Skip to content

Security: ALENOC/electrumx-ravencoin

Security

SECURITY.md

Security model

Trust boundary

ElectrumX serves indexed blockchain data from a configured Ravencoin Core daemon. It does not make a remote Electrum client trust that data. Clients must verify genesis, sampled history, trusted anchors where available, and operator independence. The custom backend RPC is evidence, not consensus proof.

Defenses

  • The server deployment must use a Core identity accepted by the signed safe-Core policy. The historical 4.8.0 threshold is metadata, not sufficient trust, and future releases require their own certification.
  • Server and daemon networks must match.
  • Mainnet block 4,487,775 is checked against the last-unaffected checkpoint.
  • Mainnet KAWPOW nHeight is checked from block 4,487,776.
  • An indexed DB tip that is not on Core's canonical history refuses startup.
  • Backend version/health is refreshed periodically. Broadcast forces a fresh check, preventing a downgrade from retaining transaction relay authority.
  • server.ravencoin_backend returns only sanitized identity, height, network, synchronization, compatibility, and observation-time fields.

Residual risks

A malicious server can lie in its self-reported backend method; a client must not treat it as cryptographic attestation. Dependency compromise, TLS-key compromise, stale server lists, resource exhaustion, and operator compromise remain deployment risks. A signed capability may be added later, but it is not a substitute for independent chain verification.

The unsafe-version override is solely for isolated development. Setting ALLOW_UNSAFE_RAVENCOIN_CORE=1 weakens the startup and periodic version gate, produces critical warnings, and is prohibited for production deployments.

Reporting

Open a private security advisory on the maintained ALENOC fork: https://github.com/ALENOC/electrumx-ravencoin/security/advisories/new. Do not include RPC credentials, wallet material, private keys, or production host details in public reports.

There aren't any published security advisories