<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet type='text/xsl' href='rfc2629.xslt'?><?rfc strict="yes"?><?rfc toc="yes"?><?rfc tocdepth="4"?><?rfc symrefs="yes"?><?rfc sortrefs="yes"?><?rfc compact="yes"?><?rfc subcompact="no"?><rfc xmlns:xi="http://www.w3.org/2001/XInclude" docName="draft-zhangb-dawn-entity-mdi-00" category="std" ipr="trust200902" submissionType="IETF">
  <front>
    <title abbrev="DAWN Entity MDI">Generalizing the DAWN MDI for All Entity Types</title>

    <author fullname="Bin Zhang" initials="B." surname="Zhang" role="editor">
      <organization>Pengcheng Laboratory</organization>
      <address>
        <postal>
          <street>Sibilong Street</street>
          <city>Shenzhen</city>
          <country>China</country>
        </postal>
        <email>zhangb@pcl.ac.cn</email>
      </address>
    </author>

    <date year="2026" month="November" day="14"/>

    <area>art</area>
    <workgroup>DAWN Working Group</workgroup>

    <keyword>DAWN</keyword>
    <keyword>discovery</keyword>
    <keyword>MDI</keyword>
    <keyword>entity</keyword>
    <keyword>DNS-SD</keyword>

    <abstract>
      <t>
        The DAWN Minimum Discoverable Information (MDI) information model
        <xref target="I-D.cui-dawn-mdi-model"/> defines a small,
        encoding-neutral common header -- three mandatory elements, a short
        set of recommended and optional elements, and the minimal
        constraints among them. In parallel, local-discovery work
        <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>
        <xref target="I-D.blwm-dawn-common-discovery-parameters"/> has
        converged on a single-DNS-SD-service-type model in which one
        generic service type carries instances of every agent protocol,
        distinguished by a protocol key, with one instance per protocol
        and a descriptor reference pointing to richer metadata.
      </t>
      <t>
        This document generalizes the MDI information model so that the
        same minimal header describes any DAWN entity. In the context of
        DAWN, entities include, but are not limited to, AI agents and AI
        resources <xref target="I-D.farrel-dawn-terminology"/>. An AI
        resource may be an individual named agent or a service such as a
        catalog that will further the discovery process, and the model is
        designed to extend cleanly to tools, tasks, data sources, models,
        workloads, brokers, and application endpoints. The generalization
        reuses the existing agent MDI without change: the element set, the
        constraints, and the DNS-SD binding are inherited, and extension
        is achieved by opening the value space of one element (entity
        type) and by per-type profiles, not by new core elements. Clients
        discover agents, MCP servers, LLM API services, and future entity
        types with a single browse; operators manage one service type.
      </t>
    </abstract>
  </front>

  <middle>
    <section anchor="introduction" title="Introduction">
      <t>
        The DAWN working group has produced two converging lines of work.
        First, the MDI information model <xref target="I-D.cui-dawn-mdi-model"/>
        defines the minimum information an entity must provide to be
        discoverable: exactly three mandatory elements (Entity Identifier,
        Entity Type, Endpoint), a short set of recommended and optional
        elements, and minimal constraints, in an encoding-neutral form
        derivable into many concrete data models. Second, local and global
        discovery work
        <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>
        <xref target="I-D.blwm-dawn-common-discovery-parameters"/>
        converged on the same structure: one generic DNS-SD service type
        (<tt>_agents._tcp</tt>) carries instances of every agent protocol,
        the protocol is named in a <tt>proto</tt> key rather than in the
        service type, one service instance is advertised per protocol, and
        a descriptor reference (<tt>desc-ref</tt>, or <tt>cardUrl</tt> for
        A2A) points to the full protocol-native description.
      </t>
      <t>
        The DAWN terminology <xref target="I-D.farrel-dawn-terminology"/>
        clarifies that, in the context of DAWN, entities include, but are
        not limited to, AI agents and AI resources. The charter further
        clarifies that an AI resource "may be an individual named agent or
        a service such as a catalog that will further the discovery
        process." This document uses "entity" as the general term for any
        discoverable object in DAWN.
      </t>
      <t>
        In parallel, several drafts define discovery for specific entity
        categories:
      </t>
      <ul>
        <li><t>
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> defines
          mDNS/DNS-SD discovery for A2A agents and MCP servers under
          <tt>_agents._tcp</tt>.
        </t></li>
        <li><t>
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/> proposes a
          separate <tt>_llm-api._tcp</tt> service type for local LLM API
          endpoints, re-implementing the same mDNS/DNS-SD discovery pattern
          already defined for agents but with a different service type and
          a different set of TXT keys.
        </t></li>
        <li><t>
          Future categories (tools, tasks, data sources, models, workloads,
          brokers, application endpoints) risk each requiring yet another
          service type and metadata schema.
        </t></li>
      </ul>
      <t>
        Reviewers have observed that the structures arrived at by the agent
        discovery work and the common discovery parameters work &#8212; a
        protocol key, a version key, an endpoint path, and one service
        instance per protocol &#8212; are equally applicable to an LLM API
        profile such as Open Responses, which could be "one more protocol
        in that model, under the same service type rather than a separate
        one." This document adopts that observation as its core design
        principle and takes the next step implied by that convergence: it
        generalizes the MDI information model so that the same minimal
        header describes any DAWN entity.
      </t>

