FinWiki
日EN中

ERC-4337 Overview · Application-Layer Implementation of Account Abstraction

Confidence: Certain Updated 2026-05-26 Review by 2026-09-21 Sources 1 Machine-translated Original (JA)
#systems#wallet#aa#account-abstraction#erc-4337#ethereum
On this page

Wiki route

This entry sits under systems index. Read it against ERC-7702 vs ERC-4337 · Ethereum AA dual-track comparison for peer / contrast context and fintech index for the broader system / regulatory boundary.

Key facts

  • Enabled on Ethereum mainnet in 2023-03
  • No protocol-layer change · implemented purely at the application layer
  • EntryPoint is an on-chain singleton contract (0x0000000071727De22E5E9d8BAf0edAc6f37da032)
  • The v0.7 upgrade (2026) simplified mempool rules · reduced gas overhead by 20-30%
  • Supports custom signatures such as WebAuthn / passkey / BLS / secp256r1 (iOS Secure Enclave)

Mechanism / How it works

The core problem solved by ERC-4337 : the capability gap between EOA and SCW. A traditional EOA is controlled by a secp256k1 private key · can only send transactions / pay gas / sign · and has no programmable logic. SCW (Safe/Argent) is programmable, but it must be triggered by a tx issued by an EOA —— users still need to hold ETH to pay gas and manage private keys.

ERC-4337 introduces a new alternative transaction pool:

  • UserOperation: an “intent” object signed by the user (not a transaction) · includes calldata / gas limits / paymaster information
  • Bundler: similar to a block builder · takes items from the UserOp mempool, packages them into standard transaction 1 件, and sends them to EntryPoint
  • EntryPoint: an on-chain singleton contract · verifies + executes all UserOps
  • Paymaster: optional contract · pays gas on the user’s behalf (or accepts gas payment in ERC-20 )
  • Aggregator: optional · batch-verifies multiple signatures (BLS / other methods)

Core innovation: UserOp is not a transaction · it is a sub-unit packaged into a transaction by a Bundler.

Origin & evolution

ERC-4337 was proposed in 2021 年 by Vitalik Buterin, Yoav Weiss, Kristof Gazso, and others · the original intent was to avoid the high barrier of “changing the Ethereum protocol layer” (earlier EIP-2938 and EIP-3074 both failed to reach mainnet). After mainnet activation in 2023-03 · progress exceeded expectations —— embedded wallet stacks such as Privy / Coinbase CDP / Alchemy integrated rapidly (contrast with Embedded-wallet network effects · the moat is in the integrator, not the wallet itself).

In 2026 年, the v0.7 upgrade simplified mempool rules · reduced gas overhead by 20-30% · and optimized the Bundler economic model. Around the same period, ERC-7702 (drafted by Vitalik from 2024-05 ) solved the problem of “upgrading existing EOA to SCW” from the opposite direction —— together they form complementary dual tracks.

Sources