<?xml version="1.0" encoding="US-ASCII"?>
<!-- edited with XMLSPY v5 rel. 3 U (http://www.xmlspy.com)
     by Daniel M Kohn (private) -->
<!DOCTYPE rfc SYSTEM "rfc2629.dtd" [
<!ENTITY rfc2119 PUBLIC "" "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2119.xml">
]>
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="3"?>
<?rfc tocindent="yes"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<rfc category="std" docName="draft-li-dmsc-mcps-agw-00" ipr="trust200902">
  <front>
    <title abbrev="MCPS-AGW">Multi-agent Collaboration Protocol Suite based on
    Agent Gateway</title>

    <author fullname="Xueting Li" initials="X" surname="Li">
      <organization>China Telecom</organization>

      <address>
        <postal>
          <street>Beiqijia Town, Changping District</street>

          <city>Beijing</city>

          <region>Beijing</region>

          <code>102209</code>

          <country>China</country>
        </postal>

        <email>lixt2@foxmail.com</email>
      </address>
    </author>

    <date day="5" month="February" year="2026"/>

    <area>IETF Area</area>

    <workgroup>DMSC Working Group</workgroup>

    <keyword>Agent Gateway, Artificial Intelligence, Protocol Suite</keyword>

    <abstract>
      <t>This document specifies a Multi-agent Collaboration Protocol Suite
      based on Agent Gateway, which enables scalable, secure, and semantically
      driven collaboration among distributed agents across heterogeneous
      networks. The protocol suite introduces Agent Gateways as control-plane
      entities responsible for agent registration, authentication, capability
      management, semantic routing and other functions, while preserving
      direct peer-to-peer semantic interactions among agents.</t>
    </abstract>
  </front>

  <middle>
    <section anchor="intro" title="Introduction">
      <t>As multi-agent systems become increasingly distributed across
      heterogeneous networks and administrative domains, efficient, secure,
      and semantically meaningful collaboration among agents becomes a
      critical challenge. Traditional service-oriented or message-based
      interaction models are insufficient to capture agent-level capabilities,
      dynamic task decomposition, and semantic intent-driven
      communication.</t>

      <t>This document specifies a Multi-agent Collaboration Protocol Suite
      based on Agent Gateway (AGW). The suite defines a set of coordinated
      protocols that enable agent registration, authentication, capability
      synchronization, semantic routing, task-based invocation, and
      peer-to-peer semantic interaction. The architecture leverages Agent
      Gateways as first-class network entities that mediate control, policy
      enforcement, and orchestration, while allowing agents to directly
      exchange semantic information once authorized.</t>

      <t>The protocol suite is aligned with the architectural principles of
      control/forwarding plane separation, least-privilege authorization, and
      session-scoped semantic communication.</t>
    </section>

    <section title="Conventions used in this document">
      <t>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 <xref target="RFC2119">
      </xref> .</t>
    </section>

    <section title="Terminology">
      <t>The following terms are defined in this draft:<list style="symbols">
          <t>AGW: Agent Gateway. A network-resident control and forwarding
          entity responsible for agent registration, local binding, capability
          management, semantic routing, and policy enforcement.</t>

          <t>Agent: An autonomous software entity capable of perception,
          planning, decision-making, and execution.</t>

          <t>Semantic Routing: The process of routing an Agent request based
          on the meaning or intent of the request, rather than solely on a
          pre-defined address or identifier.</t>

          <t>Central Auth: Central Authentication Service: A logically
          centralized authority that performs identity verification and
          authorization decisions for agents and gateways.</t>
        </list></t>
    </section>

    <section title="Multi-Agent Collaboration Protocol Suite Overview">
      <t>The Multi-Agent Collaboration Protocol Suite based on Agent Gateway
      defines a set of coordinated protocols as shown in figure 1 that
      collectively enable secure agent onboarding, ,distributed capability
      visibility, semantic request resolution, peer-to-peer semantic
      interaction, and task-oriented multi-agent orchestration. Rather than
      operating independently, these protocols are designed to be executed in
      a tightly coupled manner along the agent lifecycle and collaboration
      workflows. The Agent Gateway (AGW) serves as the anchoring point for
      control-plane coordination, while semantic interactions are
      progressively delegated to agents once resolution and authorization are
      completed.</t>

