chains.
- The Curve LUSD/3Crv pool holds $12.96M of liquidity (source reliability: unverified publisher).
- That pool did $244K of volume in 24 hours (source reliability: unverified publisher).
- The largest Uniswap v3 LUSD/USDC pool holds only $28K (source reliability: unverified publisher).
- The workspace depth measurement has no row for mainnet LUSD. The 2% depth and 0.5% depth of the main asset are therefore a data gap, not zero. The Curve figure above is pool liquidity, not depth at 2% price impact.
- The measured bridged LUSD copies on other chains are tiny: Optimism has $3,956 of depth against $79K supply.
- The 0.5% depth values are lower bounds from coarse probes, not exact capacity. Base shows $447 of depth against $8.9K supply.
- Arbitrum shows zero measured depth and a null 0.5% value.
- zkSync, Mantle and Linea LUSD copies are unmeasured and stay gaps.
The protocol's own exit path does not depend on DEX depth. Borrowers close a Trove by repaying LUSD, and any LUSD holder can redeem at face value.
Oracle Risk
Liquity V1 reads ETH/USD from Chainlink with Tellor as an on-chain fallback, and both have staleness checks while Chainlink also has a 50% deviation check, so oracle design is strong.
- A feed counts as frozen after 4 hours without an update.
- Chainlink is rejected if two consecutive updates differ by more than 50%.
- Returning to Chainlink requires both oracles to be live and within 5% of each other.
- Tellor prices are at least 15 minutes old, which gives disputers time to remove fake prices.
- Assessment: the two oracle providers are critical dependencies, and the sources cite no per-integration cap, pause or allowlist that isolates them. The fallback adds redundancy but is not isolation.
- No admin key can change the oracle choice or its parameters.
- The Chainlink ETH/USD proxy is a live feed at Ethereum block 26411621. It reports 8 decimals, phase 7 and a latest answer of $1,904.25.
Audit evidence on the oracle:
- Trail of Bits found that Chainlink could report erroneous prices after unfreezing. This is TOB-LQT-017.
- The final report marks it Fixed.
- Tellor's six-decimal format rests on social consensus, not on-chain data.
Residual oracle risks:
- If both oracles fail or are wrong together, the fallback has no third source.
- The README says a sub-100% liquidation is possible after a flash crash or an oracle failure, which would cost Stability Pool depositors.
- No incident of oracle failure causing V1 user loss was found in the pages the web research checked. This is absence of evidence only.
Flash Loan & MEV Exposure
Flash loans cannot mint or drain collateral directly, but they can inflate fees or force Recovery Mode in earlier code, and the docs describe front-running paths that stay open.
- Recovery Mode starts when the system ratio drops below 150%.
- Coinspect rated a Recovery Mode forcing issue (CI-LQY-01) high risk. It let an attacker force the mode through
closeTroveand then liquidate Troves. The team fixed it during the audit. - CI-LQY-03 (medium) describes attackers using flash loans to inflate system fees, mainly early on when total debt is low.
- The audit table marks CI-LQY-03 as not fixed.
- Assessment: the current debt of $26.9M and 53.1M LQTY staked make fee inflation harder than in the launch period the auditor described.
- Redemption fees have a 0.5% floor, which protects redemptions from arbitrageurs faster than the price feed.
Front-running in the Stability Pool:
- Deposit withdrawals are blocked while any Trove sits below 110%.
- The docs state this is only partly effective, because depositors can still front-run oracle updates.
- Flash deposits made just before a liquidation are described as helpful to system health.
- Redemptions are blocked when the system ratio is below 110%.
- Assessment: MEV searchers likely capture some liquidation gains through gas competition. This lowers returns for passive Stability Pool depositors but does not endanger solvency.
Tokenomics Risk: Yield, Leverage & Reflexivity
LUSD borrowers pay one-off fees, Stability Pool depositors earn liquidation gains and LQTY, and LQTY stakers earn real fee revenue, so yield rests on protocol fees with a small emission share.
Yield sources:
- Borrowing and redemption fees are the revenue. All of it goes to LQTY stakers.
- DefiLlama reports 30-day fees of $264K and revenue of $144K, and 365-day fees of $838K and revenue of $243K.
- LQTY community issuance, which excludes liquidity-pool incentives and the Community Reserve, follows a yearly halving schedule based on the figure 32,000,000.
- On chain, 31,346,304 LQTY of the 32,000,000 community issuance is issued at Ethereum block 26411621.
- Assessment: the remaining 654K LQTY implies about 327K LQTY of issuance in the next year, which is about $48K at $0.1469. Against $243K of yearly revenue, emissions are roughly 16% of combined paid yield.
- Stability Pool liquidation gains are not in the fee data. They depend on liquidation volume, which is not measured here, so the Stability Pool emission share is not measured.
Leverage and unwind:
- Users can borrow up to 110% collateral ratio, which allows leverage of up to about 11 times through looping.
- The leverage bound is the 110% minimum ratio.
- The unwind rule is permissionless liquidation below 110%, with the Stability Pool absorbing debt.
- The minimum debt is 2,000 LUSD.
Reflexivity:
- LQTY is not collateral and does not back LUSD. Staked LQTY is not a backstop and has no governance role.
- Redemption pressure sets the borrowing fee through a base rate that decays with a 12-hour half life.
- Assessment: a fall in the LQTY price would cut staker income in dollars but would not touch solvency. The reflexive loop is therefore weak.
- Borrowers face leverage risk. A drop in ETH of roughly 78% would be needed to reach the 110% line at the system level, but individual low-ratio Troves are liquidated much sooner.
- LUSD recently traded above peg: a 16.2% above-peg deviation peaked on 2026-10-29 and recovered after 14.9 days (source reliability: unverified publisher).
- Assessment: an above-peg event causes costs for borrowers who need LUSD to repay, not losses of backing. Redemptions cap the downside below $1.
Solvency, Liquidation & Bad Debt
Collateral is far above debt, liquidation is open to anyone, and the web research found no bad-debt event, though its newest official record ends in May 2025.
Liquidation rules:
- Anyone can liquidate a Trove below 110%.
- Liquidators receive 200 LUSD plus 0.5% of the Trove's collateral as gas compensation.
