Limit Break Payment Processor
Why it happened
Attack class: Access controlA flaw in the shared contract let a trade be treated as accepted by the owner, so old approvals could be spent at a zero price.
What would have stopped it
- Authenticate the real owner on every transfer
- Ship an emergency pause on every version
Not scored before the exploit
BlackHart had not scored Limit Break Payment Processor before this exploit.
Starting September 24, 2026, an attacker used a flaw in Limit Break's Payment Processor to move NFTs and tokens out of wallets that had approved it, without any real sale. DeFiLlama puts the loss at about $1.8 million across Ethereum, ApeChain, Polygon and Base, and other trackers cite higher figures, up to $2.8 million. About 660 WETH could not be secured, while whitehats moved more than 23,155 NFTs worth over $5.7 million to safety. The newer V3 version was paused, but V2 cannot be paused or fixed, so users should revoke their approvals.
The attacker abused standing approvals that wallets had granted to Payment Processor V2. One WETH drain transaction on September 25 took about 281.66 WETH from 25 wallets, and the NFT pulls before it used the same approvals at a zero price.
Attack flow
All 4 on-chain stepsThe attack in 3 steps. Step 1, Root control: Wallets with old approvals (source). Step 2, Drain: Attacker-controlled addresses (attacker), about $1.8M reached this step. Step 3, Rescue: Whitehat custody (exit).
Step 1: Root control
- SourceWallets with old approvalsapproved years earlierNFTs, WETH, other tokens
Step 2: Drain
- AttackerAttacker-controlled addressesNFTs at zero priceabout 281.66 WETH in one transaction
Step 3: Rescue
- ExitWhitehat custodyover 23,155 NFTsworth over $5.7M
Why each step worked
- Spoofable caller identitytrade treated as accepted by the owner
- Standing approvals never expired2024 listings still exposed
- V2 cannot be pausedmissing controlimmutable contract
- About 660 WETH not secured in timemissing controlfungible tokens could not be moved first
Full forensic detail
Step-by-step reconstruction, root cause, counterfactuals, remediation, and disclosure timeline.
Exploit anatomy
Root cause
The Payment Processor is a shared trading contract that users approve once so a marketplace can move their NFTs and tokens when a sale settles. The sources describe a flaw in how V2 decided who was acting: it accepted a caller identity that could be forged, so a trade could be treated as if the owner had agreed to it. With that, an attacker could spend an approval the owner granted long ago, at a zero price, with no signature from the owner and no key theft. The weakness did not sit in any one marketplace. It sat in the shared contract that holds every standing approval, so old listings from 2024 stayed exposed even after Magic Eden dropped V2 in October 2024 and closed its EVM marketplace in early 2026. V3 had the same flaw. V2 is immutable and cannot be paused or patched, which is why the response was rescue and revocation instead of a fix. Limit Break has not published a full post-mortem, so the detail above rests on the statements of Limit Break and 0xQuit as reported by press.
Prevention analysis
A transfer that requires a check bound to the actual owner, rather than a relayed identity, would have stopped zero-price pulls against old approvals.
V3 could be paused within hours. V2 could not, so it stayed usable and whitehats had to move assets by hand.
Approvals limited by time or amount would have left far less exposed from 2024 listings.
Similar incidents
A validation flaw let an attacker spend token allowances users had already granted, leaving standing approvals as the exposure.
Long-lived token approvals let an attacker move assets weeks after the approval was granted.
Remediation
Timeline
Get your protocol scored across 12 dimensions, or request ongoing coverage.