A treasury health check, in full, on a $116M public treasury.

How to check this document

Nothing here asks to be believed. Every number is a return value from a public Ethereum node at one specific block, and the command that produces it sits next to it. If a figure below is wrong, the command beside it is how you find that out in under a minute.

Valuation conventions, stated before the numbers

• •Liquid staking tokens are converted to their ETH equivalent using each protocol's own on-chain redemption rate, then valued at the Chainlink ETH/USD feed. This is a redemption-rate valuation, not a market-price one. Secondary market prices can sit below redemption value, and this document does not read them.
• •Stablecoins are valued at exactly $1.0000 per unit. This is a convention, and it is generous by a measurable amount: the Chainlink USDC/USD feed reads 0.99980963 at this block, so the operating wallet's USDC is overstated by $1,437.38 against that feed.
• •stETH is treated as 1 ETH. The Chainlink stETH/ETH feed reads 1.000133577293157200 at this block, so this convention understates the stETH position by $263.04.
• •ENS is valued at the Chainlink ENS/USD feed. That feed had not updated for 20h 34m at the block, which is inside its heartbeat but is the single stalest input in this document. Every ENS-denominated figure moves with it.
• •Airdropped tokens with no price feed are excluded from every total and are addressed separately under hygiene. Excluding them is a decision, not an oversight.
• •Every total is computed from raw integer values and rounded once, at the end, while each row is rounded independently for display. So adding a displayed column can land a cent away from the displayed total, and where it does the total is the right figure. It happens twice here: the ETH bucket carries a WETH dust line worth $0.000192 that prints as $0.00, and the stablecoin bucket rounds down across its rows. Both are stated rather than papered over, because a reader who checks the addition should find the explanation already written down.

Age is the block timestamp minus the feed's updatedAt. Every one of these is inside its published heartbeat; the ENS/USD and stETH/ETH ages are stated because a reader pricing a 35% ENS position should know how fresh that price is.

cast call 0x5f4eC3Df9cbd43714FE2740f5E3616155c5b8419 \
  "latestRoundData()(uint80,int256,uint256,uint256,uint80)" \
  --rpc-url https://eth.merkle.io --block 25721000

1. The perimeter, and how each address was proven

A treasury map is only as good as its address list, and an address list taken from a block explorer's label is a rumour. ENS publishes its own addresses as ENS names, which makes the enumeration self-proving: the DAO controls the name, so the name is the DAO's own statement about which address is which.

cast resolve-name wallet.ensdao.eth      --rpc-url https://eth.merkle.io
cast resolve-name endowment.ensdao.eth   --rpc-url https://eth.merkle.io
cast resolve-name controller.ens.eth     --rpc-url https://eth.merkle.io

# and the same answer from a completely different direction:
cast call 0x323A76393544d5ecca80cd6ef2A560C6a395b7E3 "timelock()(address)" \
  --rpc-url https://eth.merkle.io --block 25721000
# -> 0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7, matching wallet.ensdao.eth

2. What is held, and where

The DAO wallet: wallet.ensdao.eth

The operating treasury. Everything here is held directly, in the timelock contract itself, and none of it is deployed into a yield venue.

Total $57,512,736.39. The ENS position is 9.6294% of the 100,000,000 total supply, held by the DAO in its own token.

W=0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7
cast balance $W --rpc-url https://eth.merkle.io --block 25721000
# -> 4499789208485555923108

cast call 0xC18360217D8F7Ab5e7c516566761Ea12Ce7F9D72 "balanceOf(address)(uint256)" $W \
  --rpc-url https://eth.merkle.io --block 25721000
# -> 9629421596646019731884049

cast call 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48 "balanceOf(address)(uint256)" $W \
  --rpc-url https://eth.merkle.io --block 25721000
# -> 7550477216881

The endowment, ETH leg: endowment.ensdao.eth

Receipt tokens are converted to their ETH equivalent using each protocol's own rate function, read at the same block. The rate column is the redemption rate per whole token, so the arithmetic in the fourth column is one multiplication a reader can repeat.

Total 16,577.573912 ETH equivalent, $31,852,979.39. Rates are redemption rates read on chain, not market quotes.