- The Stability Pool absorbs debt first and redistributes the remainder across active Troves.
- The only stated loss protection is the Stability Pool plus redistribution of defaults to other Troves.
- The protocol documents name no separate insurance, reserve or compensation fund. The system also has no admin key that could create one.
- The 200 LUSD liquidation reserve is a refundable deposit, not cover.
Backing and capacity:
- Assessment: the Stability Pool's 6.44M LUSD is about 24% of the 26.88M LUSD of system debt, both at Ethereum block 26411621. Far more collateral than debt remains, so redistribution would only matter after a very large, fast crash.
- Etherscan also shows the Stability Pool as the largest LUSD holder, at 6,439,791.7421 LUSD (source reliability: unverified publisher).
Bad-debt record:
- The official V2 repository states that no redistributions occurred in V1 from launch to May 2025.
- On 19 May 2021 the pool absorbed all defaulted debt of about 93.5M LUSD.
- In 2022 the system stayed out of Recovery Mode and LUSD held its peg.
- DefiLlama's hacks list, collected 2026-10-28, has no record for this protocol.
- The research found no V1 reporting for the 2024, 2025 and October 2026 market drops. The 2025-05 statement is therefore the freshest official record, and a gap remains after it. Assessment: the no-redistribution record likely still holds, but it is not cited past May 2025.
- The docs warn that a hack or bug causing user loss cannot be excluded, and no one compensates it.
- The V1 bug bounty has been discontinued.
Incentive Alignment
Borrowers, Stability Pool depositors, liquidators and LQTY stakers each have a cash reason to act, and the main misalignment is the open liquidation-incentive finding.
- Borrowers pay a one-off fee between 0.5% and 5% of the amount drawn.
- Redeemers pay
(baseRate + 0.5%) * ETHdrawn, so heavy redemptions raise their own cost. - Stability Pool depositors earn liquidation gains because liquidations happen just under 110%. The docs say that in rare flash crashes they can take a net loss.
- LQTY stakers hold no governance power and no backstop duty, so their incentive is fee income only.
- Coinspect's CI-LQY-04 says batch liquidation lets liquidators choose profit over system health during an ETH drop.
- That finding is marked not fixed.
- Assessment: this is a design limit in a protocol that cannot be upgraded. It matters most in a fast drop with a shallow Stability Pool.
- The protocol is non-custodial, immutable and governance-free, so no team can change incentives.
- Core contracts show the zero address as owner on chain, with no live control key.
Economic Attack Vectors
The credible attacks are a fast ETH crash with a drained Stability Pool, a joint Chainlink and Tellor failure, and attacks on the thin LQTY and bridged-LUSD markets.
- Fast crash. The system would need about a 78% ETH drop to reach 110% in aggregate (Assessment). The risk is more in individual Troves, which the Stability Pool must absorb. If the pool is empty, debt goes to remaining Troves instead.
- Oracle failure. The two-source design with the 4-hour, 50% and 5% checks reduces this. A failure that hits both sources at once is the open case.
- Forced Recovery Mode. Coinspect's high finding was fixed. Recovery Mode still triggers at 150%, and today's system ratio of about 506% is far from it (Assessment).
- Fee inflation by flash-loaned LQTY. CI-LQY-03 is open. Revenue is small ($243K a year), so the prize is small too.
- Thin LQTY market. The measured LQTY depth on Ethereum is $0, and a $100K probe moves the price 18.32%.
- LQTY holder concentration. The top 10 holders hold 80.31% of LQTY.
- Staking contract share. The largest holder at 53.13% is the Liquity staking contract, not a person.
- Assessment: LQTY holds no collateral role, so a manipulated LQTY price hurts fee-share value only. Staking weight set by the contract holder is not a governance risk because there is no governance.
- Bridged LUSD. The L2 copies have supply under $0.1M each and shallow depth. Losses there would stay local to those copies.
LUSD holders are also concentrated.
- The top 10 LUSD addresses hold 67.53% and the top 5 hold 59.51% (source reliability: unverified publisher).
- 25 holders own at least 1% of supply each (source reliability: unverified publisher).
- The list includes the Stability Pool at about 24% and the Curve pool at 4.07M LUSD, which are contracts rather than single people.
- The workspace measurement of LUSD on other chains differs from the Ethereum page figure. On 2026-11-15 (Optimism block 159557798) the top 10 hold 70.24%.
- On Arbitrum (block 524354566) the top 10 hold 73.88%.
- On Base (block 57620024) the top 10 hold 75.36%.
- These are separate bridged tokens with small supply. They are not the Ethereum LUSD that the Etherscan page's 67.53% describes, which is a web snapshot without a pinned block.
- On Arbitrum (block 524354566) the largest single address holds 46.73% of the bridged LUSD.
- Assessment: the measured Base top-10 share (75.36%) and the Arbitrum top-1 share (46.73%) set the concentration level for LUSD at its lowest band, because the weakest measured row decides and no measured Ethereum figure exists.
- Assessment: exits by a large holder could still move the peg, but the redemption floor at $1 limits the downside.
Aggregator Scores
DefiSafety gives 97.0, but the figure is an archived snapshot from 2025-05-19 and is not current evidence.
- The DefiSafety capture is dated by the Wayback Machine, which may not reflect the protocol today.
- The workspace has no CertiK record for this protocol.
- DefiLlama classifies the protocol as a CDP lender.
- Code-activity data show zero commits to
liquity/devandliquity/docs-v1for each month from 2025-12 to 2026-11, as expected for immutable contracts. - The audit reports are from 2021. The Trail of Bits report covering the price feed marks its two oracle findings fixed. Coinspect marks two medium findings not fixed.
Data Completeness
Core on-chain, fee and oracle data are complete, but mainnet LUSD depth, bad-debt records after May 2025 and Stability Pool depositor concentration are missing.
- Mainnet LUSD has no row in the depth measurement, so 2% and 0.5% depth are unknown. Curve pool liquidity of $12.96M is a different measure, taken from page snapshots.
- The web snapshots for LUSD supply, holders and pools have no pinned Ethereum block. They carry 2026-11-15 timestamps only.
