EXPERIMENTALIETF (Internet-Draft)

Agent Name Service (ANS)

1 min read

DNS-based discovery system for AI agents with PKI trust.

SpecificationDocs

Overview

As AI agents proliferate, how do you find the right agent for a task? ANS proposes using DNS—the same system that resolves domain names—for agent discovery.

Key Innovation

ANS combines two proven technologies:

  • DNS: Scalable, distributed name resolution
  • PKI (X.509 Certificates): Cryptographic identity verification

How It Works

  1. Agent Registration: Agents register DNS records with their capabilities
  2. Discovery Query: Clients query DNS for agents matching criteria
  3. Identity Verification: PKI certificates prove agent authenticity
  4. Connection: Client connects to verified agent endpoint

Protocol-Agnostic

ANS focuses only on discovery. Once agents are found, they can communicate via any protocol (A2A, ANP, MCP, etc.).

Relationship to Other Standards

  • Complements A2A: Can discover A2A-compatible agents
  • Alternative to Agent Cards: Decentralized instead of well-known paths
  • Works with BANDAID: Part of broader IETF agent discovery work
Quick Info
Architecturedistributed
Message FormatDNS Records
TransportDNS
Agent-to-Agent Features
Discovery
Negotiation
Streaming
Bidirectional
Multi-Agent
Ready to Evaluate?
Agent Playground
Test against real-world scenarios
RepKit
Reputation SDK for agents
Consulting
We'll implement it for you
Stay Updated

Get notified when protocols evolve.

Subscribe
Related Resources
Evaluation Patterns Failure Modes
Strengths
  • Leverages existing DNS infrastructure
  • Cryptographic identity via PKI
  • Scalable and decentralized
  • Protocol-agnostic
Considerations
  • Early draft stage
  • Requires PKI infrastructure
  • DNS propagation delays
Tags
discoverydnsietfpki
Back to Protocols

Compare with Other Protocols

Compare against:(2/2)
Agent Name Service (ANS)(current)
Strengths
Agent Name Service (ANS)
experimental
  • Leverages existing DNS infrastructure
  • Cryptographic identity via PKI
  • Scalable and decentralized
AG-UI Protocol
beta
  • Purpose-built for frontend integration
  • Human-in-the-loop patterns built-in
  • Framework-agnostic design
Agent Network Protocol (ANP)
experimental
  • Fully decentralized—no central authority
  • Strong privacy through DID-based identity
  • Semantic interoperability
Considerations
Agent Name Service (ANS)
  • • Early draft stage
  • • Requires PKI infrastructure
  • • DNS propagation delays
AG-UI Protocol
  • • Frontend-specific scope
  • • No agent discovery
  • • Smaller community than MCP/A2A
Agent Network Protocol (ANP)
  • • Early stage (experimental)
  • • Smaller ecosystem
  • • Complex implementation
Best For
Agent Name Service (ANS)
Finding agents by capabilityVerifying agent identityCross-platform discovery
AG-UI Protocol
AI-powered chat interfacesCollaborative document editingApproval workflow UIsReal-time agent dashboards
Agent Network Protocol (ANP)
Decentralized agent marketplacesCross-organizational collaborationPrivacy-preserving networksFederated research systems
Technical Details
Agent Name Service (ANS)AG-UI ProtocolAgent Network Protocol (ANP)
Architecturedistributedevent-drivendecentralized
Message FormatDNS RecordsJSON EventsJSON-LD
TransportDNSSSE, WebSocket, HTTP WebhooksHTTP/HTTPS, WebSocket, DIDComm
Versiondraft-001.00.1
SDKs—TypeScript, Python, ReactPython, TypeScript
Architecture
Agent Name Service (ANS)distributed
AG-UI Protocolevent-driven
Agent Network Protocol (ANP)decentralized
Message Format
Agent Name Service (ANS)DNS Records
AG-UI ProtocolJSON Events
Agent Network Protocol (ANP)JSON-LD
Transport
Agent Name Service (ANS)DNS
AG-UI ProtocolSSE, WebSocket, HTTP Webhooks
Agent Network Protocol (ANP)HTTP/HTTPS, WebSocket, DIDComm
Version
Agent Name Service (ANS)draft-00
AG-UI Protocol1.0
Agent Network Protocol (ANP)0.1
SDKs
Agent Name Service (ANS)—
AG-UI ProtocolTypeScript, Python, React
Agent Network Protocol (ANP)Python, TypeScript

All Protocols & Initiatives

9 standards documented
AG-UI Protocol

The standard for streaming AI agent interactions to user interfaces.

Supporters: CopilotKit, Microsoft, Google...
Agent Name Service (ANS)(current)

DNS-based discovery system for AI agents with PKI trust.

Supporters: IETF Contributors
Agent Network Protocol (ANP)

Decentralized protocol for trustless agent discovery and collaboration.

Supporters: ANP Community, Academic Research
Agent Protocol

A standardized REST API specification for exposing AI agents as services.

Supporters: AI Engineer Foundation, AutoGPT, W3C Community Group
Agent-to-Agent Protocol (A2A)

The standard for AI agents to communicate, collaborate, and delegate tasks.

Supporters: Google, Microsoft, IBM...
Model Context Protocol (MCP)

The universal standard for connecting AI agents to external tools and data.

Supporters: Anthropic, OpenAI, Google...
NANDA

Framework for decentralized AI agent coordination and lifecycle management.

Supporters: MIT, Academic Research
Open Agent Specification (Agent Spec)

Declarative language for portable agent definitions with built-in evaluation.

Supporters: Oracle, CopilotKit
Open Agentic Schema Framework (OASF)

Standardized schemas for defining AI agent capabilities and metadata.

Supporters: AGNTCY, Cisco (Outshift)