A consensus protocol is only as secure as the cryptographic assumptions beneath it. We assess signature schemes, hash functions, randomness sources, key management, and protocol-specific constructions.
We review the choice of curve, encoding rules, signature aggregation, and replay protection. Ed25519, ECDSA, BLS, and Schnorr each have distinct malleability and verification pitfalls that must be checked against the protocol's usage.
Leader and committee selection often depends on randomness. We evaluate whether the entropy source is manipulable, predictable, or biased, and whether distributed randomness beacons or VDFs are correctly integrated.
Hash functions bind blocks, transactions, and state roots. We check for collision resistance, preimage resistance, domain separation, and canonical serialization that could lead to collisions or replay issues.
Validator keys are high-value targets. We assess generation, storage, rotation, and signing ceremonies, including HSM usage, threshold schemes, and leakage prevention in both hot and cold environments.