How Does Cetus Protocol Work?
A decentralized exchange on Sui that was hacked for $223M in May 2025 through a math bug in its code. It currently holds about $33M, down from $300M+ before the exploit. Its D grade reflects the severity of the hack, the possibility of the same bug class reappearing, and deeply eroded user trust.
TVL
$33M
Sector
DEX
Risk Grade
D+
Value Grade
D-
Core Mechanisms
AMM/Concentrated-Liquidity
CLMM pools with tick-based liquidity ranges on Sui Move
Concentrated Liquidity Market Maker (CLMM) pools similar to Uniswap V3 but implemented in Move. The May 2025 exploit originated in the integer math library used for liquidity calculations.
Math/Fixed-Point-Arithmetic
integer-mate library for u256 fixed-point math in Move
Uses a third-party math library (integer-mate) for fixed-point arithmetic. A rounding bug in checked_shlw allowed silent overflow, enabling the $223M exploit. The flawed function failed to properly guard against shift-left overflow.
AMM/Standard-Pool
Basic constant-product AMM pools for uncorrelated pairs
Standard x*y=k pools alongside the CLMM offering. Lower capital efficiency but simpler and less exploit-prone than concentrated liquidity pools.
Liquidity/Farming
Liquidity mining rewards with CETUS token emissions
Standard liquidity mining incentives to attract TVL. Post-exploit, these incentives have been insufficient to restore user confidence.
Exchange/Multi-Chain
Deployments on Sui and Aptos blockchains
Cross-deployment on Move-based chains. The exploit only affected the Sui deployment, but shared code raises questions about Aptos deployment security.
Derivatives/Launchpad
Token launchpad with liquidity bootstrapping on Sui
Provides launchpad services for new Sui tokens. The exploit demonstrated that new tokens deployed through Cetus pools were also at risk.
Oracle/TWAP
On-chain TWAP oracle derived from pool prices
Time-weighted average price oracle from pool data. During the exploit, the attacker manipulated reserves to produce distorted TWAP readings.
How the Pieces Interact
The integer overflow bug in the checked_shlw function allowed an attacker to mint positions worth billions of tokens at the cost of only 1 token, draining $223M in under 15 minutes.
The attacker deployed worthless tokens (BULLA, MOJO) into pools, manipulating price curves so the overflow bug could extract legitimate assets (SUI, USDC) far beyond deposits.
Concentrated liquidity positions create capital-efficient but fragile markets where large trades can jump across ticks rapidly, amplifying slippage and impermanent loss.
Shared Move code between Sui and Aptos deployments means a vulnerability discovered on one chain may be exploitable on the other before patches are deployed.
During the exploit, manipulated reserves fed distorted TWAP values to any protocol consuming Cetus price feeds, potentially cascading to downstream DeFi protocols.
What Could Go Wrong
- $223M exploit in May 2025 via integer overflow in liquidity math library (integer-mate)
- Concentrated liquidity positions amplify impermanent loss and manipulation risk on Sui/Move
- Post-exploit TVL collapsed from $300M+ to ~$33M, signaling deep loss of user confidence
Second Integer Overflow Exploit
ElevatedTrigger: A second integer overflow vulnerability is discovered in rebuilt math libraries or other fixed-point arithmetic paths, exploited before the low remaining TVL ($33M) can be withdrawn
- 1.Attacker discovers additional overflow condition in rebuilt integer-mate library or related CLMM math — The same exploit class that enabled the $223M drain is repeatable on remaining TVL
- 2.Attacker mints billions of tokens worth of liquidity positions at negligible cost via overflow — Remaining $33M in legitimate pool assets is drained within minutes
- 3.Sui Foundation faces pressure to invoke validator-level freeze again — Second freeze raises severe decentralization concerns for the entire Sui ecosystem
- 4.CETUS token loses all remaining value as protocol viability is questioned — Token holders face total loss; governance becomes meaningless
- 5.Protocol effectively ceases operation; no recovery possible after second total drain — Users who returned post-first-exploit lose everything; trust is permanently destroyed
Risk Profile at a Glance
Overall: D+ (58/100)
Lower score = safer