- The Stability Pool share of supply (about 24%) compares a token balance with a supply figure. The pool's own deposit total (
getTotalLUSDDeposits) reads 6,439,791.74 LUSD at block 26411621, which matches Etherscan's token balance. - The on-chain file lists many V2 and Chicken Bond contracts that are not V1. This analysis uses only the V1 core contracts.
- Liquity V1 liquidation and bad-debt reporting after May 2025 was not found, including the 2024, 2025 and October 2026 market drops.
- Only aggregate top-holder data exists for LUSD. Per-depositor Stability Pool concentration was not found.
- Venue-wide Uniswap v3 and Balancer coverage is a ranked sample, not complete.
- The two audits are from 2021, and the link from audited commits to the deployed bytecode was not shown in the workspace. The address list for DSProxy has no attributed current address.
- Stability Pool share of liquidation gains and the emission share of Stability Pool yield are not measured.
Score: 7.4/10
Rubric security-score-v2.6; data completeness 1.00
- Solvency redemption
- 7.82 points
- Oracle design
- 9.50 points
- Dependency risk
- 5.50 points
- Liquidity concentration
- 3.50 points
- Yield design
- 7.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Solvency redemption | L3 | supported | 0.35 | 7.82 (loss cover +4.2%) |
| Oracle design | L4 | supported | 0.25 | 9.50 |
| Dependency risk | L2 | supported | 0.15 | 5.50 |
| Liquidity concentration | L1 | supported | 0.10 | 3.50 |
| Yield design | L3 | supported | 0.15 | 7.50 |
92 citations · click markers to view evidence
Team & Transparency
Social & Team Transparency: Liquity V1
Liquity V1 is built by a named team under a Swiss company, Liquity AG, and the protocol itself runs immutably without that company. The sources give a clear picture of team identity, published audits, a published contract list and an official post-mortem for a 2022 oracle flaw. Two points weigh against it. The V1 bug bounty has been discontinued, and the team's V2 product (another product of the same team) had a Stability Pool issue in February 2025. This conclusion is fairly firm on identity and transparency. It is weaker on community size, because Discord and X data are missing.
Team Identity
Team identity is clear: several named people and a named Swiss legal entity are published on the official site.
- Named people (official team page): the page is the current company page on the Liquity V2 site, so it lists the present roster, not only the V1-era team. The page names Robert Lauko as and Rick Pardoe as A June 2022 V1-era wrap-up names Rick Pardoe as Lead Engineer . The sources do not say when the other listed people joined or whether they worked on V1.
- Rick Pardoe, Lead Engineer:
- Bingen Eguzkitza, Head of Development:
- Samrat Lekhak, CEO, described on the same page as a growth and business specialist.
- Robert Lauko, listed as a research consultant and advisor:
- Public record of the founder: the V1 community-resources page lists many podcast and conference appearances by Robert Lauko, including a Bankless AMA .
- Legal entity: Liquity AG, Zurich, Switzerland . The V1 readme calls it "the project company" and says it developed the software but does not operate the protocol .
- Admin address: the readme says the Liquity admin address is a regular Ethereum address controlled by Liquity AG. It restricted LQTY transfers during the first year and holds no extra protocol privileges .
- Backers and advisors: the team page lists investor logos and a venture-investor advisor, Cédric Waldburger.
- Verification limit: the professional histories come from the team's own page. No independent press or filing source is in the available material.
- Frontends: the team does not run a frontend and says it has not vetted community frontend operators .
Track Record
V1 has run since April 2021 with no loss from an exploit in the sources, but it has had a disclosed oracle flaw, and the team's V2 had an issue in 2025.
- Operating history: the current V1 core contracts were created on 5 April 2021 . The core contract list is published in the docs.
- Stress test, 19 May 2021: The team says ETH fell , that and that This is the team's own account, and the source does not call the event market-wide. A team blog from July 2021 adds that during the crisis . That is about 3% below $1, and its duration is not stated. The measured LUSD price history starts in November 2022 , so it does not cover this event.
- Stress event, June 2022: the official Q3 2022 report describes bear-market conditions. It says and . It also says . Monthly TVL points were about $656 million on 1 June 2022 and $369 million on 1 July 2022 . Assessment: this is a second stress event with TVL above $100 million and no stated exit halt, depeg or bad debt. These are the team's own statements, and no independent price record for mainnet LUSD in June 2022 is in the sources.
- Hacks datasets: DefiLlama hacks and DeFiHackLabs hold no records for this protocol. The history file lists no incident dates.
- Incident 1, Tellor fallback oracle flaw (2022): a bug report on 17 September 2022 showed that Liquity used its fallback oracle Tellor unsafely. The official post-mortem is dated 22 November 2022. It says The team says a fake price on the ETHW fork minted trillions of LUSD there. It says Tellor's team implemented a fix and that mainnet never switched to Tellor. Loss on mainnet: none stated. TVL on the incident date is not in the history file. The row is classed minor because the post-mortem states mainnet funds were not at risk. That is the team's own statement.
- Incident 2, V2 Stability Pool ("Earn") issue (February 2025, another product of the same team): the official update says The team stated that no users had been impacted to its knowledge. The deployment became a legacy protocol, and V2 was relaunched on 19 May 2025 after an audit contest . The pages checked give no root cause, no loss accounting and no V2 TVL on the incident date. The row is classed major as the conservative rule-based default. It is not a proven loss. The V1 contracts are not named as affected in any page checked, and the V2 FAQ describes V1 as ongoing .
- Own-token peg history: LUSD is measured on six chains (Optimism, zkSync Era, Arbitrum, Mantle, Linea, Base). All six show no depeg runs against USD in the measured history. The deepest daily deviation was -1.55% on 2024-03-20 . The sidecar lists three of the six bridged deployments, the maximum a row allows. The mainnet LUSD address is not in the measured set, so its peg history is a gap here.
- Roadmap delivery: the V2 redeployment was first planned for mid-March 2025 . The team later moved the target to mid-May and relaunched on 19 May 2025 , about two months after the first target.
Social Presence
GitHub activity is steady, driven by V2 and API repositories; community-size data for X and Discord is not available.