      <t>The protocol suite consists of the following functional
      components::<list style="symbols">
          <t>Agent Registration Protocol (ARP) and Agent Authentication and
          Authorization Protocol (AAAP), which jointly establish agent
          identity, trust, and operational scope.</t>

          <t>Capability Synchronization Protocol (CSP), which maintains
          distributed visibility of agent capability digest across
          gateways.</t>

          <t>Semantic Resolution and Routing Protocol (SRRP), which enables
          semantic request discovery and routing across gateway domains.</t>

          <t>Task-based Invocation Protocol (TIP), which extends semantic
          routing to multi-agent task decomposition and orchestration.</t>
        </list></t>

      <t>Each protocol operates at a specific phase of the collaboration
      lifecycle and may be invoked independently or in combination with
      others. The following sections describe how these protocols are
      integrated into coherent operational flows.</t>

      <figure align="center">
        <artwork><![CDATA[
 User        Agent A      AGW1          AGW3        Central Auth        AGW2        Agent B    Agent C
  |             |            |             |               |              |<--------------Register|
  |             |Register--->|             |               |              |<--Register|           |       
  |             |            |--Auth Req------------------>|<---Auth Req--|           |           |     +-----------------------------+      
  |             |            |<-----------------Auth Grant |--Auth Grant->|           |           |<----| Agent Registration process  |
  |             |<-Local Bind|             |               |              |           |           |     +-----------------------------+ 
  |             |            |             |               |              |--Local Bind (B,C)---->|
  |             |            |             |               |              |           |           |
  |             |            |             |               |              |           |           |     +-----------------------------+   
  |             |            |<----------->|<-----Capa Digest and Sync -->|           |           |<----| Agent Gateway Interaction   |
  |             |            |             |               |              |           |           |     +-----------------------------+ 
  |             |            |             |               |              |           |           |
  |             |            |             |               |              |           |           |
  |             |            |             |               |              |           |           |
  | --Req-----> |-SemR Req-->|             |               |              |           |           |
  |             |            |             |               |              |           |           |
  |             |            |---SemR Req->|----------------------------->|           |           |
  |             |            |<------------|<------SemR Resp--------------|           |           |
  |             |<-SemR Resp |             |               |              |           |           |     +-----------------------------+   
  |             |============================ Semantic Session========================|           |<----| Semantic Routing process    |
  |             |>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>|           |     +-----------------------------+ 
  |             |<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<|           |
  |             |            |             |               |              |           |           |
  |             |            |             |               |              |           |           |
  |             |            |             |               |              |           |           |
  |--Task Req-->|               +-----------------------------+           |           |           |
  |             | ---------->   | Semantic Routing process    |---------->|           |           |
  |             |               + ----------------------------+           |           |           |     +-----------------------------+ 
  |             |--------------------------------------Invoke------------------------------------>|<----|Task-based Invocation process|
  |             |<--------------------------------------Executing---------------------------------|     +-----------------------------+   
  |             |            |             |               |              |           |           |                                                                                               

                           Figure 1 The overall sequence diagram of MCPS-AGW  
]]></artwork>
      </figure>

      <section title="Agent Registration and Authorization Process">
        <t>An agent MUST register with its locally attached Agent Gateway
        before participating in any collaboration. This process is governed
        jointly by ARP and AAAP and establishes the agent&rsquo;s identity,
        trust status, and capability binding. Upon receiving a registration
        request, the Agent Gateway performs preliminary validation of the
        agent&rsquo;s identity attributes and initial capability description,
        and creates a provisional local binding. The gateway then initiates an
        authentication and authorization request to the Central Authentication
        Service, conveying the agent identity, gateway identity, and requested
        operational scope.</t>

        <t>The Central Authentication Service evaluates the request and
        returns an authorization grant or denial. Upon successful
        authorization, the Agent Gateway finalizes the registration by
        assigning the agent a globally unique Agent Identifier and Capability
        Identifier(s). An agent MUST NOT be considered active, discoverable,
        or invocable until this process completes successfully. This combined
        registration and authorization procedure ensures that all subsequent
        semantic routing and task invocation operate on authenticated
        identities and policy-approved capability scopes.</t>
      </section>

