What can go wrong

Assets locked on one chain to mint a representation on another are stolen, or the minting authority is subverted, leaving the wrapped tokens unbacked.

Evidence: Determine whether the bridge verifies the source chain with a light client and proofs or trusts an attestation committee, and if it is a committee, how many signers exist, what the threshold is, whether the signers are independent, and whether their keys are…

Most of the network runs a single software implementation, so one bug in that program becomes the network's bug rather than a contained failure.

Evidence: Client distribution dashboards report the share of nodes or stake by execution and consensus client, and the protocol's own thresholds give the reference points that matter, such as the one-third of stake that can delay finality and the two-thirds that can…

A defect in the consensus rules or in their implementation makes nodes disagree about valid history, halting the chain or splitting it into two.

Evidence: Look for a public incident history of halts, deep reorganizations, and emergency releases, and for whether the chain has a documented restart procedure. Client diversity data shows how many independent implementations validate the rules, and time-to-finality…

A hash function, signature scheme, or proving system that a network depends on proves weaker than assumed, undermining ownership, history, or validity.

Evidence: Identify which primitives and curves the chain uses, whether any proving system involved required a trusted setup and how many independent participants took part in the ceremony, and whether the circuits have been independently audited or formally verified.…

The transaction data needed to reconstruct or challenge a chain's state is not published, so users cannot prove what they own or exit on their own.

Evidence: Establish where data is posted, what the retention window is, and who is expected to archive it after that window closes. For committee-based designs, look at the number of members, the signing threshold, and whether members are independent entities. Where…

A protocol reads a price from an external feed, and someone moves that price cheaply in order to trigger borrowing, liquidations, or settlement in their favor.

Evidence: Read the oracle configuration: how many independent sources feed it, whether it is a spot read or a time-weighted average and over what window, what deviation and heartbeat thresholds trigger an update, and what the contract does when the feed goes stale.…

A sufficiently large error-corrected quantum computer would break the elliptic-curve signatures that authorize transactions, though no machine near that scale is known to exist.

Evidence: On-chain data shows how much supply sits at addresses whose public keys are already exposed through reuse or early output types, which is the directly measurable part of this exposure. Check whether the protocol has an upgrade path that allows new signature…

A rollup relies on one operator to order transactions, so if that operator stops or censors, users need a working escape hatch to the settlement layer.

Evidence: Check whether the sequencer is permissioned and who runs it, whether a forced-inclusion mechanism exists in the deployed contracts, what its delay window is, and whether anyone has demonstrably used it. Read who can upgrade the bridge and the proof system,…

A flaw in deployed contract code lets funds be moved, locked, or destroyed in ways the designers never intended.

Evidence: Check whether the deployed bytecode matches published verified source, when the implementation last changed, and how much value the contract has held without incident, since value held multiplied by time live is a cruder but harder-to-fake signal than an…

The data a full node must store and process grows over time, so fewer people can run one and verification concentrates in a smaller set of operators.

Evidence: Chain size, state size, and their growth rates are published, as are the hardware requirements in official documentation and the practical requirements for archive nodes. Count reachable full nodes and look at their geographic and hosting distribution, and…

A protocol upgrade or token migration goes wrong, splitting the network, stranding holders on an old contract, or breaking dependent applications.

Evidence: Read the activation mechanism and the share of nodes or stake signaling readiness before the fork block, and read client release notes, public testnet runs, and shadow-fork results to see how much rehearsal happened. For a token migration, check whether the…