Zebra CVE-2026-44497

Exploitation requires the attacker to construct a transparent transaction output with a custom Bitcoin Script containing specific opcode sequencing: a valid signature check operation (OP_CHECKSIGVERIFY or OP_CHECKSIG) that succeeds, immediately followed by another signature check operation (OP_CHECKSIG) using an undefined ZIP 244 V5 hash-type byte. The target Zebra node must be running version 4.3.1 specifically, which contains the prior GHSA-8m29-fpq5-89jj fix that introduced the incomplete error handling. The vulnerability exists only at the Rust-to-C++ FFI boundary during transparent transaction verification, so exploitation is limited to transactions with transparent inputs (not shielded Sapling or Orchard transactions). Network-accessible attack vector with no authentication required-any peer can submit crafted transactions via standard P2P protocol. Exploitation does NOT require miner cooperation, special node configuration, or user interaction, but practical impact depends on network topology: since most Zcash miners currently run zcashd, malicious blocks created by Zebra nodes accepting invalid transactions would likely be orphaned, limiting attacker profit from double-spends unless targeting specific Zebra-dependent services.

This is a critical consensus-layer vulnerability with potential for coordinated network attacks despite no confirmed active exploitation. The vendor rates severity as Critical due to the consensus failure impact class: blockchain consensus vulnerabilities can enable double-spend attacks, miner revenue loss through orphaned blocks on minority chains, and service disruption across the network. Attack complexity is moderate-exploitation requires understanding of Bitcoin Script opcodes, Zcash transaction structure, and the specific OP_CHECKSIGVERIFY-then-OP_CHECKSIG chaining pattern, but the GitHub advisory provides a complete attack blueprint. Real-world risk is currently mitigated by deployment topology: the advisory explicitly notes most Zcash miners run zcashd rather than Zebra, limiting attacker return-on-investment for consensus attacks since minority-chain blocks would be orphaned. However, any infrastructure depending on Zebra for transaction validation (exchanges, payment processors, layer-2 protocols) faces immediate risk of accepting invalid transactions that the broader network rejects. CVSS metrics are not provided, but consensus vulnerabilities typically score CVSS:3.1/AV:N/AC:L/PR:N/UI:N with integrity impact due to network-accessible exploitation without authentication. The technical disclosure is complete with proof-of-concept transaction structure details, lowering the barrier for weaponization. SSVC decision framework indicates this warrants immediate patching: exploitation status is public (PoC available in advisory), automatable (transaction crafting can be scripted), technical impact is total (consensus failure affects all validation outcomes), and mission prevalence is situational (depends on Zebra adoption in critical infrastructure). Patch availability is confirmed with explicit version 4.4.0 fix, making delayed patching indefensible.

An attacker creates a Zcash transaction spending a transparent output secured by a custom script containing two consecutive signature checks: first an OP_CHECKSIGVERIFY with a valid signature and standard SIGHASH_ALL type, immediately followed by OP_CHECKSIG with an identical signature but an undefined hash-type byte not recognized by ZIP 244. When Zebra processes this transaction, the first opcode executes normally-the Rust callback computes the correct sighash, writes it to the C++ buffer, and signature verification succeeds. The second opcode triggers the callback with the invalid hash type, causing it to return None, but the FFI bridge skips the buffer write. The C++ verifier reads the stale buffer containing the prior valid sighash and verifies the signature against it, accepting the transaction. Meanwhile, zcashd nodes correctly reject the transaction due to the undefined hash type. The attacker broadcasts this transaction to partition the network: Zebra nodes accept it into blocks, zcashd nodes reject those blocks, creating competing chain tips and enabling potential double-spend attacks against services relying on Zebra validation.