E=0x4F2083f5fBede34C2714aFfb3105539775f7FE64

# ether.fi weETH: balance, then the rate that turns it into eETH
cast call 0xCd5fE23C85820F7B72D0926FC9b05b43E359b7ee "balanceOf(address)(uint256)" $E \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 7449133773802619493417
cast call 0xCd5fE23C85820F7B72D0926FC9b05b43E359b7ee "getEETHByWeETH(uint256)(uint256)" 1000000000000000000 \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 1101127921644879141

# Stader ETHx: the rate lives on the pool manager, and the manager is derived,
# not assumed. ETHx.staderConfig() -> StaderConfig.getStakePoolManager().
cast call 0xA35b1B31Ce002FBF2058D22F30f95D405200A15b "staderConfig()(address)" \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 0x4ABEF2263d5A5ED582FC9A9789a41D85b68d69DB
cast call 0x4ABEF2263d5A5ED582FC9A9789a41D85b68d69DB "getStakePoolManager()(address)" \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 0xcf5EA1b38380f6aF39068375516Daf40Ed70D299
cast call 0xcf5EA1b38380f6aF39068375516Daf40Ed70D299 "getExchangeRate()(uint256)" \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 1094783933789241951

The endowment, stablecoin leg

Where a position is an ERC-4626 vault, the figure below is the vault's own answer to how many underlying units the holding is worth, not a number this document computed. Where it is a rebasing receipt, the balance already is the underlying.

Total $26,995,714.83. Deriving these from convertToAssets independently reproduces the vault-reported figures exactly for fUSDC, fUSDT and fGHO, and to twelve decimal places for sUSDS.

# Fluid fUSDC: shares held, then what the vault says they are worth
cast call 0x9Fb7b4477576Fe5B32be4C1843aFB1e55F251B33 "balanceOf(address)(uint256)" $E \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 8210632302044
cast call 0x9Fb7b4477576Fe5B32be4C1843aFB1e55F251B33 "maxWithdraw(address)(uint256)" $E \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 9938670831761

# Aave and Compound receipts are already denominated in USDC
cast call 0x98C23E9d8f34FEFb1B7BD6a91B7FF122F4e16F5c "balanceOf(address)(uint256)" $E \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 2501043736715
cast call 0xc3d688B66703497DAA19211EEdff47f25384cdc3 "balanceOf(address)(uint256)" $E \
  --rpc-url https://eth.merkle.io --block 25721000          # -> 1500550842665

3. Concentration, stated as numbers rather than colours

The single largest line is the DAO's own token, at 35.51% of mapped value. That is the structural exposure every token-governed treasury carries, and it is the one number in this document that is not a criticism: a DAO cannot diversify out of its own token without selling it, and selling it is a governance question rather than a treasury one. What matters for a health check is that it is stated, sized, and that the remaining 64.49% is not correlated to it by accident. Here it is not: the non-ENS balance is roughly half ETH and half dollars, and both halves are held in instruments that redeem rather than in positions that need a buyer.

Counterparty and venue exposure

Every venue below is a distinct set of smart contracts with a distinct failure mode. Percentages are of the endowment, then of total mapped value, because those two denominators answer different questions.

Sums to $58,848,694.22, the endowment total, to the cent. Largest single venue 26.78% of the endowment; top three 72.88%.

4. Who can move this money

This is the section a treasurer should read first and the one most treasury dashboards do not have. Holdings tell you what is at risk; the control surface tells you who has to agree before it moves. The role set below was not assembled by checking addresses we already suspected. It was derived by replaying every access-control event the timelock has ever emitted, which is the only method that can find a role holder nobody thought to look for.

Eight events, from the timelock's deployment in November 2021 to the pinned block. This is the complete access-control history of the contract, not a sample.

cast rpc eth_getLogs '{
  "address": "0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7",
  "topics": [[
    "0x2f8788117e7eff1d82e926ec794901d17c78024a50270940304540a733656f0d",
    "0xf6391f5c32d9c69d2a47ea670b442974b53935d1edc7fd64eb21e047a839171b"
  ]],
  "fromBlock": "0x0", "toBlock": "0x1887968"
}' --rpc-url https://gateway.tenderly.co/public/mainnet

# topic[0] values are keccak("RoleGranted(bytes32,address,address)") and
# keccak("RoleRevoked(bytes32,address,address)"). Confirm each result with:
cast call 0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7 "hasRole(bytes32,address)(bool)" \
  $(cast keccak "PROPOSER_ROLE") 0x323A76393544d5ecca80cd6ef2A560C6a395b7E3 \
  --rpc-url https://eth.merkle.io --block 25721000          # -> true

The endowment's own control surface

The endowment is a Safe, but not the kind of Safe most treasuries run. Its owner set has exactly one entry, and that entry is the DAO timelock.

cast call $E "getOwners()(address[])"   --rpc-url https://eth.merkle.io --block 25721000
# -> [0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7]   (the DAO timelock, and only it)
cast call $E "getThreshold()(uint256)" --rpc-url https://eth.merkle.io --block 25721000   # -> 1
cast call $E "VERSION()(string)"       --rpc-url https://eth.merkle.io --block 25721000   # -> "1.3.0"

cast call $E "getModulesPaginated(address,uint256)(address[],address)" \
  0x0000000000000000000000000000000000000001 20 --rpc-url https://eth.merkle.io --block 25721000
# -> [0x703806E61847984346d2D7DDd853049627e50A40,
#     0xCFbFaC74C26F8647cBDb8c5caf80BB5b32E43134]

