Lesson 3 — Drainer attacks from a victim's view: what the signature you almost approved actually does

There are four signature types modern wallets present, and three of them are routinely weaponised. The first is an on-chain transaction — a state-changing call to a smart contract. You see the gas fee, the contract you're calling, and (in modern wallets) a decoded summary. The second is a token approval — a specific kind of on-chain transaction that grants a contract permission to move tokens on your behalf, often for unlimited amounts. The third is an EIP-712 typed signature — an off-chain message with structured data. The fourth is a personal_sign message — an unstructured off-chain message. Drainer attacks use approvals and EIP-712 signatures most heavily because they can be made to look innocent.

The Permit (EIP-2612) and Permit2 patterns are the most-exploited variants. Both let you authorise token transfers via off-chain signature instead of on-chain transaction — convenient because it saves gas, dangerous because the signature carries no transaction fee and no MetaMask 'are you sure?' urgency. A Permit signature, in plain reading, says 'this address can move up to N of my tokens until time T.' Drainers craft these signatures with unlimited amounts and far-future expirations. You sign, they immediately submit the signature to the chain, and the tokens move.

The setApprovalForAll pattern is the NFT-equivalent threat. Most NFT marketplaces and trading interfaces ask you to approve setApprovalForAll(operator, true) on your NFT collection once — after which the operator can transfer any NFT in that collection. Drainer sites that mimic real marketplaces request this approval against an attacker-controlled operator address. Once signed, every NFT in that collection is one transaction away from being moved.

Address-poisoning is the dust-attack variant that catches users who copy-paste from transaction history. The attacker sends you a zero-value (or one-wei) transfer from a vanity address that matches the first and last characters of an address you've genuinely used. The fake address appears in your wallet's 'recent activity.' On a later send, you copy what looks like your friend's address from history — except this time it's the attacker's. The funds go to them. Defence: never copy addresses from transaction history; always re-verify the full address against the original source.

What modern drainer kits do under the hood: a fake site convinces you to click 'connect wallet' and asks you to 'verify your wallet' by signing a message. The signed message is either a Permit (giving them unlimited spending on a specific token) or a setApprovalForAll (giving them every NFT of a specific collection). Within milliseconds of your signature, an automated drainer-as-a-service backend sweeps every approved asset to attacker-controlled addresses, often through privacy mixers in the same transaction batch.

The single most defensive habit is reading the signature payload. In MetaMask and most modern wallets, hovering or expanding the signature request shows the decoded EIP-712 fields. If you see 'spender' set to an unknown address, 'value' set to a number larger than what you intended to spend, or 'deadline' set decades in the future, that is the moment to cancel. Tools like Pocket Universe, Rabby's transaction simulator, and Tenderly's web simulator do this decoding automatically and flag known drainer patterns.