The decentralized finance ecosystem faced a stark reminder of its fragility in June 2024 when UwU Lend, a non-custodial lending protocol, suffered a devastating oracle manipulation attack resulting in losses exceeding $19.4 million. The exploit, which unfolded on June 10 and was followed by a second breach on June 13 that drained an additional $3.7 million, exposed fundamental weaknesses in how DeFi protocols handle price data from external sources.
The Exploit Mechanics
At the heart of the UwU Lend exploit lies a critical vulnerability in how the protocol sourced pricing data for its sUSDe stablecoin markets. The platform employed a median-based oracle system that aggregated prices from 11 different sources. While this approach appears robust on the surface, the implementation contained a fatal flaw: five of those 11 price feeds relied on Curve Finance liquidity pools that could be manipulated with relatively modest capital.
The attacker executed a sophisticated flash loan attack, borrowing approximately 40,000 ETH to manipulate the price feeds. By exploiting the get_p function from Curve Finance pools, which returns the instantaneous spot price without any smoothing mechanism, the attacker depressed the sUSDe price by roughly 4 percent to $0.99. This artificial price depression enabled the attacker to borrow against undervalued collateral. When the oracle price reverted to its true value of $1.03, forced liquidations generated substantial profits for the attacker while UwU Lend and its lenders absorbed the losses.
The attack was amplified by the protocol’s aggressive rehypothecation strategy, where deposited assets were reused as collateral in repeated borrowing and lending cycles. This leverage multiplied the impact of the price manipulation far beyond what a single borrowing transaction would have achieved.
Affected Systems
The primary victim was UwU Lend itself, a decentralized non-custodial liquidity market protocol operating on Ethereum. The protocol’s total value locked dropped significantly following the exploit. However, the ramifications extended to Curve Finance, whose liquidity pools served as the manipulated price source. Curve had explicitly warned developers against using the get_p function for price oracle purposes, but this guidance went unheeded by UwU Lend’s development team.
On-chain security firm Cyvers detected the attack in its early stages when losses stood at approximately $14 million and immediately alerted UwU Lend. Despite this early warning, the attacker continued draining funds until losses surpassed $20 million before the protocol was paused. The attacker’s address (0x841dDf093f5188989fA1524e7B893de64B421f47) and contract (0x21C58d8F816578b1193AEf4683E8c64405A4312E) were identified on Etherscan, with the stolen funds tracked through multiple wallets using MetaSleuth.
The Mitigation Strategy
Preventing this type of oracle manipulation requires a multi-layered approach to price feed design. First, protocols must implement price smoothing mechanisms that calculate time-weighted average prices rather than relying on instantaneous spot values. Curve Finance’s own documentation recommends against using raw spot prices for oracle purposes, and this attack demonstrates exactly why.
Second, the quality of each price source matters more than the quantity. UwU Lend’s median of 11 feeds seemed secure, but when five of those feeds had low liquidity and could be easily manipulated, the median became predictable and exploitable. Protocols should weight their price feeds based on the liquidity depth and manipulation resistance of each source.
Third, flash loan resistance must be a core design principle for any DeFi protocol handling price-sensitive operations. Implementing delayed price updates, multi-block confirmation requirements, and circuit breakers that halt operations when prices move beyond reasonable thresholds can prevent flash loan-driven manipulation.
Lessons Learned
The UwU Lend exploit underscores that DeFi security extends well beyond smart contract code audits. The protocol’s code may have functioned exactly as designed, but the design itself failed to account for the economic attack surface created by its oracle dependencies. With Bitcoin trading at approximately $66,191 and Ethereum at $3,565 on June 15, 2024, the capital available for flash loan attacks in the broader DeFi ecosystem made the $19.4 million loss entirely preventable with proper oracle design.
The second attack on June 13, which stole an additional $3.7 million using tokens acquired during the first breach, reveals another critical failure: insufficient incident response. After the initial exploit, the protocol should have conducted a thorough assessment of all attack vectors that the compromised tokens could enable. Instead, the attacker leveraged the stolen sUSDe tokens to execute a second manipulation against the same oracle vulnerability while the protocol was still processing reimbursements.
User Action Required
For users who interact with DeFi lending protocols, this incident highlights the importance of evaluating oracle security before depositing funds. Research which price feeds a protocol uses, whether those feeds implement time-weighted averages, and whether the protocol has undergone audits specifically focused on economic attack vectors. Protocols that rely on single-source oracles or unsmoothed spot prices present elevated risk regardless of their smart contract audit history. In a market where Bitcoin hovers above $66,000 and total DeFi value locked exceeds $100 billion, the stakes are too high to ignore oracle security.
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Always conduct your own research before engaging with any DeFi protocol.
11 price sources and 5 of them were manipulable curve pools. thats not decentralization, thats just distributed vulnerability
median-based oracle with 5 out of 11 compromised sources means the median itself becomes unreliable. basic statistical failure that should have been caught in design review
5 out of 11 being curve pools means the median was trivially gameable. you only need to shift 1-2 sources to push the median past your liquidation threshold
40,000 ETH borrowed for the attack, approximately $142 million at the time. The sheer scale of flash loan attacks in 2024 is staggering. Over $2.2 billion stolen across all exploits.
the second attack on june 13 for another $3.7m after the first $19.4m drain is the wildest part. how do you not pause everything after the first breach?
first breach drains 19.4m and they leave markets running? the second 3.7m theft is entirely on the team. inexcusable incident response
the second attack is wild. protocol gets drained for $19.4m and nobody thought to pause all markets? thats negligence not just bad luck
5 out of 11 oracle sources being curve pools is the kind of design flaw that should have killed the audit. somebody signed off on this
flash loan attacks in 2024 alone crossed $2.2B. the UwU exploit was sophisticated but the underlying pattern is always the same: cheap oracle + big liquidity = target