zkLink Nova suffered a $805K exploit via rollup batch-root forgery, allowing an attacker to forge L2-to-L1 withdrawals and drain the L1 bridge escrow without corresponding L2 burns.
Security & Exploits ·
zkLink Nova experienced an $805,000 exploit on October 28 involving a forged rollup batch root that allowed an attacker to drain the Layer 1 bridge escrow without corresponding Layer 2 token burns. The attack exploited an owner-gated contract (0xd5f1...ff07) to install a fraudulent L2 logs root through zkLink's gateway, then repeatedly called the L1 bridge's finalize withdrawal function with forged Merkle proofs that validated against the fake root. The attacker at 0xaca0655a withdrew approximately 245.74 pufETH ($635K), 22.43 ETH ($52.8K), various other liquid staking tokens, and stablecoins totaling around $805K from the escrow.
This type of attack mirrors output-root and withdrawal-forgery vulnerabilities previously seen in Optimism Stack systems, where attackers circumvent the standard two-way peg mechanism by forging the cryptographic proofs that attest to L2 transactions. The vulnerability required only the ability to inject a malicious batch root into the synchronization process, after which legitimate-looking but false withdrawal proofs could be generated at will.
It remains unclear whether the owner-gated harness was left intentionally unguarded, misconfigured, or compromised, and no public technical post-mortem has detailed the root cause or whether other funds on zkLink Nova remain at risk.