Analysis of Agent Name Service Frameworks: OWASP, IETF, GoDaddy Implementations and the Path Forward with Capability Negotiation and Binding Protocols
First of all, congratulations to GoDaddy on the successful public launch of the alpha release of their Agent Name Service (ANS) API and the accompanying ANS Standards site as of November 20, 2025
Godaddy’s work is based on the OWASP Agent Name Service (ANS) proposal as well as IETF ANS draft and share foundational concepts but differ in scope, focus, and implementation details.
OWASP Agent Name Service (ANS)
OWASP’s ANS is a security-first framework focused on secure discovery, registration, and identity management of AI agents, as part of the OWASP GenAI Security Project. It features a centralized registry enabling agent registration, certificate issuance, and discovery with cryptographic assurance (PKI-backed ID, digital signatures, and Zero-Knowledge Proofs for capability validation). The architecture addresses cross-protocol interoperability by bridging among Google’s A2A, Anthropic’s MCP, and others through a modular protocol adapter layer. Security controls include threat modeling with the MAESTRO 7 Layers framework covering impersonation, poisoning, and DoS attack mitigation. OWASP ANS offers code-level implementations with clear API usage and registration/resolution examples.
IETF ANS Proposal (Draft narajala-ans-00)
The IETF draft formalizes the ANS as a universal directory for secure AI agent discovery at the protocol and architecture level. It focuses on a DNS-inspired naming convention for agents with a structure encoding protocol, agent ID, capability, provider, version, and trust domain. The draft defines formal algorithms for agent registration, resolution, certificate-based verification, and lifecycle management, emphasizing secure and scalable directory services. Emphasis is placed on interoperability, multi-protocol support (A2A, MCP, ACP, etc.), and modular extensibility via protocol-specific metadata fields in agent records. It covers detailed security steps such as validation by Registration Authority and Certificate Authority following a trust model for agent identity.
GoDaddy’s ANS Implementation
GoDaddy implements ANS with a strong focus on real-world operational trust, integrating customer identity vetting (KYC), certificate issuance, and attestation mechanisms building a comprehensive trust chain. Their agent naming syntax strictly follows the IETF draft format: Protocol://AgentID.Capability.Provider.vX.Y.Z.Extension. The system provisions multiple DNS record types (TXT, HTTPS, TLSA) to enable comprehensive discovery and endpoint security. GoDaddy’s system includes a public-facing API and internal lifecycle management tools, combining domain DNS infrastructure with agent card hosting and certificate pinning. Special focus is on dynamic DNS provisioning and transparency logs for registration audits, enhancing assurance and public verifiability.
Comparative Summary
Suggestions for Future Work
Semantic Discovery: Enhance semantic search and querying beyond structured DNS names to include richer agent capability and context-aware discovery.
Cross-Protocol Standardization: Further define common schemas to unify agent metadata across heterogeneous protocols for seamless interoperability.
Dynamic Trust Evaluation: Incorporate AI-driven anomaly detection and trust scoring to assess agent behaviors and reputational metrics dynamically.
Ecosystem Expansion: Encourage open-source, multi-vendor ANS registries with federation capabilities to allow agent discovery across competing domains.
Privacy Enhancements: Integrate privacy-preserving techniques like zero-knowledge proofs more extensively in agent registration and capability validation.
Adding Agent Capability Negotiation and Binding Protocol (ACNBP) as the Next Step
Our recent work on the Agent Capability Negotiation and Binding Protocol (ACNBP) builds on ANS infrastructure by providing a comprehensive framework for capability negotiation and binding between AI agents post-discovery. It introduces a structured 10-step protocol including:
Capability discovery via the ANS.
Candidate pre-screening and selection based on capability compatibility, reputation, and security credentials.
Secure, multi-round negotiation phases covering service terms, SLAs, and cryptographically protected message exchanges.
Binding commitment creating legally and cryptographically verifiable agreements.
Execution with real-time monitoring of service quality and security compliance.
Commit/abort decision with cleanup or compensation as needed.
Distributed commitment update for auditability and consistency.
ACNBP integrates robust security measures such as digital signatures, capability attestation, and threat mitigation based on the MAESTRO threat model. It supports backward-compatible protocol evolution via an extensible protocolExtension mechanism to maintain interoperability across heterogeneous agent systems.
OWASP’s security-centric ANS, the IETF’s formal DNS-inspired directory protocol, and GoDaddy’s operational DNS-backed trust system provide a secure foundation for AI agent discovery. The ACNBP protocol extends this foundation by enabling secure, verifiable, and efficient capability negotiation and binding—crucial for collaboration in diverse multi-agent ecosystems.
Together, these form a comprehensive ecosystem: starting with secure agent discovery (OWASP, IETF, GoDaddy ANS), moving to advanced negotiation and binding (ACNBP), and supported by future directions in decentralized trust, semantic discovery, dynamic trust scoring, and privacy.
References: