Aave resolved a CAPO oracle misconfiguration that triggered wstETH E-Mode liquidations worth $700K. Here’s what happened.
Aave recently dealt with an oracle incident that caught the DeFi community’s attention.
A misconfiguration in its CAPO oracle system triggered unexpected wstETH E-Mode liquidations on the protocol. The event resulted in 345 ETH, roughly $700,000, flowing to liquidators as an excess windfall.
No bad debt was recorded, and service providers have committed to fully reimbursing all affected borrowers.
What Triggered the Aave stETH Liquidation Event
The issue stemmed from Aave’s Correlated Asset Price Oracle, known as CAPO.
According to Chaos Labs co-founder Omer Goldberg on X, CAPO is designed to cap exchange rate values and protect against inflation attacks or price manipulation.
1/ stETH CAPO Misconfiguration
Today, a misconfiguration on Aave's CAPO oracle caused wstETH E-Mode liquidations, resulting in a loss of 345 ETH.
No bad debt was incurred, and all affected users will be fully reimbursed.
More below.
— Omer Goldberg (@omeragoldberg) March 10, 2026
The system uses an offchain Chaos Oracle to compute and submit maximum exchange rate updates, while BGD’s smart contracts serve as the onchain guardrail.
The Chaos Oracle went live with a contract that had been initialized back in early 2024.
That old contract stored a snapshot ratio of 1.15. At 11:46 UTC on the day of the incident, the Chaos Oracle calculated the correct wstETH/stETH snapshot ratio at approximately 1.2282.
The contract rejected the update because it exceeded the 3% growth limit over a 3-day window.
How the Oracle Misconfiguration Caused Artificial Price Depression
After the rejection, the system did what it was built to do. It queried the contract for the maximum allowable value and submitted that figure, approximately 1.19.
Under normal operating conditions, this fallback is the right behavior. But conditions were far from normal.
The snapshot ratio was capped at 1.19, yet the snapshot timestamp still reflected a 7-day-old reference point. CAPO calculates its price cap using the ratio combined with elapsed time.
A low ratio paired with a timestamp assuming a full 7-day growth window produced a cap that sat below the current market rate. That gap artificially depressed the oracle price by roughly 2.85%. Besides, that was enough to push wstETH E-Mode positions across the liquidation threshold.
Goldberg pointed to the root cause clearly: the smart contract applied different constraints to the snapshot ratio and the snapshot timestamp.
The ratio was rate-limited; the timestamp was not. When the two fell out of sync, the system entered an inconsistent state that produced an inaccurate price feed.
Related reading: AAVE Holders Eye Labs’ Bold Proposal as v4 Migration Looms
Aave’s Response and the Reimbursement Plan
Chaos Labs and BGD moved quickly once the issue was identified.
They reduced wstETH borrow caps to 1 on all affected instances and manually aligned the snapshot ratio through the Risk Steward. The CAPO exchange rate was then fully restored.
Aave founder Stani Kulechov also addressed the incident on X, contextualizing its scale.
He noted that the 345 ETH represented approximately 0.00274% of Aave’s total protocol size. He further emphasized that the Chaos Risk Oracles system has processed over 1,200 payloads and 3,000 parameters historically with zero prior incidents.
TL;DR on the stETH CAPO configuration:
– Chaos Risk Oracles is an external tool used within Aave.
– It has historically processed over 1,200 payloads and 3,000 parameters without issues.
– CAPO is a separate defense mechanism designed to protect against inflation attacks or… https://t.co/LcoThpx7p9
— Stani.eth (@StaniKulechov) March 10, 2026
As for reimbursements, Goldberg confirmed that 141 ETH had already been recovered through BuilderNet refunds.
Moreover, Aave DAO service providers are working to finalize a full reimbursement plan. This is for all affected borrowers, using fees previously earned from liquidations.
The team also plans to publish a detailed breakdown of the specific changes made to prevent this class of issue in the future.
