Agent ERCs and EOA migration drafts advance

Ethereum Magicians hosts saw concentrated technical work on agent reputation, settlement, and safety proposals, alongside a pre-ERC for EOA code adoption. The heaviest threads refined transferable trust models and verification rules after testnet data.

ERC-8004 transferable reputation refined

Participants on ethereum-magicians.org continued the ERC-8004 Trustless Agents proposal across 134 messages from 24 contributors. Discussion tightened a transferable reputation model that uses ownership_epoch provenance and drew cross-ERC comparisons informed by recent testnet measurements. Developers tracking agent coordination primitives will care because the changes affect how reputation can move with ownership without relying on centralized trust anchors.

Pre-ERC for EOA code adoption via SETCODEFROM

An author posted a pre-ERC describing an adoption profile and reference kernel for EOAs that take on code through SETCODEFROM, referencing EIP-8298 and EIP-8151. The draft lists failure modes and proposes a migration guard. Readers building account abstraction or upgrade paths should note the concrete security boundaries it tries to codify for EOA-to-code transitions.

ERC-8183 agentic commerce verification rules

The ERC-8183 Agentic Commerce thread, with 114 messages from 22 participants, refined recomputability and verification-scope rules that govern reputation in agent settlement flows shared with ERC-8004. The work aims to make reputation checks reproducible within defined scopes. This matters for anyone implementing multi-agent payment or commerce protocols that depend on verifiable trust signals.

Draft Agent Safety Framework registry

A draft ERC for an Agent Safety Framework (ASF) was posted, defining a registry of versioned safety specifications and attributable adoption relations that bind agents to them. Three participants discussed suspension mechanics and intervener fixes across four revisions. The proposal gives a structured place to record which safety specs an agent has adopted, useful for auditability in agent deployments.