      <section anchor="motivation" title="Motivation: Stopping Per-Entity-Type Discovery Proliferation">
        <t>
          A clear pattern has emerged in the DAWN progress:
          every new category of discoverable entity is being given its own
          discovery protocol definition, even when the underlying discovery
          mechanism is identical.
        </t>
        <t>
          The most concrete example is
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/>. That draft
          defines discovery for local LLM API services using a brand-new
          DNS-SD service type, <tt>_llm-api._tcp</tt>, with a set of TXT
          keys (<tt>txtvers</tt>, <tt>api</tt>, <tt>protovers</tt>,
          <tt>scheme</tt>, <tt>base</tt>, <tt>auth</tt>) that are
          functionally equivalent to the keys already standardized for
          agent discovery in
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/>
          (<tt>txtvers</tt>, <tt>proto</tt>, <tt>proto-ver</tt>,
          <tt>path</tt>, <tt>auth</tt>). The discovery transport (mDNS with
          PTR/SRV/TXT/A records), the two-stage discovery flow (browse then
          resolve), and the per-instance rules are identical to those in
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>. The only
          substantive difference is the entity category: an LLM API service
          instead of an agent.
        </t>
        <t>
          If this pattern is not corrected, the DAWN working group will face
          the following consequences:
        </t>
        <ul>
          <li><t>
            <strong>Duplicated standardization work.</strong> Each new
            entity category will require a dedicated discovery draft, a
            dedicated IANA service-name registration, a dedicated set of
            TXT keys, and a dedicated manifest format &#8212; even though
            the discovery mechanism is the same.
          </t></li>
          <li><t>
            <strong>Inconsistent discovery behavior.</strong> Clients that
            need to find both agents and LLM APIs on the same local network
            must browse two different service types, parse two different TXT
            key sets, and implement two different manifest formats, even
            though the underlying information is structurally identical.
          </t></li>
          <li><t>
            <strong>Increased client implementation complexity.</strong>
            Every new entity type multiplies the client code paths for
            browsing, parsing, and dispatching &#8212; code that differs
            only in field names, not in logic.
          </t></li>
          <li><t>
            <strong>Poor extensibility.</strong> Adding a new entity type
            (e.g., a tool, a data source, a model, a workload) would require
            a new service-type registration and a new discovery draft,
            creating a linear growth of standards work for a constant
            discovery mechanism.
          </t></li>
          <li><t>
            <strong>Operator burden.</strong> Network operators would need
            to configure, firewall, and monitor a growing number of service
            types, one per entity category, even though all of them carry
            the same kind of information.
          </t></li>
        </ul>
        <t>
          The root cause is that the entity <em>category</em> (what the
          entity <em>is</em>) and the discovery <em>transport</em> (how the
          entity is <em>found</em>) have been coupled. Once they are
          separated, a single discovery transport serves all entity
          categories, and new categories are added by registering an entity
          type value rather than by defining a new discovery protocol.
        </t>
        <t>
          This document makes that separation. It defines one generalized
          MDI information model and one extensible entity type registry,
          layered on top of the existing <tt>_agents._tcp</tt> service type
          and the common discovery parameters. The LLM API discovery
          defined in
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/> is absorbed
          as the <tt>type=service</tt> case with
          <tt>proto=openresponses</tt>, eliminating the need for
          <tt>_llm-api._tcp</tt> entirely. Future entity categories will be
          added the same way: a new type value, the same service type, the
          same DNS keys, the same model.
        </t>
      </section>

