Bottom line
The Performance Safe has 9,000.1206 wstETH supplied to Aave V3 Ethereum Prime, 11,010,171.58 GHO of variable debt and 11,078,476.77 GHO of sGHO assets. Current LTV is 40.36% and health factor is 2.0564 — below the 2.25 review boundary in this memo’s operating conditions. The position grew from 2,500.03 wstETH / 2.20m GHO at the 21 August review; the transactions behind the scale-up were not traced in this refresh, and the previous debt-linked sGHO attribution was not maintained. The Safe also holds 18,614.68 wstETH outside Aave; that balance does not support the loan until supplied.
On 1 September Aave governance raised the sGHO target from 4.25% to 4.50% and the GHO base borrow rate from 2.00% to 2.50%. The borrow leg moved twice as much as the savings leg: at the pinned block, sGHO’s 4.50% target APR minus the GHO debt’s 3.7378% APY leaves 76.2bp on GHO debt — below the 100bp review boundary (critical below 50bp). GHO utilisation is 86.09%, above the 85% review boundary; the 92% rate kink is 3.157m GHO of additional venue borrowing away, and crossing it takes the spread to approximately −1,624bp. Including Aave’s 3.70bp wstETH supply APR, the position adds approximately 34.5bp on posted wstETH over holding it. The full exit path (redeem → repay → withdraw) was fork-simulated same-day at block 25,882,194 and cleared with a 68,254 GHO surplus.
The carry remains exposed to Aave rate policy, sGHO funding, wstETH oracle and liquidation mechanics, and concentration of both legs in Aave governance and contracts.
Mechanism and true exposure
This is one collateralised loan, not a recursive loop: the Safe supplies wstETH to Aave V3 Ethereum Prime, borrows GHO and deposits GHO into sGHO. The staking return remains embedded in wstETH; Aave supply interest and the sGHO-minus-debt spread are incremental.
The GHO asset and liability are denomination-matched, but not operationally netted. sGHO sits outside the Aave account and does not improve its health factor. Aave’s oracle hard-prices GHO at $1, so a secondary-market GHO depeg does not change the loan’s health factor; the wstETH oracle value remains the liquidation driver.
Balances and rates are pinned to Ethereum block 25,883,293 (1 September 2026, 15:20 UTC). The Safe’s sGHO is shown as one balance; per-clip attribution to the loan was not maintained after the August scale-up.
Net economics, benchmark and capacity
| Live position | Value |
|---|---|
| wstETH supplied / GHO debt | 9,000.1206 wstETH / 11,010,171.58 GHO |
| Total sGHO assets | 11,078,476.77 GHO |
| Current LTV / health factor | 40.36% / 2.0564 |
| GHO borrow APR / APY | 3.6697% / 3.7378% |
| sGHO target APR | 4.50% |
| Spread on GHO debt | 76.2bp |
| Aave wstETH supply APR | 3.70bp |
| Incremental carry on posted wstETH | approximately 34.5bp |
The position was first opened in two clips: 500,000 GHO on 7 August and 1,700,000 GHO on 13 August. Between 21 August and 1 September the supplied collateral rose to 9,000.12 wstETH and the debt to 11.01m GHO while the Safe’s sGHO fell from 12.405m to 11.078m GHO; the transactions behind these changes were not reconstructed in this refresh.
The 21 August venue sweep (Aave V4 Bluechip/Prime Hub 2.722%, Aave V3 Prime 2.972%, Aave V4 Core 3.612%, Aave V3 Core 3.750%, Fluid 5.630%) was not refreshed in this review; at this block the live venue’s rate is 3.6697% APR. The venue comparison must be re-run before any migration or new clip.
The GHO interest-rate model on Prime, read from the rate strategy contract at this block: 2.50% base (raised from 2.00% on 1 September), 1.25% slope to the 92% optimal utilisation, 35% slope above it. Utilisation is 86.088% (45.974m GHO debt against 53.403m of supply-side capacity). Additional venue borrowing reprices all existing debt:
| Additional GHO debt | Utilisation | Borrow APR / APY | Spread to sGHO | Carry on wstETH at 40.4% LTV, incl. supply APR |
|---|---|---|---|---|
| 0 | 86.09% | 3.6697% / 3.7378% | 76.2bp | 34.5bp |
| 1m | 87.96% | 3.6951% / 3.7642% | 73.6bp | 33.4bp |
| 2m | 89.83% | 3.7206% / 3.7907% | 70.9bp | 32.3bp |
| 3.157m | 92.00% | 3.7500% / 3.8212% | 67.9bp | 31.1bp |
| 5m | 95.45% | 18.8425% / 20.7346% | −1,623.5bp | −651.5bp |
Exit, cost and latency
At the pinned block the Safe can redeem all 11,078,476.77 GHO of sGHO assets (coverage 1.0062× of the debt) and the Aave reserve has 42,737.07 wstETH available against the Safe’s 9,000.12 aWstETH (4.75×).
The normal exit is: redeem sGHO, repay the full GHO variable debt, withdraw all wstETH. This sequence was fork-simulated same-day at block 25,882,194 and executed: 11,078,279.40 GHO redeemed, debt repaid to zero, all wstETH withdrawn, 68,254.36 GHO of surplus remaining with the Safe. The simulation ran the three calls sequentially from the Safe; the single MultiSend batch encoding was not built. It needs three of the Safe’s six owner signatures. The contracts charge no exit fee; gas is the remaining direct cost.
