Privacy L1 (CryptoNote / Zarcanum)
Immunefi review of the Zano privacy L1 (CryptoNote / Zarcanum lineage) C++ daemon (zanod) and wallet, spanning the admin RPC surface, the mempool acceptance path, the wallet↔daemon transport, and the wallet RPC authentication layer.
The handler calls `purge_transactions()` with no token, header, or caller check, so against an admin-API node bound to a reachable interface any remote party can clear the entire mempool with a single unauthenticated POST — and repeat it to suppress transactions and stall confirmations.
`tx_memory_pool::add_tx` admits transactions with no early upper bound on size, input/output count, or Zarcanum proof complexity, so one crafted oversized transaction (200+ inputs/outputs) is accepted and only then verified — spiking node CPU over 30% and degrading RPC responsiveness; repeated submissions amplify it.
On the `https` daemon path the wallet enables SSL but installs no certificate verification (no CA, hostname, or expiry check), so an attacker on the network path can transparently MITM the wallet↔daemon channel and read or tamper with all RPC traffic.
Consumed JWT anti-replay salts (`m_jwt_used_salts`) live only in process memory, so after a wallet-RPC restart a previously-seen, still-unexpired token plus its exact request body can be replayed to re-execute sensitive methods such as `transfer` and `sweep_below`.