This standard outlines an interface for oracle contracts leveraging multi-agent councils to enable decentralized resolution of information queries. It supports trust-minimized data aggregation and validation, with agents handling off-chain processing and on-chain coordination for enhanced scalability.
It defines a core flow with lightweight methods for request creation, commitments, reveals, judging, and rewards, while providing optional hooks for disputes and reputation systems.
We hope that the discussion around this proposal brings new insight into strengths and weaknesses of the core flow, as well as opening it up to questions from the community.
We also hope to work closely with individuals in the data oracle and prediction market spaces to bring further improvements.
We acknowledge Davide Crapis (Ethereum Foundation) for his time and feedback.
Agent councils is a very important direction that we should make progress on. Important for validation tasks in relation with ERC-8004. Looking forward for this to go live and start the community discussion.
hi guys, any updates about this? I built a project for the chainlink hackathon, kind of resolving the same problem using Chainlink CRE, suddenly this ERC came up in my google session, it would be nice to know where this is going. Any project working on this?
Hi @gilbertsahumada , it’s still early but there is active work happening.
ERC-8033 is currently a draft ERC for multi-agent oracle resolution, with a commit/reveal/judge flow. Public discussion has been fairly limited so far.
We’ve been building an implementation/prototype here:
The demo covers the entire request lifecycle, agent commits/reveals, judge selection, and reward distribution, so this is at least one concrete project working in the ERC-8033 direction.
Would be great to compare notes with your Chainlink CRE hackathon build.
Hey @jonmross, thanks for sharing the contract and demo, will take a closer look at the implementation this week.
Quick context: I built CREsolver for the Chainlink hackathon, a prediction market resolution system using multi-agent BFT consensus, cross-examination, and ERC-8004 reputation. While my use case is narrower (prediction markets specifically), the underlying pattern converged on
something very similar to what ERC-8033 proposes for decentralized oracles more broadly.
Still digesting the full spec and your implementation, but I have some initial observations from building this that could be relevant for the standard. Will follow up with more detailed thoughts once I’ve gone through the contract properly.
What’s the best way to contribute beyond this thread? Should I build against the spec and share findings, open issues on the ERC PR directly, or is
there another channel where the working discussion is happening?
