Zcash Zebra CVE-2026-123559

Exploitable against a zebrad node running any release before 7.0.0 in its default configuration when the node is synced close to tip (mempool enabled) AND either accepts inbound peer connections / relays peer transactions (the default) or exposes sendrawtransaction to untrusted callers. No private key, authentication, or non-default setting is required - the invalid signatures need only be well-formed enough to pass standardness and DER-encoding checks before failing. LIMITING FACTORS: nodes still initial-syncing with the mempool disabled are not exposed; the attacker must hold ~1-3 ZEC across ~39,000 small outputs to satisfy ZIP-317 fee headroom (never actually spent, and reusable); impact is confined to Rayon-backed verification (no crash, RPC/peer handling continue slower). No configuration on older releases removes the exposure - upgrade is the only complete fix; limiting inbound peers or not exposing RPC only reduces, not eliminates, exposure.

This is a genuine availability threat whose signals are internally consistent but modest in blast radius. The input CVSS 4.0 vector (AV:N/AC:L/PR:N/UI:N/VC:N/VI:N/VA:H, score 8.7) and the vendor's own CVSS 3.1 (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5) agree: remote, unauthenticated (PR:N), low-complexity, availability-only impact with no confidentiality or integrity effect - so the 'Information Disclosure' tag in the metadata is simply wrong for this CWE-407 algorithmic-complexity DoS. The vuln.today decision label is 'this_week' (not act_now/emergency), and there is no CISA KEV listing and no EPSS score in the input, so there is no evidence of active or widespread exploitation; a knowledgeable reporter (@asafmod) demonstrated a working Regtest proof of concept with scaling measurements and two independent rediscoveries, which raises credibility but is not the same as public weaponized exploit code. Net: a real, prioritize-this-week defect for anyone running a Zebra node near chain tip, bounded by the fact that impact is CPU exhaustion of block verification (no crash, no data exposure, connections preserved) and by Zebra's relatively narrow deployment footprint. The low-attacker-cost, no-ban, reusable-outputs characteristics push real-world risk above what the availability-only impact alone would suggest for exposed nodes.

A peer connects to a synced Zebra node (default behavior) and relays a single ≤2 MB transaction carrying roughly 39,000 transparent pay-to-public-key inputs with well-formed but invalid signatures. The quadratic ZIP-244/ZIP-243 sighash recomputation runs on the shared Rayon thread pool and is not cancelled on rejection, stalling block verification for about 7 minutes on a 4-core node (≈1,700 CPU-seconds measured), and the same outputs can back unlimited distinct invalid transactions with no fee spent and no peer ban. The reporter built and ran a working Regtest proof of concept, though no publicly published exploit code is identified in the provided sources.