- GitHub: commits landed in 12 of the last 12 months, with the latest on 2026-11-12 . These commits come mainly from V2-era and API repositories (for example
liquity/api.liquity.organdliquity/bold). The V1 repositoryliquity/devshows no commits in the window. Contributor count is not in the sources. - Telegram: the official channel shows 954 subscribers (source reliability: unverified publisher) .
- X/Twitter and Discord: the links exist on the official site. Follower and member counts and discussion quality are not in the sources.
- Blog: the official blog index lists V2 posts from January 2025 to 19 February 2026 . Posts are not monthly: the listing shows none between 15 August and 5 November 2025.
Transparency Indicators
Documentation, addresses and audits are public; the V1 bug bounty is gone.
- Audits: the V1 docs list a Trail of Bits assessment from January 2021, a Coinspect audit from March 2021, and two further Trail of Bits reports from March 2021 .
- Bug bounty: Reports on V2 go through Cantina. V1 questions go to a security email address.
- Open source: the V1 code and docs are in public repositories, and core contracts show Sourcify matches in the contracts list.
- Financials: DefiLlama fee data exists (30-day fees 264,128 USD, 30-day revenue 144,497 USD) . No treasury or expense report from the company is in the sources. V1 has no protocol treasury or governance by design.
- Post-mortems: the Tellor issue has a detailed post-mortem. The V2 issue has no public technical root cause in the pages checked.
Red Flags
- The V1 bug bounty program is discontinued, so V1 has no active paid disclosure channel in the sources.
- The 2022 Tellor flaw sat unnoticed in the deployed oracle logic until an outside researcher reported it. The ETHW-fork exploit shows it was exploitable.
- The V2 Stability Pool issue of February 2025 lacks a public root cause or loss accounting in the pages checked (another product of the same team).
- Professional histories of team members are not independently confirmed in the sources.
Positive Signals
- Named team and Swiss legal entity, with a founder who has a long public record.
- Immutable V1 design: the company states it does not operate or control the protocol.
- Published audits, a published address list and an official post-mortem for the Tellor flaw.
- The May 2021 and June 2022 stress events, as the team describes them: the Stability Pool absorbed defaulted debt in 2021 and the system functioned uninterrupted in 2022. A LUSD:3pool dip to about 97 cents in May 2021 had no stated duration.
- No peg breach beyond 1.55% in the measured bridged LUSD history, which starts in November 2022.
- Continuous GitHub activity across the team's repositories.
Data Completeness
- Available: official team page, company address in docs, audit list, contract list, 2021 to 2022 blog posts, V2 incident updates, GitHub counts, bridged LUSD peg measurements, a Telegram snapshot.
- Missing: X and Discord metrics, contributor counts, independent press verification of team histories, V2 incident root cause and loss accounting, mainnet LUSD peg measurement, treasury reports.
- No Blockworks (ex-Messari) reports are present; the article set is empty.
- This analysis rests on collected files, not live social media scraping.
Score: 7.5/10
Rubric security-score-v2.6; data completeness 1.00
- Team identity
- 7.50 points
- Disclosure
- 7.50 points
- Development activity
- 7.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Team identity | L3 | supported | 0.45 | 7.50 |
| Disclosure | L3 | supported | 0.35 | 7.50 |
| Development activity | L3 | supported | 0.20 | 7.50 |
Track record: 6.9/10
Rubric security-score-v2.6; data completeness 1.00
- Operating history
- 5.50 points
- Incident record
- 7.50 points
- Resilience
- 7.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Operating history (computed) | L2 | supported | 0.30 | 5.50 |
| Incident record (computed) | L3 | supported | 0.50 | 7.50 |
| Resilience | L3 | supported | 0.20 | 7.50 |
38 citations · click markers to view evidence
Governance
Governance Analysis: Liquity V1
Liquity V1 has no governance. Its core contracts (borrowing, Stability Pool, price feed, LQTY staking) have no owner, so no key can upgrade them, pause them or move user funds. This is confirmed on-chain on Ethereum at block 26411621 and matches the protocol docs and the 2021 Trail of Bits audit. The main trade-off is that a bug or a failed assumption cannot be fixed by anyone.
Governance Structure
- Type: none. The docs call the protocol "governance free", and there is no DAO, multisig-run upgrade path or timelock.
- Voting: none. The docs state "No. LQTY is not a governance token, as there is no Liquity governance." Staked LQTY earns fee revenue and is not used for governance.
- Proposal process, quorum, timelock, emergency bypass: none exist, because there is nothing to propose or execute.
- Vote concentration: not applicable. The vote-concentration file lists no venue (
venues: []). No Snapshot space or Governor is named in the sources, so there are no proposals to cite. - Core control: the V1 contracts use the OpenZeppelin
Ownablepattern (one owner address). The owner is the deployer, and ownership is renounced at deployment or right after the address setter is called.
Admin Key Analysis
Core V1 contracts have no admin key, and no other privileged account was found that can move user funds, mint LUSD or pause the system.
- Protocol claim: "Liquity has no admin key, and nobody can alter the rules of the system in any way."
- Audit statement (Trail of Bits, January 2021, publication provenance: trusted auditor source): ownership is renounced after deployment and "There are no other privileged accounts."
