Introduction to ERC-8004: The Trust Layer for AI Agents

January 9, 2026 · 8004, Ethereum, Agents

ERC-8004 creates a public discovery and trust layer for autonomous AI agents through three lightweight on-chain registries: Identity, Reputation, and Validation.

Introduction

We're entering the age of AI agents, autonomous software that doesn't just answer questions, but coordinates with other software to deliver real-world outcomes.

Imagine a travel agent that syncs with a budgeting agent and a carbon-impact agent to plan your trip. Or factory agents that negotiate supply, price, and routing across continents. Or health agents that assemble second opinions within minutes.

But there's a problem: how do agents from different organizations trust each other?

When a human hires a contractor, they check references, read reviews, and verify credentials. AI agents need the same capabilities, but at machine speed and without human intermediaries. This is the "trust gap" that ERC-8004 is designed to solve.

What you will learn:

By the end of this article, you will:

This article is the first of a series of guides to help you understand and build using the new ERC-8004 standard.

What problems does ERC-8004 Solve?

Today's agent communication protocols like Google's Agent-to-Agent (A2A) and the Model Context Protocol (MCP) handle how agents talk to each other. But they don't address three critical questions:

Without answers to these questions, the agent economy fragments into walled gardens controlled by large platforms, or becomes a playground for fraud.

What is ERC-8004?

ERC-8004, officially titled "Trustless Agents", is an Ethereum standard that creates a public discovery and trust layer for autonomous AI agents. It was proposed in August 2025 by a cross-industry team including contributors from MetaMask, the Ethereum Foundation, Google, and Coinbase.

At its core, ERC-8004 extends existing agent communication protocols (like A2A and MCP) with three lightweight on-chain registries:

  1. Identity registry: Gives every agent a portable, verifiable identity
  2. Reputation registry: Records feedback from clients who have worked with an agent
  3. Validation registry: Enables third-party verification of an agent's work

ERC-8004 deliberately keeps things minimal. It doesn't try to solve everything, it provides just enough on-chain infrastructure for trust, while leaving application-specific logic off-chain. This design philosophy means:

How ERC-8004 relates to other standards

ERC-8004 doesn't replace existing protocols, it complements them:

ProtocolWhat It DoesERC-8004's Role
A2A (Agent-to-Agent)Agent authentication, messaging, task orchestrationAdds discovery and trust layer
MCP (Model Context Protocol)Servers list capabilities (tools, prompts, resources)Agents can advertise MCP endpoints in their registration
x402HTTP-native micropayments between agentsAgents can pay each other on-chain via HTTP and JSON payloads

Why ERC-8004 matters

As AI agents proliferate, dozens of projects are creating their own discovery and trust systems. Without a shared standard, the ecosystem risks fragmentation, agents from one platform can't interact with agents from another.

This is the coordination problem ERC-8004 addresses. Just as ERC-20 unified the token economy and ERC-721 enabled the NFT market, ERC-8004 aims to create a common foundation for the agent economy.

Why Ethereum?

You might ask: why build this on Ethereum instead of a centralized platform?

The answer comes down to neutrality and permanence:

Ethereum isn't where models run, it's where their trust runs. A discovery and accountability backbone that any agent can plug into.

How does ERC-8004 work?

ERC-8004 introduces three registries that work together. Let's explore each one.

1. Identity Registry

Before ERC-8004, there was no standard way to discover agents across organizations. The Identity Registry creates a global, crawlable directory of agents—enabling the equivalent of "web search" for AI agents.

The Identity Registry gives every agent a unique, portable identity on-chain. It's built on ERC-721, which means:

How it works:

When an agent registers, it receives an agentId (the NFT token ID). This ID links to a registration file (hosted on chain as Base64, IPFS, or any URL) that contains:

Think of it as a machine-readable "passport" for AI agents.

2. Reputation Registry

Reputation is collateral for AI agents. Agents with transparent track records can command higher prices and access more opportunities. The Reputation Registry makes this reputation portable, it follows the agent regardless of which platform or application it uses.

The Reputation Registry creates an on-chain audit trail of client feedback. When you hire an agent to perform a task, you can leave a rating that becomes part of the agent's permanent record.

How it works:

  1. When an agent accepts a task, it signs a feedbackAuth authorizing the client to leave feedback
  2. After the task, the client submits feedback: a score (0-100), optional tags, and a link to detailed off-chain data
  3. The feedback is recorded on-chain and linked to the agent's identity
  4. Anyone can query an agent's reputation history

Key design decisions:

3. Validation registry

Different tasks require different levels of assurance. Ordering pizza? Reputation feedback is probably enough. Managing a DeFi portfolio or diagnosing medical conditions? You want cryptographic or economic guarantees. The Validation Registry provides a unified interface for all these trust models.

For high-stakes tasks, reputation alone isn't enough. The Validation Registry enables independent third-party verification of an agent's work.

How it works:

  1. An agent requests validation by specifying a validator contract and providing a hash of the task data
  2. The validator performs the check (using various methods)
  3. The validator records a response (0-100) on-chain

Validation methods:

ERC-8004 supports multiple validation approaches:

How the 3 registries work together

Here's how the registries combine in practice:

  1. A client searches the Identity registry to discover agents with the right capabilities
  2. The client checks the Reputation registry to filter for agents with strong track records
  3. For the chosen agent, the client may check the Validation registry to see if the agent's work has been independently verified
  4. After completing the task, the client adds feedback to the Reputation registry, building the agent's reputation over time

This creates a virtuous cycle: good agents build reputation, attract more clients, and can afford validation, which further increases trust.

ERC-8004 use cases

So what can you actually build with this? Here's where the ERC-8004 standard is gaining traction and possible use cases:

DeFi and trading

Enterprise coordination

AI Agent marketplaces

Governance & DAOs

What's next

The community is exploring several enhancements:

How to Contribute

Resources

Official resources

Community

Developer tools

Scanners and directories

Conclusion

ERC-8004 represents a foundational step toward an open, permissionless agent economy. By providing standardized registries for identity, reputation, and validation, it enables AI agents to discover each other and build trust, without relying on centralized intermediaries.

Key takeaways:

The agent economy is coming. ERC-8004 ensures it can be built on open, credibly neutral infrastructure rather than proprietary platforms.

Ready to get involved?