Ethereum agent ERCs refined

Ethereum Magicians threads advanced several draft ERCs on trustless agents, task tokens, and related safety and reputation mechanics. Discussions centered on transferable models, verification rules, and registry designs after testnet work and cross-standard comparisons.

ERC-8004 trustless agents reputation model

Participants on ethereum-magicians.org refined the ERC-8004 proposal for trustless agents, focusing on a transferable reputation model that incorporates ownership_epoch provenance. The thread compared the approach with other ERCs following testnet measurements. Developers tracking agent settlement primitives should note the provenance and cross-standard alignment work.

Token-bound task tenders as reverse assets

A new ERC-9999 draft on ethereum-magicians.org proposes task tokens as reverse assets that act as token-bound task tenders. The design is framed as a demand-side inverse to ERC-8338 skill tokens. The proposal matters for readers building agent marketplaces that need explicit demand-side primitives.

ERC-8183 agentic commerce verification rules

The ERC-8183 thread refined recomputability and verification-scope rules for reputation used in agent settlement, with links to ERC-8004. Twenty-four participants examined how those rules apply in agentic commerce flows. The clarifications affect implementers who must decide what counts as recomputable evidence in cross-agent transactions.

Agent pre-sign risk report draft

A small thread refined draft ERC-8481 for agent pre-sign risk reports, adding 7702 delegate analysis and test vectors. Three participants iterated on the pre-ERC text. The additions give early implementers concrete vectors for assessing delegate risk before signature.

Agent Safety Framework registry draft

An author posted a draft ERC for an Agent Safety Framework that registers versioned safety specifications and scoped, attributable adoption relations binding agents to them. Six participants discussed suspension and intervener fixes across four revisions. The registry approach is relevant to anyone coordinating attributable safety claims across agent deployments.