      <section anchor="design-principles" title="Design Principles">
        <t>
          The principles of <xref target="I-D.cui-dawn-mdi-model"/>,
          Section 4 &#8212; encoding-neutral model, thin core with rich
          references, discovery-time only, freshness as first-class
          &#8212; are inherited without change. This document adds three
          principles that govern the generalization:
        </t>
        <dl>
          <dt>Reuse before extension:</dt>
          <dd>
            No new core element is introduced for any current or future
            entity type. A type is expressed through the value space of the
            Entity Type element, through profile-specific values of existing
            elements (e.g., capability-summary tags), and through the
            extension mechanism. An implementation that ignores Entity Type
            entirely still obtains a valid MDI instance for an agent: the
            generalized model degrades to the agent MDI.
          </dd>
          <dt>Orthogonal axes:</dt>
          <dd>
            Entity type says WHAT the resource is; the protocol keys say
            HOW to speak to it. The two are independent. A model resource
            may expose an Open Responses endpoint; an agent may expose both
            an A2A and an MCP endpoint. Neither axis is encoded in the
            DNS-SD service type.
          </dd>
          <dt>Uniform multiplicity:</dt>
          <dd>
            The settled rule &#8212; one service instance per protocol, with
            instances of the same logical entity carrying the same
            descriptor reference so clients can tell they belong together
            <xref target="I-D.blwm-dawn-common-discovery-parameters"/>
            &#8212; extends uniformly: one service instance per (entity,
            protocol) pair. The unit of multiplicity is the instance, not
            the record or the key.
          </dd>
        </dl>
      </section>

      <section anchor="requirements-language" title="Requirements Language">
        <t>
          The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
          "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY",
          and "OPTIONAL" in this document are to be interpreted as
          described in BCP 14 <xref target="RFC2119"/>
          <xref target="RFC8174"/> when, and only when, they appear in all
          capitals, as shown here.
        </t>
      </section>
    </section>

    <section anchor="terminology" title="Terminology">
      <t>
        This document uses the terms defined in
        <xref target="I-D.farrel-dawn-terminology"/> and
        <xref target="I-D.cui-dawn-mdi-model"/>, including Entity, AI
        resource, Discovery Information, Discoverable Object, Capability
        Card, Trust Indicator, Minimum Discoverable Information (MDI),
        publisher scope, and binding. In the context of DAWN, entities
        include, but are not limited to, AI agents and AI resources
        <xref target="I-D.farrel-dawn-terminology"/>. This document uses
        "entity" as the general term and adds:
      </t>
      <dl>
        <dt>Entity type:</dt>
        <dd>
          The coarse kind of an entity, carried in the Entity Type element
          (<xref target="elements"/>). Values are registered in the DAWN
          Entity Type registry (<xref target="iana"/>).
        </dd>
        <dt>Profile:</dt>
        <dd>
          A document that binds the MDI model to one entity type (and,
          where needed, to specific protocol values), specifying the
          applicable value spaces, the native description document that the
          descriptor reference locates, and any type-specific keys. A
          profile reuses the core model and MUST NOT relax its rules.
        </dd>
        <dt>Named agent:</dt>
        <dd>
          An entity of type <tt>agent</tt> with a stable name, e.g., an A2A
          agent advertising an AgentCard.
        </dd>
        <dt>Discovery-furthering service:</dt>
        <dd>
          An entity of type <tt>catalog</tt> whose descriptor reference
          locates a catalog document that lists further entities (or
          references to them), extending the discovery process beyond the
          link-local browse.
        </dd>
      </dl>
    </section>

    <section anchor="scope" title="Scope and Non-Goals">
      <t>
        This document defines the abstract content of the generalized MDI
        common header and the framework for extending it. It does NOT
        define:
      </t>
      <ul>
        <li><t>a discovery, query, resolution, or transport protocol, and it does not bind MDI to DNS or to any other carrier;</t></li>
        <li><t>a wire format, serialization, or schema; representations are illustrative (<xref target="appendix-json"/>);</t></li>
        <li><t>a new capability card, or a replacement for AgentCard, MCP server manifests, model cards, or any existing descriptor;</t></li>
        <li><t>identity verification, authentication, authorization, policy evaluation, or attestation procedures;</t></li>
        <li><t>capability exchange, negotiation, selection, registration, or invocation semantics.</t></li>
      </ul>
      <t>
        The non-goals of <xref target="I-D.cui-dawn-mdi-model"/>, Section 3
        apply unchanged. MDI remains advertised information, positioned
        strictly below verification: by itself it asserts nothing a
        discovering party should treat as verified.
      </t>
    </section>