- On-chain check (Ethereum, block 26411621, read 14 November 2026):
owneris the zero address onTroveManager,StabilityPool,ActivePool,PriceFeed,SortedTroves,CommunityIssuanceandLQTYStaking. The privilege rows in the same file list one owner-gated setup function each, with the zero address as holder:setAddressesonBorrowerOperations,PriceFeedandActivePool, andsetParamsonSortedTroves. No one can call those functions now. The file also lists aBorrowerOperationsrow formoveETHGainToTrove, whose guard the parser could not resolve. The sources do not show that row as an admin power. - No proxies: the control file records no
implementationoradminslot for these contracts. A missing slot does not by itself prove the code cannot be upgraded. The docs describe the code as immutable, and the audit found no privileged accounts. - LQTY "Liquity multisig": the
LQTYTokenstores an immutablemultisigAddressand minted the team and investor allocation to it. Its power is limited to moving LQTY, and only to verified lockup contracts with unlock dates at least one year after deployment. After that year the restriction is lifted. The key moves the team's own LQTY, not user collateral or LUSD. - Safe behind
multisigAddress: the control file resolves a Safe at0xb8a9fada75c6d891fb77a7988ff9bad9e485ca1cwith threshold 3 of 8 owners and no modules. The file lists ten owner or threshold changes, the latest on 25 October 2023. A later web read of the Safe gateway (source reliability: unverified publisher) lists the same eight owner addresses, for example . A web search of the Liquity docs, team page and block explorers found no public mapping from these addresses to named people, so the signers are treated as undisclosed. - Chicken Bonds (LUSD bonding product, separate contracts): a "migration mode" can be triggered only by one admin, the Yearn Finance governance address
0xFEB4acf3df3cDEA7399794D0869ef76A6EfAff52. The control file shows that address as a Safe with threshold 6 of 9 and a guard. It also has three enabled modules:sudoImplant,ejectImplantandAllowanceModule. The file listsejectOwneramong theejectImplantprivilege rows. A Safe module can act without the owner threshold, so 6 of 9 is not the only control path. This control belongs to Yearn and to another product, not to the core V1 borrowing system. It is context here, and the sources do not show it can reach V1 core funds. - Benefit and cost of this design: with no key, there is no insider-rug or key-compromise path on the core. There is also no emergency pause, so a defect or oracle failure cannot be stopped by anyone. The protocol relies on built-in oracle fallbacks and Recovery Mode instead.
No control graphic is shown. The core V1 contracts have no upgrade or pause path to draw.
Upgradability
The V1 core is immutable; the sources show no upgrade route.
- Pattern: plain contracts with no standard proxy slot recorded.
- Who can upgrade: nobody. There is no timelock because there is no upgrade path.
- Core invariants: they cannot be changed. The 110% minimum collateral ratio and 150% critical ratio in the deployed
StabilityPoolviews are fixed values. - Contracts outside the core: the Chicken Bonds contracts and the V2 ("BOLD") contracts are separate products. This report does not assess them. Their ownership is not analysed here.
Parameter Controls
Parameters are fixed at deployment or move by algorithm; no person or vote can change them.
- Base rate and fees: the redemption and borrowing fees rise with redemption volume and decay over time. The redemption fee has a 0.5% floor outside Recovery Mode, and the borrowing fee is capped at 5%. The on-chain
BORROWING_FEE_FLOORread is raw5000000000000000, and this is consistent with 0.5% only under the 18-decimal denominator that the same contract exposes asDECIMAL_PRECISION(1000000000000000000). - Oracle settings: the
PriceFeedview shows a 14400-second timeout and a 50% maximum deviation from the previous round, and the contract has two oracles with a maximum difference of 5%. These raw values read as 50% and 5% only under the 18-decimal denominator, which is theDECIMAL_PRECISIONvalue in the same read. The external feeds (Chainlink, Tellor) can fail, and nobody inside the protocol can replace them. ThePriceFeedread names apriceAggregatorat0x5f4ec3df9cbd43714fe2740f5e3616155c5b8419, whose owner is a separate Safe,0x21f73d42eb58ba49ddb685dc29d3bf5c0f0373ca(threshold 4 of 9). It has one enabled module,ConfirmedTransactionModule, which can act without that threshold. This Safe sits outside the Liquity contracts. Its powers over the aggregator are not assessed in this report, and the module's authorization guard is not resolved in the on-chain file. - Harm potential: a collateral ratio below 100% cannot be set by anyone. Harm would come from a market or oracle failure, not from a parameter change.
Decentralization Assessment
Liquity V1 is at the fully decentralized end of the spectrum for core contracts.
- Today: no admin, no vote, no upgrade path. The docs say the contracts are accessible through multiple interfaces hosted by different frontend operators.
- Roadmap: none needed; the end state was reached at deployment.
- Key-person risk: low for the core contracts. The 3-of-8 LQTY Safe can only move the team's own LQTY. It cannot halt or drain the core.
- Token distribution: LQTY carries no governance power, so concentration does not translate into control. The measured LQTY holders are 53.13% in the
LQTYStakingcontract, which holds staked LQTY, with the rest across exchanges, pools and wallets. Only 1.14% sits in the protocol-ownedCommunityIssuancecontract. The off-chain frontends are the practical control point for user access and are not governed on-chain.
Governance Risks
- No-fix risk: this is the main risk. A bug or failed oracle dependency cannot be patched or paused. The audit record (Trail of Bits and Coinspect reviews from 2021) is the only mitigation; this report does not assess audit depth.
- Governance attacks: flash-loan or bribery attacks on votes do not apply, since there are no votes.
- Voter apathy: not applicable.
- Centralization creep: not possible for the core, since no role exists to extend. Related products (Chicken Bonds under Yearn governance) show that other Liquity-branded contracts can have admins, so users should check each product separately.
- Single-point scenarios: the Chainlink and Tellor feeds, and the frontends, are outside Liquity's control. The V1 rules stay immutable if either fails.
- Residual team power: the LQTY Safe (3 of 8, signers undisclosed) can move team LQTY, which affects LQTY market supply and not protocol solvency.
Data Completeness
- Available: official docs, a 2021 audit statement, an on-chain owner read at Ethereum block 26411621 for the core contracts, the LQTY Safe configuration and its change history, and holder concentration.
- Missing or limited: no governance venue or proposals exist to measure. Signer identities for the LQTY Safe stay undisclosed after a web search of Liquity docs, the team page and explorers. The on-chain file lists many roles as unresolved (115 privilege rows were unresolved and 96 unclear of 275) , so the absence of a power outside the core owner reads rests on the docs and audit, not on a full on-chain proof. The audit statement dates from 2021, and the current on-chain owner read confirms the ownership part.
Score: 9.5/10
Rubric security-score-v2.6; data completeness 1.00
- Admin control
- 9.50 points
- Change delay
- 9.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Admin control | L4 | supported | 0.45 | 9.50 |
| Change delay | L4 | supported | 0.30 | 9.50 |
| Decision process | — | not applicable | — | — |
41 citations · click markers to view evidence
Cross-Analysis
Cross-Risk Analysis: Liquity V1
The Liquity V1 core contracts have no admin key, so its main combined risks come from things nobody can fix: 2021-era code, two outside price feeds and market stress. Nobody can pause the system, replace an oracle or patch an open finding. Collateral is far above debt today (about 506% system ratio), so the most likely holder loss is a narrow one: Stability Pool depositors or leveraged borrowers in a fast ETH drop. Evidence for the mechanics and control state is strong. Evidence is weaker for how the deployed code relates to the audited code and for LUSD market depth on Ethereum.