Atomicity prevents a partial unwind, but does not guarantee execution. An sGHO pause or funding shortfall, an Aave reserve pause, or another supplier consuming wstETH liquidity first can make the sequence fail. A DEX-assisted fallback adds GHO basis, slippage, MEV and routing risk.
At current debt, a 2.74% fall in Aave’s wstETH collateral value takes health factor to 2.0 (the critical boundary) and a 51.37% fall reaches liquidation. The 18,614.68 wstETH held outside Aave is operational liquidity, not on-chain protection against those moves.
Risks
| ID | Risk | Current evidence | Failure surface |
|---|---|---|---|
| R1 | Rate and policy | 76.2bp debt spread, below the 100bp review boundary; the 1 September governance change raised the sGHO target 25bp and the GHO base rate 50bp, a net −27bp to the spread; utilisation 86.09% against the 92% kink | Aave can change the borrow curve or sGHO target; utilisation can reprice all venue debt |
| R2 | sGHO funding | 106,177 GHO excess, approximately 5.3 days of target yield at current assets and the raised 4.50% target (9.3 days on 21 August) | Yield needs continuing GHO top-ups; withdrawals are first-come if the vault is underfunded |
| R3 | Liquidation and oracle | HF 2.0564, below the 2.25 review boundary; 2.74% collateral-value fall reaches the 2.0 critical boundary; oracle wstETH ratio matches stEthPerToken() at the snapshot | sGHO is not collateral; a price or oracle error can liquidate wstETH before repayment |
| R4 | Exit liquidity | 1.0062× GHO and 4.75× wstETH immediate coverage; full exit fork-simulated same-day at block 25,882,194 | Coverage can disappear between approval and execution; pause states can block the sequence; GHO coverage is 0.62% above 1× |
| R5 | Contract and governance concentration | Collateral, debt and savings all depend on Aave; staking value depends on Lido | A common Aave control or implementation failure affects both sides of the carry |
| R6 | Wallet control | Safe v1.4.1, three-of-six threshold, six bare EOAs (codesize checks 28 August), no modules | Three owner keys can move the position; the signer compromise graph was not assessed |
Aave’s March 2026 wstETH CAPO parameter mismatch produced an approximately 2.85% effective exchange-rate fall and liquidated about 10,938 wstETH across 34 accounts, without bad debt. It is direct evidence that an internal-rate oracle can become a liquidation surface (Aave post-mortem, accessed 21 August 2026). At the current health factor, a repeat of that 2.85% move exceeds the 2.74% distance to the 2.0 critical boundary.
Ordinary Aave protocol updates pass through a one-day executor timelock; core governance changes use the long executor’s seven-day timelock (Aave governance documentation, accessed 21 August 2026). The minimum key or governance compromise required to exercise each relevant emergency and upgrade power was not re-derived in this run.
Operating conditions
These are analytical tripwires, not evidence that runtime monitoring is deployed. Boundaries are unchanged from the 21 August review; three are currently crossed.
| Measure | Current | Review boundary | State |
|---|---|---|---|
| Aave health factor | 2.0564 | below 2.25; critical below 2.0 | crossed |
| Prime GHO utilisation | 86.09% | above 85%; rate cliff at 92% | crossed |
| sGHO target minus debt APY | 76.2bp | below 100bp; critical below 50bp | crossed (1 September rate changes) |
| Total sGHO / GHO debt | 1.0062× | below 1× | within |
| sGHO pause | false | true | within |
| GHO and wstETH exit coverage | 1.0062× / 4.75× | either below 1× | within |
| Aave wstETH ratio vs wrapper ratio | effectively equal | divergence above 0.5% | within |
The coverage measures are reported on the Safe’s total sGHO because per-clip debt attribution was not maintained through the untraced August scale-up.
Evidence boundary
Balances, account data, reserve state, rate-model parameters, oracle values, Safe configuration and sGHO state were read at Ethereum block 25,883,293, hash 0x192a3398dd935d2b28119577480bc7e023a08af52ff34902678aa03ac528ea75 (1 September 2026, 15:20 UTC), via the strategy’s scripts/verify-monitor.ts plus direct reads of getUserAccountData and the GHO interest-rate strategy — including the same-day governance changes (sGHO target 425 → 450, GHO base rate 2.00% → 2.50%). The full exit sequence was executed on an anvil fork of same-day block 25,882,194, before those rate changes. The 21 August review’s pinned state is preserved in the git history; snapshot.json now holds this pin.
Addresses and interfaces are unchanged from the 21 August review (official Aave address books, sGHO documentation, Safe contracts; access dates as cited there).
Not verified in this run: the transactions behind the 21 August → 1 September position growth and sGHO reduction (attribution of debt to sGHO clips lapsed with it); the venue comparison sweep (21 August values retained, marked stale); historical rate persistence; stressed exit depth; the exact Safe MultiSend batch encoding; gas and MEV cost; the Lido staking APR; runtime monitor deployment; and the minimum Aave governance or emergency-control compromise set. The D2 source beside this memo reproduces the capital-flow chart.