Euler Review

How Euler scores in crypto lending

This entry rates Euler v2, the lending protocol built from separate vaults deployed with the Euler Vault Kit (EVK) and linked by the Ethereum Vault Connector (EVC), in our lending ranking. The basis is desk research of the Euler documentation, the EVK whitepaper and the GitHub repositories, checked on November 1, 2026. No position was opened. The category uses five criteria, each marked 0 to 10 and weighted equally, as described in our ratings methodology.

The overall score is the plain average: (8 + 4 + 6 + 7 + 10) ÷ 5 = 35 ÷ 5 = 7.0. That places Euler behind Kamino at 7.2 and ahead of Aave at 6.6.

One mark was re-scored from current sources on this check. Audit disclosure rose from 9 to 10 after the documentation published a dated review table of more than 60 engagements, replacing the 12-report archive the previous mark was based on. That matches the 10 that Aave holds for a dated, firm-named archive. The overall score moves from 6.8 to 7.0.

Strongest criteria: audits and fee limits

Audit disclosure (10)

Euler's documentation keeps a review record table with a date, scope, review type, firm and report link on each row. It covers the EVK, the EVC, the price oracles, periphery contracts and Fee Flow. Firms include OpenZeppelin, Spearbit, ChainSecurity, Trail of Bits, Certora, Omniscia, yAudit, Electisec, MixBytes and Enigma Dark, and the most recent entry is an Enigma Dark review from July 2026. The security page puts the public record at more than 60 engagements and lists a Cantina bug bounty of up to $7.5 million and a SEAL Safe Harbor agreement. The repositories hold the reports themselves: 12 files in the vault kit's audits folder and 11 in the EVC's. Some rows are private or marked "not publicly available", including a February 2024 Certora formal verification of v2 and a March 2026 Cantina review still pending publication. They are disclosed as such, and the public reports cover every core component, so they do not hold the mark down. Audits narrow risk and do not insure deposits.

Fee disclosure (8)

Euler's Fee Flow page states that in April 2026 Fee Flow was turned off and protocol fees were set to 0% across live and future deployments; the DAO can bring them back by vote. The limits on vault fees are written into the code, as the whitepaper describes. A vault's interest fee, the share of borrower interest kept back from lenders, must be between 10% and 100% to stay within the guaranteed range. If the protocol config asks for a protocol share above 50% of that fee, the vault ignores it and uses 50%. What the documentation cannot give is each vault's actual interest fee, because curators set it. That is the reason the mark is not higher.

Weakest criteria: licence and per-vault terms

Licence & code openness (4): a deficiency

The vault kit LICENSE is Business Source License 1.1, with Euler Labs Ltd. as licensor and the additional use grant set to "None". Under BUSL terms that means the code may be copied, modified and used outside production, but production use is not licensed until the change date of April 24, 2029, when it converts to GPL-2.0-or-later. The EVC, which links vaults together, is already GPL-2.0. The rubric places delayed-open licences mid-table. A conversion date two and a half years away, with no use grant, keeps this mark at 4.

Liquidation terms (6): per vault by design

The liquidation mechanics are documented. Once an account's health falls below 1, the discount starts at zero and grows with the shortfall until a liquidator acts or the vault's maximum discount is reached. A governor can set a cool-off time that blocks liquidation in the block after a healthy check, to resist oracle manipulation. Bad-debt socialization, which spreads leftover debt across depositors, is on by default and can be switched off. None of this comes with protocol-wide numbers. Borrowing and liquidation LTVs, the maximum discount and the cool-off are set by each vault, and the docs warn that the maximum discount defaults to 0 if the creator does not set it. A borrower has to read the specific vault before planning around it.

Governance & timelock (7)

Control is also per vault. The whitepaper sets out four cases, from upgradeable and governed, to immutable with no governor, which no one can change. The governor suite Euler recommends uses admin and wildcard timelocks with a minimum delay of one day. A CapRiskSteward can adjust caps and swap rate models within preset limits without a timelock. Use of the suite is recommended, not required. Euler's own Q1 2026 retrospective shows the cost of the other end of the range: markets with no governor built up bad debt after the Stream Finance default and the March 2026 Resolv exploit, because no one could pause them.

What the score does not cover

• Vault quality. Individual curators, oracles and collateral lists were not assessed.
• Euler Earn and EulerSwap. The aggregator vaults and the AMM are outside the five criteria.
• Rates. Supply and borrow rates were not measured.
• History. The March 2023 exploit of Euler v1, a different codebase, is not scored here.

Euler vs Kamino and Aave

Kamino scores 7.2. It marks 10 for fee disclosure and 7 for liquidation terms, against Euler's 8 and 6, but 6 for governance and timelock. The two share a 4 for licence, and Euler edges Kamino on audits, 10 against 9. Aave scores 6.6, and the gap is mostly fees: 4 against Euler's 8. Aave leads on governance (9 against 7), and the two match on licence, liquidation terms and audits (10 each). Euler's per-vault design is the common thread in its marks: strong where limits are set in code, weaker where each vault decides.

Who this rating suits

The score suits readers who value a dated, linked audit record and fee limits enforced in code. Readers who want one set of liquidation terms and one governance process for every market should weigh the 6 and the 7, and read each vault's settings before borrowing; our guide to reading a DeFi lending market covers what to check. The rating describes published documentation and is not advice to lend or borrow.