Liquity V1 is not a bridge, so the Conditions of Safe Use section is omitted. The bridge-route data holds 110 routes of CCIP (Chainlink's cross-chain protocol). The data does not attribute any of them to a Liquity token, and the dependency graph has no bridge edge for LUSD or LQTY .
Top Risks for a Holder
The three risks, most serious first, are an unfixable failure of code or oracle, a fast ETH drop that drains the thin Stability Pool, and an exit squeeze for LUSD and LQTY holders.
-
Unfixable code or oracle failure (all products: Troves, Stability Pool, LUSD, LQTY staking).
- What can go wrong: a latent bug in the 2021 contracts, or a joint Chainlink and Tellor failure, hits a system with no pause and no upgrade path. The core contracts have no owner, so "no key can upgrade them, pause them or move user funds" . The governance report adds that "There is also no emergency pause, so a defect or oracle failure cannot be stopped by anyone." Review depth is thin for the live code: the available material does not link the audited commits to the deployed bytecode, and the bug bounty is discontinued . Both oracle providers are critical dependencies with no isolation .
- Magnitude: up to the full collateral base, about 71,473 ETH, or about $136M at the Chainlink price on 2026-11-14 . DefiLlama shows $129.4M TVL . Staked LQTY ($7.77M) is a separate figure.
- Likelihood: low. Basis: V1 has run since April 2021 with no hack record in the DefiLlama and DeFiHackLabs data, and two independent firms audited the code. The mainnet oracle flaw of 2022 was reported by an outsider and was not exploited on mainnet . The missing fact that would change the band is a comparison of the deployed bytecode with the audited commits.
-
Fast ETH drop drains the Stability Pool (Stability Pool depositors and leveraged Trove owners; LUSD).
- What can go wrong: a sharp ETH fall liquidates many Troves faster than the pool can absorb. The pool holds 6.44M LUSD against 26.88M LUSD of debt, about 24% . The remainder is redistributed and "Redistribution spreads the loss across remaining Troves." . Two Coinspect medium findings (liquidator incentives and flash-loan fee inflation) are still open and the described code is unchanged in the deployed
TroveManager. No insurance or reserve fund exists . - Magnitude: the pool's 6.44M LUSD is the first loss layer. Only 64 Troves are open, so a few large positions drive system health . The whole system would need about a 78% ETH fall to reach 110% in aggregate , so system-wide insolvency is remote.
- Likelihood: low for a net loss to depositors or any redistribution; routine for leveraged Troves in a sharp drop. Basis: the pool absorbed about 93.5M LUSD of defaulted debt on 19 May 2021, and the official V2 repository says zero redistributions occurred in V1 up to May 2025 . No V1 reporting covers the 2024, 2025 and October 2026 drops . Per-depositor pool concentration is unmeasured and would change the band.
- What can go wrong: a sharp ETH fall liquidates many Troves faster than the pool can absorb. The pool holds 6.44M LUSD against 26.88M LUSD of debt, about 24% . The remainder is redistributed and "Redistribution spreads the loss across remaining Troves." . Two Coinspect medium findings (liquidator incentives and flash-loan fee inflation) are still open and the described code is unchanged in the deployed
-
Exit squeeze for LUSD and LQTY holders (LUSD, bridged LUSD copies, LQTY).
- What can go wrong: LUSD holds its $1 floor by redemption, but depth on open markets is thin and concentrated, and redemption stops if the system ratio is under 110% . LQTY has no measured depth on Ethereum: "The measured LQTY depth on Ethereum is $0, and a $100K probe moves the price 18.32%." . The top 10 LQTY holders hold 80.31% . The largest holder, at 53.13%, is the Liquity staking contract, not a person , so free-float concentration is lower than 80.31% suggests.
- LUSD concentration (source reliability: unverified publisher): the top 10 LUSD addresses hold 67.53% and the top 5 hold 59.51%, and 25 holders own at least 1% of supply each . The list includes contracts such as the Stability Pool (about 24%) and the Curve pool . The economic report notes the figure is a web snapshot without a pinned block.
- Magnitude: LUSD supply on Ethereum is about $26.7M against $12.96M in the Curve pool and $28K in the largest Uniswap v3 pool (source reliability: unverified publisher for the pool and supply figures). Bridged copies on Optimism, Base and Arbitrum have supply of $79K, $8.9K and $70K with depth of $3,956, $447 and zero . Losses on the copies stay local to them.
- Likelihood: medium for a price dislocation, low for principal loss. Basis: LUSD already went 16.2% above peg from 2026-10-29 for 14.9 days (source reliability: unverified publisher) . That costs borrowers who need LUSD to repay, not backing . The missing fact is mainnet LUSD 2% and 0.5% depth, which the depth measurement lacks.
- LUSD on-chain market cap
- 26.71 USD m
- Curve LUSD/3Crv liquidity
- 12.96 USD m
- Stability Pool (LUSD held)
- 6.45 USD m
- Largest Uniswap v3 pool
- 0.03 USD m
Combined Attack Vectors
Four vectors combine findings from several reports. None needs a key compromise of the core Liquity contracts, because they have no admin key. One separate branch of Vector 2 concerns a Safe outside those contracts that owns the oracle aggregator.
Vector 1: Crash, liquidation-incentive gap and shallow pool
- ETH falls fast. Many Troves cross 110% while the pool covers about 24% of debt .
- Coinspect's open finding says liquidators prefer Troves that matter least for the system ratio . The code check found the described code still in the deployed
TroveManager. - Pool depositors can watch the price feed and withdraw before liquidations. Trail of Bits listed this as a known issue (medium, AI-estimated; the audit states no rating) and it is not marked fixed . The docs call the withdrawal block "only partially effective" .
- The pool runs short, so debt and collateral are redistributed onto the remaining Troves. No admin can change parameters or add a backstop .
