Operational Security

Phishing in crypto is not limited to fake login pages. Approval-phishing tricks a user into signing an ERC-20 approval (or Permit2 signature) that grants the attacker unlimited spend on the victim's tokens. The attacker then calls transferFrom at leisure. Address poisoning is subtler: the attacker sends a zero-value or dust transaction from an address that looks similar to one the victim has used before, hoping the victim will copy-paste it as a destination.

These attacks exploit human habits, not cryptographic weaknesses. The defence is procedural: always verify the full address character-by-character, simulate transactions before signing, and revoke stale approvals regularly.

A safe signing workflow adds a verification step between intent and broadcast. Before signing a transaction, simulate it using tools like Tenderly, Rabby's built-in simulator, or a local Foundry fork. The simulator shows you exactly what the transaction will do: which tokens move, which approvals are set, which state changes occur.

For high-value transactions, the gold standard is: draft the transaction on Device A, review the simulation output on Device B, and sign on a hardware wallet (Device C). This three-device workflow defeats most single-point-of-compromise attacks.

A multi-signature wallet requires M-of-N signers to approve a transaction. Safe{Wallet} (formerly Gnosis Safe) is the dominant implementation on EVM chains. A 2-of-3 multisig means any two of three keyholders must sign — eliminating single-point-of-failure for the keys.

Social recovery is a variant: a single key controls the wallet day-to-day, but a set of guardians (friends, family, colleagues) can collectively rotate the controlling key if it is lost. Vitalik Buterin has advocated for this model for individual wallets. The trade-off is operational complexity and the trust you place in your guardians.