AM=0xCFbFaC74C26F8647cBDb8c5caf80BB5b32E43134
cast call $AM "getDelegates(address,uint48,uint8)(address[],uint48)" $E 0 20 \
  --rpc-url https://eth.merkle.io --block 25721000
# -> [0x91c32893216dE3eA0a55ABb9851f581d4503d39b]

cast call $AM "getTokenAllowance(address,address,address)(uint256[5])" \
  $E 0x91c32893216dE3eA0a55ABb9851f581d4503d39b \
  0x0000000000000000000000000000000000000000 \
  --rpc-url https://eth.merkle.io --block 25721000
# -> [30000000000000000000, 0, 36000, 29761613, 26]
#     amount, spent, resetTimeMin, lastResetMin, nonce

5. How much of it could actually move today

A holdings table says what is owned. A liquidity ladder says what could be turned into dollars before the end of the day, which is the number that matters the moment anything goes wrong. Every figure in the exitable column is the venue's own answer at the pinned block, not an estimate.

Stablecoin leg: $19,472,772.28 of $26,995,714.83 exitable in a single block, 72.13%. The $7,522,942.55 that is not exitable is one position.

V=0x4Ef53d2cAa51C447fdFEEedee8F07FD1962C9ee6
cast call $V "maxWithdraw(address)(uint256)"   $E --rpc-url https://eth.merkle.io --block 25721000  # -> 0
cast call $V "canSendShares(address)(bool)"    $E --rpc-url https://eth.merkle.io --block 25721000  # -> true
cast call $V "totalAssets()(uint256)"             --rpc-url https://eth.merkle.io --block 25721000  # -> 20373297502416
cast call 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48 "balanceOf(address)(uint256)" $V \
  --rpc-url https://eth.merkle.io --block 25721000                                                  # -> 0

6. The idle cash line

The operating wallet holds 7,550,477.216881 USDC and 4,499.789208485555923108 ETH, both sitting directly in the timelock contract, neither deployed anywhere. Together that is $16,196,597.19, or 13.92% of mapped value, earning nothing.

The benchmark used below is deliberately not a Kerne product. It is the Sky Savings Rate, which this same treasury is already earning on $2,008,180.54 of sUSDS in its own endowment, read from the same chain at the same block. Comparing an idle balance to a rate the owner has already chosen elsewhere is the only comparison that does not smuggle in a recommendation.

The 3.5200% figure is computed from the Sky Savings Rate ray read on chain, 1000000001096988989836188433, compounded over 31,536,000 seconds. The ETH balances are not benchmarked because staking them changes their liquidity profile, and a health check that quietly assumes a treasury wants that has stopped measuring and started advising.

cast call 0xa3931d71877C0E7a3148CB7Eb4463524FEc27fbD "ssr()(uint256)" \
  --rpc-url https://eth.merkle.io --block 25721000
# -> 1000000001096988989836188433   (per-second rate, 27 decimals)
# APY = (ssr / 1e27) ** 31536000 - 1 = 3.5200%

7. Where the balance has been

One balance is a snapshot. Five are a shape. Each row below is one historical read at a specific block, so the trend is as reproducible as the snapshot.

Blocks chosen at roughly 12, 6, 3 and 1 month before the pinned block.

for B in 23093000 24425000 25073000 25505000 25721000; do
  cast call 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48 "balanceOf(address)(uint256)" \
    0xFe89cc7aBB2C4183683ab71653C4cdc9B02D44b7 --rpc-url https://eth.merkle.io --block $B
done

Two things fall out of that table. ETH accumulates steadily and almost monotonically, which is what a treasury funded by a fee stream looks like. The dollar balance does not: it fell to $379,665.36 in February 2026, then was refilled twice. That is a lumpy replenishment cycle rather than a running balance, and the useful question it raises is whether the February low was intentional. A treasury that plans to hold a floor of operating dollars and got within a rounding error of zero has learned something about its own process. A treasury that meant to run that lean has learned nothing and needs no change. Public data can show the shape. It cannot tell you which of those two happened, and this document does not pretend otherwise.