- Cross-domain point: the code report's coverage map does not link any audit to the deployed
StabilityPool, which is bound to a different commit than the other core contracts. The finding sits in the component with the least confirmed audit link . - Domains: economic security, code security, governance, protocol overview.
- Likelihood: low (net depositor loss); medium (liquidation of leveraged Troves). Basis: past absorption record above, and a system ratio near 506%.
- Impact: moderate. Loss is bounded by pool size and the Troves involved, not by the full $129M.
Vector 2: Oracle failure with no pause and external control of the feed
- Chainlink and Tellor are both wrong or both stale. The fallback has no third source .
- Precedent from the 2022 post-mortem: on the ETHW fork Chainlink did not push updates, so the fallback logic switched the Liquity instance to Tellor, as designed. The flaw was Liquity's unsafe Tellor usage, with no dispute period, and a fake price then minted trillions of LUSD there . The team says mainnet never switched to Tellor and that Tellor's fix returns prices at least 15 minutes old . The team's own account is the only source.
- Wrong prices trigger wrong liquidations. Nobody can pause the system or swap the oracle, and depositors take a net loss when Troves are liquidated below 100% .
- Domains: economic security (oracle), governance (no pause, external Safe), social (2022 precedent) .
- Likelihood: low for the joint-failure scenario above. Basis: both feeds must fail together, and the team says mainnet never switched to Tellor in the 2022 incident.
- Impact: severe to catastrophic. Wrong liquidations would reach collateral across the system.
- Separate, unresolved branch (not covered by the band above): the
PriceFeedread names apriceAggregatorwhose owner is a separate Safe (threshold 4 of 9) with an enabled module,ConfirmedTransactionModule. The on-chain file could not resolve the module'sauthorizedguard on its functions, so who can execute through it and what it can change on the aggregator are not established . This branch needs only the Chainlink-side aggregator to be controlled, not both feeds to fail, and it is the one place where a key-based path may exist. Likelihood: low (assessment, weak basis; no sign of misuse in the sources). The fact that would decide the band is the module's executor set and powers over the aggregator.
Vector 3: Open finding plus an unfixable defect, with no paid disclosure
- A new flaw is found in the 2021 code, or an open finding becomes exploitable. The available material does not show which audit fixes are live .
- The bounty (maximum pool USD 250,000 before it ended) is discontinued, and the sources show no active paid V1 disclosure channel .
- The code cannot be changed: "any later-discovered issue cannot be fixed in place" . Users can only exit.
- Known open audit items outside the economic findings: Trail of Bits's March 2021 proxy review has two high-severity
DSProxydelegatecall findings, which "could result in the proxy being self-destructed and losing funds" . Both are not marked fixed, and the code report ran no code check. Their live status is unconfirmed, and the affectedDSProxyhas no attributed deployed address . They concern the peripheral proxy scripts, not the core contracts. - Domains: code security, social (bounty), governance (no upgrade path).
- Likelihood: low. Basis: no new finding appears in the sources, and
liquity/devshows zero commits from 2025-12 to 2026-11 . - Impact: catastrophic if the flaw touches collateral custody. Moderate if it is economic, such as the fee-inflation finding.
Vector 4: Unvetted frontends over immutable contracts
- The team runs no frontend and has not vetted operators .
- The governance report calls frontends "the practical control point for user access" .
- A malicious or compromised interface can send a user's approvals or transactions elsewhere. Users who reach the contracts through proxy-script wrappers also touch the
DSProxylayer, whose two high-severity audit findings have unconfirmed live status (see Vector 3). On-chain transactions cannot be reversed by any admin, and the sources show no active V1 bounty.
- Domains: social, governance, code security. The social report also records that the team does not run a frontend .
- Likelihood: low to medium. Basis: this vector rests on the frontend design and unvetted operators, not on a recorded incident in the sources.
- Impact: moderate. Loss is per user, not systemic.
Systemic Risk Factors
- Dependency graph (collected 2026-11-15). One edge is
IS_BACKED_BYWETH at share 1.0, with the WETH node unresolved, and one oracle edge to Chainlink . The graph lists no protocol with a published score and no warnings, so it shows no circular backing and no downstream holders. The economic report scores dependency risk at level 2. - Bridges carrying LUSD. The sources list LUSD copies on Optimism, Base, Arbitrum, zkSync Era, Mantle and Linea. Supply on each is under $0.1M where measured . The graph has no
bridgeedge for them, and no bridge's route settings could be matched to a Liquity token. The bridged tokens are not covered by any audit in the sources . - Chain-level. Ethereum is the only TVL chain, so an L2 sequencer outage does not touch core collateral. Assessment: Ethereum congestion would delay liquidations, because liquidation is an on-chain call anyone must make.
- Macro. Collateral is ETH only, so a market-wide drop hits every Trove at once. The 30-day TVL fall of $50.3M was almost entirely price effect (-$49.8M) .
- Correlation. Assessment: a market-wide ETH drop can raise liquidation volume (Vector 1) and put the oracle under stress (Vector 2) at the same time. The economic report lists these as separate attacks and calls the joint oracle failure the open case . No report says a crash causes an oracle failure.
Risk Amplifiers
- No pause + oracle dependency: a bad price cannot be stopped, so any oracle error is permanent in effect (Vector 2).
- Open audit findings + immutability: the two open Coinspect findings are confirmed in deployed code and cannot be patched .
- Few Troves + shallow pool: 64 Troves and a pool that covers about 24% of debt let one large position move system health.
- Thin markets + redemption block: if the system ratio fell under 110%, the par exit for LUSD stops while open-market depth is thin.
- Discontinued bounty + unlinked audits: the sources show no active paid V1 disclosure channel, and the available material does not show which audited fixes are live.
Risk Mitigators
- No admin key on the core contracts. On-chain reads on 2026-11-14 (block 26411621) show the zero address as owner of the core contracts, so the core contracts have no insider-rug or key-compromise path. The Safe that owns the oracle aggregator is outside this statement (see Vector 2) .
- High collateral cover. A system ratio near 506% against 110% leaves a wide buffer.
- Oracle design. Chainlink primary, Tellor fallback, a 4-hour staleness check, a 50% deviation check and a 5% return check. The economic report scores oracle design at level 4.
