The isolated kUSD/USDC market on Euler Base is deployed. It is permissionless, so this is a
notification and an underwriting packet, not a governance request. Nothing here needs a vote.
Addresses
| USDC vault, borrowable | 0xe87c294E1139C31770727193e3Be0ccEd73d6AA3 |
| kUSD vault, escrow collateral | 0xD3800ceb6bBeB90101ed5d5A46017fE846c9509e |
| EulerRouter | 0xB95407727d33fB2E444966fd0B7D0E7767Ce019E |
Oracle, FixedRateOracle(kUSD, USDC, 1000000) |
0x8f27228f02E798c17B7d6b270F32F8EC6afDD2D3 |
| Deployed via EdgeFactory | 0x4B930F0222349c2092b8531A42295262cc4F0e4A |
| Deploy transactions | 0xbacaf5445effce5d32b57ee512ab54123306e24de437ab93dac36e3ca024c9ad, 0xcda40c26a970869efef971912a39b2ac7acae68b647ed2d54d72453f6380bef7, block 49,726,126 |
Parameters
Borrow LTV 60.00%, liquidation LTV 65.00%, unit of account USDC, IRM
0xC6dCfFE18cd9f532628aF15da2541c01499E3bcE (kink 90%, 0% base, about 6.00% APR at kink), max
liquidation discount 15% and liquidation cool-off 1 second, both fixed by the factory.
Governance on both vaults and on the router is address(0). EdgeFactory renounces it in the same
transaction that creates them, so Kerne cannot change the oracle, the LTV or anything else. The
consequence a reader should hold onto is that no supply or borrow cap can ever be set on this
market either. Caps live in a curator’s vault, the same as on Morpho Blue. Any number we name is a
recommendation to an allocator, not a control we hold.
1. Summary
kUSD is a synthetic dollar on Base whose backing today is entirely USDC in three on-chain Peg
Stability Modules, not the delta-neutral basis position the design describes. The basis position
exists, it is small, and it backs no kUSD at present. That was established in topic 1849 by an
independent reviewer and we agreed with it on the record rather than arguing.
Reserves 1,110.888006 USDC against 1,109.707154 kUSD outstanding, a ratio of 1.00106411, read at
Base block 49,722,336 on 2026-08-08 and reproducible in six read calls.
2. Background
Topic 1849 has been open since 2026-07-03. Over thirty-five days
@Oleg_Aleksandrov tested the design in public: the
canonical versus aggregate solvency split, founder-wallet contamination of the hedge base and its
correction, signed account enumeration, negative funding absorption, the timelock deployment, the
guardian bytecode, and a two day seventeen hour admin overlap window that he reconstructed more
tightly than we had published it. He verified most of it independently and he found real errors in
our posts, which we corrected in the thread rather than quietly.
That record is the reason this post exists. It is also, in our view, the only unusual thing about
Kerne at this size.
3. Risk analysis
Oracle risk. The oracle is a fixed rate of 1 kUSD = 1.000000 USDC using Euler’s own
FixedRateOracle adapter. It is depeg-blind: it will not mark kUSD down and the market will not
liquidate into an impairment. We chose it over the backing-ratio oracle we specified earlier in 1849
because a bespoke, unaudited oracle written by the asset issuer inside a live lending market is a
worse risk than a conservative LTV. The protection is the LTV and the allocator’s cap, not the feed.
Liquidity risk, and this is the sharp one. The only DEX venue for kUSD is a single Aerodrome
stable pool holding 5.978637 kUSD and 7.088931 USDC. About thirteen dollars of depth. A liquidator
cannot sell seized kUSD into a market. The only real exit is redemption through the PSM at the
backing ratio, minus a size-tiered fee of 10 bps below 50k, 8 bps to 250k, 7 bps to 1M and 5 bps
above, all four boundaries measured on chain today. That is a materially different liquidation model
from normal collateral and it should be priced as one.
Centralization risk. DEFAULT_ADMIN_ROLE on kUSD and on all three PSM modules sits with a
TimelockController at 172,800 seconds. PROPOSER, EXECUTOR and CANCELLER are all the same
2-of-3 Safe; the Pause Guardian holds no timelock role. MINTER_ROLE nets to exactly one holder by
full event replay, the live mint PSM.
Two things follow that we would rather state than have found:
- The staking vault skUSD is outside the handover. Its admin is still the Safe, not the timelock,
and skUSD holds 88.37% of kUSD supply. - A 48 hour timelocked
grantRolecan create an unbounded kUSD minter. At par pricing that mints
collateral from nothing and drains any market that accepts kUSD. This is true of this market, of
the existing Morpho market, and of any kUSD market anyone deploys. It is the binding constraint on
what a sane cap looks like. The 48 hours are public, and as of today nothing subscribes to
CallScheduledon that timelock, which the same reviewer pointed out and which is not fixed.
Smart contract risk. Hexens final report 2026-07-31, SHA-256
655e7126030c750e9d58f2ab30b58215ce604942f06997d5c01532d6f687a4ca, reviewed commit 0912c870.
Zero critical, two high, two medium, four low, two informational. Eight fixed, two acknowledged. All
ten are in KerneVault, and all ten are live on the deployed vault bytecode, which predates the
audited commit by about three weeks. That vault holds no user funds, has never issued a share, cannot
accept a deposit and no longer holds MINTER_ROLE. The live mint PSM, which is the contract on the
path that matters here, drew no findings and its deployed bytecode is Sourcify exact_match and
byte-identical to the reviewed source.
Two further live divergences between deployed and current source are published in full at
kerne.fi/security/deployed-vs-source, including an ungated burnFrom on kUSD.
4. What we have actually delivered, since it argues against us
Realized yield on skUSD is 0.0022% annualized over the trailing 30 days, measured as on-chain share
price growth, against an advertised forward model of 4.7%. That is about 0.05% of what we advertise.
Our own public comparison board puts fourteen other synthetic dollars beside us and the comparable
ones pay 91% to 102% of what they advertise. We are last on a board we publish ourselves. The single
distribution the live staking vault has ever received was a 0.10 kUSD plumbing test on 2026-07-09.
The relevance to a lender is limited, since you would be lending USDC against collateral rather than
buying the yield, but a protocol that cannot pay what it advertises may have other things it cannot
do, and you should weigh that.
5. The honest base rate
A permissionless Morpho Blue market with kUSD collateral and USDC loan has existed since 2026-06-16,
id 0xdc6a28b2941b3d32affc71cdeeb10e9779ff2a8ba1dcabc258e7b61f9a11beb3. Supply, borrow and shares are
all zero and lastUpdate still equals its creation timestamp fifty-three days later. It has never
been touched. A permissionless listing with nobody behind it is worth nothing, and we are not going to
present this deployment as if it were worth more.
6. What we are asking for
Nothing that requires a vote, and nothing from Euler Labs or the DAO.
One thing, from a risk curator: read the underwriting memo and tell us the number you would allocate,
including if that number is zero and why. A reasoned zero from someone who underwrites for a living is
more useful to us than silence, and we will publish it either way.
The memo is dated 2026-08-08, states every figure with the call that produces it, leads with the five
facts that argue against us, and quotes the reviewer in topic 1849 verbatim rather than paraphrasing
him: contracts-public/underwriting/kusd-underwriting-memo-2026-08-08.md at main · kerne-protocol/contracts-public · GitHub
Everything above is reproducible from chain. If any number here is wrong, say so in the thread and we
will correct it in the thread, which is what we did the last four times.