zkLink Nova Bridge Drained of About $786,000

An attacker has pulled ether and a batch of tokens out of the Ethereum-side escrow of zkLink Nova’s bridge, after feeding it a forged record of what had happened on the network. Blockchain security firm Blockaid flagged the exploit early on 29 October and put the losses at about $786,000. SlowMist, another security firm, estimated about $783,100.

Bridges hold the assets that users lock on one chain so they can use them on another, which makes their escrow contracts some of the most valuable targets in crypto. In this case, Blockaid says the attacker did not break the escrow directly. They tricked it into accepting withdrawals that were never genuinely authorised. zkLink, the project behind Nova, had not commented on X by the morning of 29 October.

How Blockaid says the attack worked

Blockaid’s alert on X, posted at 02:24 GMT on 29 October, described the exploit as still in progress. The attacker “used a public relayer to inject a forged batch root on Ethereum,” it said, “then finalized ETH and ERC20 withdrawals from zkLink Nova L1 escrow to attacker-chosen receivers.” It added: “~$786k drained from L1 ETH escrow and ERC20 bridge so far.”

A batch root is a compact fingerprint of a block of activity on the Layer 2 network. The Ethereum contracts trust it as proof of which withdrawals are valid. If a fake root is accepted, the contracts will release funds against withdrawals that the root says exist, even if they were never made. Blockaid listed the attacker and contract addresses in a follow-up post.

SlowMist’s analysis goes further. It says a function in zkLink’s Arbitrator contract, claimMessage, failed to clear a temporary value, finalizeMessageHash, after use. According to SlowMist, that stale value could then be reused through a public relayer contract to push an arbitrary batch root into the main zkLink contract. That is SlowMist’s reading of the code. zkLink has not confirmed it.

What the Ethereum record shows

TSN read the transactions directly from Ethereum. The main withdrawal came at 22:21:35 GMT on 28 October, in block 26,294,188, four hours before Blockaid’s alert. In that single transaction, 11 different tokens left zkLink’s ERC-20 bridge contract. The largest amounts were 245.74 pufETH, a liquid restaking token, about 18,113 USDC and about 3,361 USDT. The rest included smaller amounts of ezETH, rsETH, STONE, wrapped bitcoin and several other ether-based tokens.

The chain also shows a second transaction at 01:43 GMT on 29 October, from a different address. It moved dozens of further tokens out of the same bridge contract, mostly in tiny amounts. The biggest by count was about 216 million ZKL, zkLink’s own token, alongside about 2,133 GAL and small amounts of USDe and other tokens.

Blockaid’s figure also covers ether taken from the separate ETH escrow. When TSN checked the escrow contract on 29 October, it held about 0.0025 ETH. TSN could not independently confirm how much ether was taken, so the totals here rely on Blockaid’s and SlowMist’s estimates. Both are early figures that may change as prices move and the full set of transfers is counted.

A bridge exploit with no word from the bridge

This is TSN’s analysis. The most striking thing about the zkLink Nova exploit is the silence. By the morning of 29 October, the project’s official X account had not posted since 26 September, so users learned about the loss from security firms rather than from the team that runs the bridge. For anyone with funds still in Nova, that leaves basic questions unanswered: whether withdrawals are paused, whether the remaining assets are safe, and whether there is any plan to make users whole.

The method also shows why bridges stay a weak point. The escrow did what it was designed to do: it paid out against a root it accepted as valid. Blockaid’s description puts the failure upstream, in the path that let a forged root be accepted at all. If SlowMist is right, the flaw was a value that was not cleaned up after use. That small mistake would have given anyone who noticed it a way to rewrite what the bridge believed. TSN has covered similar trust failures, from a Tranche vault drained of $1.01m on Avalanche to the X100 governance exploit, and the loss here is small by those standards. The pattern is not.

Response speed matters too. The main withdrawal came about four hours before Blockaid’s public alert, and the second came after more than three hours. Blockaid has tracked how much stolen crypto ends up beyond anyone’s reach to freeze. Most of the tokens taken here, apart from USDC and USDT, have no central issuer able to freeze them. Bridge operators elsewhere, such as Zilliqa after its X-Bridge incidents, have shut bridges and gone public quickly. zkLink’s users are still waiting.

What is still unknown

zkLink has not confirmed the exploit, the amount, the cause or whether it has paused the bridge. There is no word on recovery efforts, and no reports that funds have been frozen or returned. TSN will update this story if zkLink responds. Other incidents are tracked on TSN’s crypto hacks hub.