      <section title="Capability Digest and Synchronization Process">
        <t>Each Agent Gateway maintains detailed capability information only
        for its locally registered agents. Gateways do not synchronize full
        agent capability states with each other. Instead, to support
        inter-gateway semantic resolution, gateways exchange capability
        digests using the Capability Synchronization Protocol (CSP). A
        capability digest is a locally generated, abstract summary of
        available capabilities, designed solely to indicate what kinds of
        capabilities exist behind a gateway, rather than how those
        capabilities are internally implemented or executed by agents.The
        structure and semantics of capability digests are intentionally
        decoupled from agent-internal capability representations, allowing
        gateways to evolve local capability models without impacting
        inter-gateway interoperability.</t>

        <t>CSP distributes these capability digests incrementally. An initial
        exchange establishes basic inter-gateway visibility, while subsequent
        updates convey only digest changes, such as newly advertised
        capabilities, capability updates, or withdrawals. Digest updates are
        versioned and acknowledged to support consistency and conflict
        resolution. Through this digest-based mechanism, gateways maintain a
        scalable and privacy-preserving view of distributed agent capabilities
        without requiring centralized directories or full capability
        replication.</t>
      </section>

      <section title="Semantic Resolution and Routing Process">
        <t>When a user issues a request to an agent (e.g., Agent A), the agent
        abstracts the request into a semantic request and submits it to its
        locally attached Agent Gateway (AGW1). This interaction is governed by
        the Semantic Resolution and Routing Protocol (SRRP). Upon receiving
        the semantic request, AGW1 performs semantic parsing and normalization
        and consults its local capability directory. If no matching capability
        identifier is found, AGW1 forwards the semantic request to a peer or
        upstream gateway (e.g., AGW3), which repeats the same resolution
        procedure. If the request remains unresolved, it is further forwarded
        to another gateway (e.g., AGW2).</t>

        <t>When a gateway identifies a matching capability in its local
        directory, it generates a semantic resolution response containing the
        resolved capability identifier and the corresponding target agent
        information. This response is propagated hop-by-hop back to the
        originating gateway and ultimately delivered to Agent A.</t>

        <t>Following successful resolution, Agent A and the target agent
        (e.g., Agent B) directly establish a semantic session. During the
        lifetime of this session, semantic data is exchanged directly between
        agents in a peer-to-peer manner, while gateways remain responsible for
        resolution, authorization scope enforcement, and security policy
        application during session establishment.</t>
      </section>

      <section title="Task-based Multi-Agent Invocation Process ">
        <t>Task-based collaboration extends semantic resolution to scenarios
        requiring multiple agents and coordinated execution, as defined by the
        Task-based Invocation Protocol (TIP). When a user initiates a task
        request, the request is delivered to Agent A, which performs semantic
        understanding of the task and decomposes it into one or more sub-tasks
        along with the required capabilities. If Agent A does not possess task
        decomposition capabilities, its attached Agent Gateway MAY act as a
        proxy to analyze and decompose the task on behalf of the agent.</t>

        <t>For each sub-task, Agent A submits a semantic request to its local
        gateway, triggering the same multi-hop semantic resolution process
        defined by SRRP. Unlike pure point-to-point semantic communication,
        gateways additionally apply task-level constraints, policy
        considerations, and capability selection logic to identify suitable
        target agents.</t>

        <t>The resolved results are returned to Agent A, which then directly
        invokes the selected agents and establishes the necessary semantic
        sessions for execution. Through this mechanism, multiple agents can be
        dynamically selected and coordinated to collaboratively execute
        complex tasks, while maintaining consistent authorization and security
        enforcement through gateway-mediated control-plane functions.</t>
      </section>
    </section>

    <section title="Conclusion">
      <t>By explicitly separating control-plane functions from semantic
      interaction flows, and leveraging gateways as control-plane coordination
      points, the proposed protocol suite enables scalable and secure
      multi-agent collaboration without compromising agent autonomy.</t>
    </section>

    <section anchor="iana" title="IANA Considerations">
      <t>TBD</t>
    </section>

    <section title="Acknowledgement">
      <t>TBD</t>
    </section>
  </middle>

  <back>
    <references title="Normative References">
      <?rfc include="reference.RFC.2119"?>
    </references>
  </back>
</rfc>
