Researchers Propose Agent Name Service to Standardize AI Agent Discovery

To address the growing need for standardized communication between AI agents, researchers have introduced the Agent Name Service (ANS), a protocol-agnostic registry system inspired by the Domain Name System (DNS).

The Need for ANS

With the rise of autonomous AI agents, there's a pressing need for protocols to govern their interactions. Currently, multiple standards exist:

• Google's Agent2Agent (A2A) for B2B communication
• Anthropic's Model Context Protocol (MCP) for enterprise use
• IBM's Agent Communication Protocol (ACP) for agent delegation

However, none provide a universal way to discover and verify agents. ANS aims to fill this gap by offering a PKI-based identity verification system, ensuring secure and trusted interactions.

How ANS Works

ANS integrates Public Key Infrastructure (PKI) to verify agent identities. Key components include:

• Agent Registry: Stores credentials, capabilities, and policies
• Certificate Authority (CA): Issues X.509 certificates
• Registration Authority (RA): Validates new agents

Agents publish metadata in JSON format, including fields like:

• Protocol
• Agent ID
• Capabilities
• Provider
• Certificate

Challenges and Governance

The paper acknowledges hurdles like:

• Decentralization vs. consistency
• Latency and operational costs
• Name collisions and squatting

Governance models, potentially similar to ICANN for DNS, may be required. The project has garnered interest from MIT's Media Lab and undisclosed large companies.

Reference Implementation

A reference implementation is available on GitHub. The full preprint paper can be found here.

ANS represents a significant step toward standardizing AI agent interactions, but widespread adoption will depend on industry support.