Agent Collaboration Protocols Architecture for Internet of Agents

Status of This Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶

This Internet-Draft will expire on 2 August 2026.¶

Copyright Notice

Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved.¶

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶

Table of Contents

1. Introduction

With the rapid development of artificial intelligence (AI), particularly large language model (LLM) technology, the number of AI agents has grown dramatically. With the capability of autonomous perception, decision-making, and execution, agents' applications are becoming increasingly widespread.¶

To overcome the limitations of single-agent systems, and to break free from the constraints of proprietary multi-agent frameworks developed by various vendors, the Internet of Agents (IoA) has emerged. IoA aims to enable seamless connectivity and efficient collaboration among agents, through standardized communication protocols and interfaces.¶

This draft proposes the functional architecture of Agent Collaboration Protocols (ACPs) [ACPs-Github], which is designed for IoA to enable wide-area connectivity, cross-domain interoperability, and secure collaboration among heterogeneous agents. The main characteristics of ACPs are as follows:¶

• Multi-centralized architecture, which consists of multiple autonomous domains, each domain containing its own management nodes (such as registration, authentication, and discovery functions), to support efficient, reliable, and manageable large-scale agent interconnection scenarios.¶

• Standardized communication mechanisms with peer-to-peer and grouping mode, allowing agents to self-organize and negotiate autonomously, to facilitate rapid and accurate information exchange, as well as efficient task Collaboration.¶

• Robust registration and authentication mechanisms, to ensure the trusted access of agents and prevent unauthorized access and data breaches.¶

• Reliable registration and management of agent capability, along with intra-domain and cross-domain discovery based on capability matching, to support universal and efficient discovery of collaborative agents.¶

• Real-time monitoring of agent status and behavior, to support more complex application requirements such as agent auditing and transactions.¶

ACPs cover all stages of agents in the IoA, from their access to collaboration, to construct the key infrastructure for agent communication, task collaboration and resource allocation.¶

2. Conventions

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].¶

3. Terminology

Agent: An agent is a software or hardware entity with autonomous decision-making and execution capabilities, capable of perceiving the environment, acquiring contextual information, reasoning, and learning. Agent can perform tasks independently or collaboratively with other agents. Each agent is created by a specific agent provider and needs to complete processes such as registration and authentication before providing services to obtain a legitimate identity and relevant credentials.¶

Agent Identity Code (AIC): AIC is a certifiable, globally unique identity that represents the identity of an agent. AIC MAY contain the following information: the registration service center, the agent provider, the serial number of agent entity and instance, and the check code.¶

Agent Capability Specification (ACS): ACS is a detailed description of an agent's capabilities and information that can be saved and retrieved. ACS MAY use the JSON [RFC8259] format, typically including the following information: AIC, the functional capabilities of agent, the technical characteristics, and the service interfaces.¶

Agent Credential: Agent Credential is a tamper-resistant data object issued by a credential authority (e.g., certificate, token) , used by an agent to prove identity attributes and/or authorization to a relying party.¶

Tool: A tool is a device, software component, or service that provides a specific function and can be accessed and used by an agent.¶

Agent Autonomous Domain: Agent Autonomous Domain is an administrative and governance domain organized and managed by a specific IoA service provider. It MAY include service functions such as identity management, credential management, capability description management, agent discovery, message distribution, and agent monitoring.¶

4. ACPs Architecture from Functional Perspective

This section defines a functional architecture for agent autonomous domain in IoA. The interconnection environment is divided into different conceptual domains. Each conceptual domain includes several functional components, and there are different interfaces between functional components.¶

+--------------------------+        +--------------------------+
| Identity Management      |        | Interconnection Service  |
| Domain                   |        | Domain                   |
|                          |        |                          |
|   +-------+  +-------+   |--------|   +-------+  +-------+   |
|   | AIMgF |  | CMgF  |   |        |   | ADMgF |  |  ADF  |   |
|   +-------+  +-------+   |        |   +-------+  +-------+   |
|   +-------+              |        |   +-------+  +-------+   |
|   |  IVF  |              |        |   |  MDF  |  | DTMgF |   |
|   +-------+              |        |   +-------+  +-------+   |
+-------------+------------+        +-------------+------------+
              |                                   |
              |                                   |
              |                                   |
+-------------+-----------------------------------+------------+
|                         Agent Domain                         |
|                                                              |
|     +------+ +------+ +------+ +------+ +------+ +------+    |
|     | AIMF | | ADMF | | IAF  | | AIF  | | TAF  | | AMF  |    |
|     +------+ +------+ +------+ +------+ +------+ +------+    |
+-------------+-----------------------------------+------------+
              |                                   |
              |                                   |
              |                                   |
+-------------+------------+        +-------------+------------+
| Resource Access Domain   |        |        User Domain       |
|                          |        |                          |
|         +------+         |        |                          |
|         | TSF  |         |        |                          |
|         +------+         |        |                          |
+--------------------------+        +--------------------------+

User Domain
  - User

Agent Domain
  - Agent
  - Agent Identity Maintenance Function (AIMF)
  - Agent Description Maintenance Function (ADMF)
  - Interconnection Authentication Function (IAF)
  - Agent Interaction Function (AIF)
  - Tool Access Function (TAF)
  - Agent Monitoring Function (AMF)

Identity Management Domain
  - Agent Identity Management Function (AIMgF)
  - Credential Management Function (CMgF)
  - Identity Verification Function (IVF)

Interconnection Service Domain
  - Agent Description Management Function (ADMgF)
  - Agent Discovery Function (ADF)
  - Message Distribution Function (MDF)
  - Domain Trust Management Function (DTMgF)

Resource Access Domain
  - Tool Service Function (TSF)

5. Typical Operational Scenarios

This section illustrates how the functional architecture with conceptual domains, functional components, and interfaces is applied in typical operational scenarios.¶

6. Conclusions

This draft introduces the Agent Collaboration Protocols (ACPs) architecture for Internet of Agents. By defining a functional architecture with conceptual domains, functional components and reference interfaces, ACPs can support trusted registration, identity authentication, agent discovery, agent interaction, tool invocation, and agent monitoring. ACPs address the core requirements for interoperability and secure collaboration among heterogeneous agents. The long-term goal of the ACPs is to enable IoA to serve as the critical infrastructure for agent collaboration on complex tasks, as well as to support the future large-scale and diverse applications.¶

7. Security Considerations

This document focuses on the Agent Collaboration Protocols architecture for IoA. Security of IoA is not detailed in this document. Security considerations relevant to deployment with multiple agent service providers are suggested to be deeply discussed through other proposals.¶

8. IANA Considerations

This document makes no request for IANA action.¶

9. References

Authors' Addresses