- Stress record. The team states the pool absorbed all defaulted debt on 19 May 2021, and the system "functioned uninterrupted" in June 2022 . These are the team's own statements.
- Review depth. Two independent firms audited the V1 code in 2021, with three Trail of Bits engagements .
- Team LQTY key is narrow. The 3-of-8 LQTY Safe can only move the team's own LQTY .
- Not a mitigator: the DefiSafety score of 97.0 is an archived May 2025 snapshot, not current evidence.
Contradictions & Inconsistencies
- Peg history. The social report lists "No peg breach beyond 1.55% in the measured bridged LUSD history, which starts in November 2022." . The economic report cites a 16.2% above-peg excursion from 2026-10-29 that lasted 14.9 days (source reliability: unverified publisher) . The two are not reconciled. The measured series records only a downside minimum of 1.55% below peg and no mainnet LUSD address, so the above-peg event does not appear in it. Treat the "no breach" signal as covering downside moves in bridged copies only.
- Price ceiling. The overview and economic report describe a price floor and ceiling from redemption and the 110% minimum ratio. A $1.16 peak (source reliability: unverified publisher) is a measured excursion above that ceiling logic, and no report reconciles the two. The cause is not stated in the page.
- Oracle strength versus oracle control. The economic report scores oracle design at level 4 and does not mention the Safe (4 of 9, with a module) that owns the Chainlink-side
priceAggregator. The governance report notes it, says its powers over the aggregator are not assessed there, and says the module's authorization guard is not resolved in the on-chain file . No report assesses the Safe's powers. This is a coverage gap, not a measured disagreement. - Tellor precedent versus "strong" oracle design. The social report records an exploitable Tellor usage flaw in the deployed logic, shown on the ETHW fork. The economic report states "No incident of oracle failure causing V1 user loss was found" and calls the design strong. Both can hold: no mainnet loss, but a latent flaw existed. The economic report does not cite the incident.
- LQTY admin address. The social report says the readme describes a regular Ethereum address controlled by Liquity AG . The governance and overview reports read a Safe with 3 of 8 owners on-chain at block 26411621. The immutable
multisigAddressinLQTYTokenresolves to that Safe . The publication date of the readme is not established, so the conflict stays unresolved. - Audit coverage versus review depth. The code report counts 22 deployments with no component shown audited at its deployed version, yet scores audit depth at level 3. The depth credit rests on audit quality, not on deployment match.
- Absence claims. The overview says no hack records appear in DefiLlama or DeFiHackLabs. The social report documents two incidents, both outside mainnet V1 losses: the 2022 Tellor flaw (no mainnet loss stated) and the February 2025 V2 Stability Pool issue (another product of the same team; no root cause or loss accounting published in the pages checked). Neither is a V1 mainnet loss event. The V2 issue does not name V1 contracts as affected.
Overall Risk Profile
Liquity V1 has a low governance and key risk for its core contracts and a moderate code and market risk. Immutability removes insider and upgrade risk and also removes any response to failure. The strongest protections are high over-collateralization, open liquidation and a two-source oracle.
The weakest points are listed here:
- unlinked audits of 2021 code;
- two open economic findings confirmed in deployed code;
- two high-severity
DSProxydelegatecall audit findings not marked fixed (live status unconfirmed, no attributed deployed address); - no active bounty;
- a Stability Pool that covers about 24% of debt;
- thin open-market depth for LUSD and LQTY;
- concentrated LUSD holdings (top 10 at 67.53%, an unverified-publisher snapshot, part of it held by contracts).
The most likely failure mode is a sharp ETH drop that costs leveraged Trove owners and, in a fast move, Stability Pool depositors, with LUSD price dislocation in either direction. The worst case is a joint oracle failure or an unknown code defect, with no pause and no fix; that would reach about $136M of collateral.
Assessment: against typical CDP stablecoins with admin or governance control, V1 has lower governance risk and fewer dependencies (ETH and one oracle pair). It has older code and no live security programme. No peer data was supplied for a measured comparison.
Data Gaps
-
Some source material could not be collected; the report may be incomplete.
-
Some protocol documentation could not be read because the page was unavailable.
-
Some source material is unavailable because it was not found publicly.
-
Current deployment addresses could not be confirmed for some audited contracts.
-
External rating https://skynet.certik.com/projects/liquity-usd is named by CertiK but was not measured: family member unclear (CertiK lists it as 'Liquity USD', proposed for this protocol; the name fits more than one family member).
-
External rating https://exponential.fi/protocols/liquity/07d1c31c-ee43-416d-8c7f-1f259eb0e750 is named by Exponential but was not measured: family member unclear (Exponential lists it as 'Liquity', the name of the family of liquity-v1, liquity-v2).
-
External rating https://skynet.certik.com/projects/liquity is named by a manual CertiK check (2026-11-13) but was not measured: a live scored page that no run staged; the CertiK lookup tries this protocol's name and slug, not its family slug.
-
The deployed bytecode has not been compared with the audited commits. The code report lists 94 files with 100.0% "unknown" overlap. This stops any assessment of whether audit fixes are live.
-
Mainnet LUSD has no depth measurement and no peg history in the measured set. The pool, supply and holder figures are web snapshots with no pinned Ethereum block.
-
Per-depositor concentration in the Stability Pool is not measured, so the loss split in Vector 1 is not estimated.
-
No V1 liquidation or bad-debt reporting was found after May 2025, including the 2024, 2025 and October 2026 market drops.
-
No report assesses the powers of the Safe that owns the
priceAggregator. -
The on-chain control file leaves many roles unresolved: 115 privilege rows were unresolved and 96 unclear of 275 . The absence of a power outside the core owner therefore rests on the docs and the audit, not on a full on-chain proof .
-
The cause of the October 2026 above-peg event is not stated in the available page.
-
Material dependencies with no published report in the workspace: Chainlink (oracle), Tellor (fallback oracle) and WETH (unresolved collateral token). No dependency report index exists for this protocol. The bridges that carry LUSD to other chains are not identified in the data.
-
The bridge-route file reports a completed read, but its 110 routes are all unattributed, so none was matched to a Liquity token .
-
X and Discord metrics and independent verification of team histories are missing, which limits the social side of the analysis.
78 citations · click markers to view evidence