8. Housekeeping, including one thing that would break a naive dashboard

Besides ENS, USDC and WETH, the DAO wallet shows a balance in 58 other token contracts, none of which it asked for. Fifteen of the 58 carry a web address or a claim-your-reward lure in their own name, which is a nuisance rather than a risk as long as nobody ever interacts with them. One of the remaining forty three is worth naming.

cast call 0x91235900178f3Ce970028fAd5f1Be25Cd385A2E4 "name()(string)" \
  --rpc-url https://eth.merkle.io --block 25721000     # -> "Tether USD"
cast call 0x91235900178f3Ce970028fAd5f1Be25Cd385A2E4 "decimals()(uint8)" \
  --rpc-url https://eth.merkle.io --block 25721000     # -> 18
cast call 0xdAC17F958D2ee523a2206206994597C13D831ec7 "decimals()(uint8)" \
  --rpc-url https://eth.merkle.io --block 25721000     # -> 6   (the real one)

There is no action implied here. Unsolicited tokens cannot be refused, sweeping them costs gas for no benefit, and interacting with them is the actual risk. The finding is about how the treasury is measured, not about what it holds.

9. Allocation memo, dated 2026-08-09

What the numbers say

1. 1. The control architecture is stronger than most treasuries of this size, and the evidence is on chain rather than in a policy document.

   A Safe whose only owner is a 2-day timelock cannot be drained by a compromised signer, because there is no signer. A manager that operates through a permission modifier rather than as an owner cannot exceed the permissions the DAO scoped. A security council that can cancel but has no bytecode path to schedule cannot become an attacker. And the previous council's term expired and its role was actually revoked on chain, which is the difference between a governance control and a governance intention.

2. 2. The concentration that dominates the balance sheet is the one nobody can fix, and it is disclosed rather than hidden.

   35.51% in the DAO's own token is structural. The meaningful question is not whether to hold it but whether the other 64.49% would survive the same shock that hit it, and here the answer is reasonable: the non-ENS balance is split roughly evenly between ETH-denominated and dollar-denominated assets, and the dollar half is spread across five protocols with no venue above 22.88% of the endowment.

3. 3. Same-day dollar liquidity is 72.13% of the dollar leg, and the gap is one position with a known second step.

   $19.47M of the $27.00M stablecoin leg exits in one block. The remaining $7.52M needs a deallocation first. That is a design choice of the vault rather than a defect, but it is the kind of fact that should be written down before it is needed rather than discovered during a week when it is.

4. 4. The staking allocation is concentrated away from the market leader, and that is a policy question this document cannot answer.

   ether.fi and Stader together are 50.00% of the endowment against Lido at 3.35%. There are defensible reasons to underweight the dominant provider and defensible reasons not to. What a treasurer should want is for that split to be the result of a decision that was written down, rather than the residue of a series of individually sensible allocations.

5. 5. $16.20M sits idle, and the cost of that is $265,777 a year at a rate this treasury already accepts elsewhere.

   This is the only line in the memo with a dollar figure attached to inaction, and it is also the line most likely to be correct-as-is. Operating cash is supposed to be boring. The number is offered so the choice can be made deliberately rather than by default.

The case against everything above

A memo that only argues one way is marketing. Here is the strongest version of the opposite case, which a reader should weigh at least as heavily.

• •Every observation here is a snapshot of one block. A treasury is a moving thing, and a manager who rebalanced yesterday will look different tomorrow. Nothing in this document establishes a pattern, because one block cannot.
• •The idle-cash figure assumes the balance would otherwise sit still for a year. If it is working capital with a known drawdown schedule, the real foregone yield is a fraction of $265,777 and possibly close to zero.
• •The liquidity ladder measures venue capacity, not market capacity. Withdrawing $19.47M from lending markets in one block is possible per the contracts; doing it without moving rates against yourself is a different question this document does not touch.
• •The Roles Modifier scope is unmeasured. If that scope is wide, the 2-of-8 manager pod is a materially larger control surface than the structure suggests, and this document would have understated it. It reports the shape it can prove and flags the part it cannot.
• •An outside reader with no access to the subject's plans will always find things that look like gaps and are actually decisions. The value of a health check is that it produces a list worth arguing with, not a list that is right.

10. What this document could not determine

Every one of these is a boundary rather than an omission. A memo that does not have this section has one and is not telling you.

• •Income. Registration revenue arrives at the DAO wallet as an internal ETH transfer, which emits no event and is not readable from a public node without trace access. No revenue or runway figure appears anywhere above.
• •Liabilities. Grant commitments, streams, vesting and anything agreed in a forum post are invisible to state reads. Every figure here is an asset figure.
• •The exact permission scope of the Zodiac Roles Modifier, which is not retrievable in a single call.
• •Withdrawal-queue depth at Lido, Stader and ether.fi, and therefore the true same-week liquidity of $31.85M of ETH-denominated assets.
• •The underlying allocation inside the two Morpho Vault V2 positions. The vaults are readable, and following them one layer deeper is a larger piece of work than a fixed-tier check contains.
• •Whether any observation here matters, which depends entirely on intent that public data does not carry.