    <section anchor="model" title="Generalized MDI Information Model">
      <t>
        An MDI instance is the common header published or returned for one
        entity. Only three elements are mandatory &#8212; Entity Identifier,
        Entity Type, and Endpoint &#8212; so that an entity can be
        discoverable with a very small record; in constrained carriers, a
        binding may derive mandatory values from the carrier itself. The
        element set is exactly that of <xref target="I-D.cui-dawn-mdi-model"/>,
        Section 5, read with "entity" used as the general term.
        <xref target="element-table"/> states the correspondence;
        <xref target="structure"/> restates the structure for convenience.
      </t>

      <table anchor="element-table">
        <name>Element Correspondence with the MDI Model</name>
        <thead>
          <tr><th>This document</th><th><xref target="I-D.cui-dawn-mdi-model"/></th><th>Semantics</th></tr>
        </thead>
        <tbody>
          <tr><td>Entity Identifier</td><td>Entity Identifier</td><td>unchanged</td></tr>
          <tr><td>Entity Type</td><td>Entity Type</td><td>value space opened (<xref target="iana"/>)</td></tr>
          <tr><td>Entity Summary</td><td>Entity Summary</td><td>unchanged</td></tr>
          <tr><td>Endpoint</td><td>Endpoint</td><td>unchanged</td></tr>
          <tr><td>Capability Summary</td><td>Capability Summary</td><td>unchanged; tag space per type</td></tr>
          <tr><td>Capability Reference</td><td>Capability Reference</td><td>unchanged</td></tr>
          <tr><td>Trust Reference</td><td>Trust Reference</td><td>unchanged</td></tr>
          <tr><td>Authentication Hint</td><td>Authentication Hint</td><td>unchanged</td></tr>
          <tr><td>Freshness</td><td>Freshness</td><td>unchanged</td></tr>
          <tr><td>Provenance</td><td>Provenance</td><td>unchanged</td></tr>
          <tr><td>Scope Hint</td><td>Scope Hint</td><td>unchanged</td></tr>
          <tr><td>Operational Hint</td><td>Operational Hint</td><td>unchanged</td></tr>
          <tr><td>Extension</td><td>Extension</td><td>unchanged</td></tr>
        </tbody>
      </table>

      <section anchor="structure" title="Structure">
        <t>
          The structure of an MDI instance is unchanged from
          <xref target="I-D.cui-dawn-mdi-model"/>, Figure 1:
        </t>
        <figure>
          <artwork type="ascii-art" name="mdi-structure.txt"><![CDATA[
+--------------+   describes exactly one
| MDI Instance |--------------------------->  entity
+--------------+
  |- entity-identifier                          (1)
  |- entity-type                                (1)
  |- entity-summary                             (0..n)
  |- Endpoint { locator-type, locator,
  |             protocols (1..n) }              (1..n)
  |- capability-summary                         (0..n)
  |- capability-reference : Reference           (0..n)
  |- trust-reference      : Reference           (0..n)
  |- authentication-hint { scheme, metadata? }  (0..n)
  |- freshness { version?, valid-from?,
  |              valid-until?, refresh-hint? }  (0..1)
  |- provenance { publisher,
  |               responsible-party? }          (0..1)
  |- scope-hint                                 (0..n)
  |- operational-hint { status?, detail? }      (0..1)
  |- extension { name, value }                  (0..n)

  Reference = { ref-type, locator, digest? }
]]></artwork>
        </figure>
        <t>
          A Reference is a typed pointer, unchanged from the MDI model: the
          ref-type says how to interpret the locator, and the optional
          digest is an integrity digest over the object the locator
          identifies.
        </t>
      </section>

      <section anchor="elements" title="Elements">
        <t>
          Element definitions, status, and classes (static /
          mainly-static / dynamic) are those of
          <xref target="I-D.cui-dawn-mdi-model"/>, Section 5.2 and are not
          repeated here. The only semantic delta is Entity Type:
        </t>
        <t>
          <strong>Entity Type:</strong> the kind of entity, from the DAWN
          Entity Type registry (<xref target="iana"/>). The initial value
          space is <tt>agent</tt>, <tt>catalog</tt>, <tt>service</tt>,
          <tt>tool</tt>, <tt>task</tt>, <tt>data-source</tt>,
          <tt>model</tt>, <tt>workload</tt>, <tt>broker</tt>, and
          <tt>application-endpoint</tt>; it is designed to be extended.
          Entity Type lets a discovering party triage before fetching
          anything. The value <tt>agent</tt> is the default
          (<xref target="type-agent"/>): an instance that does not state an
          Entity Type denotes a named agent, so every advertisement formed
          under <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> remains
          valid under this model.
        </t>
      </section>

      <section anchor="constraints" title="Minimal Constraints">
        <t>
          Constraints 1-5 of <xref target="I-D.cui-dawn-mdi-model"/>,
          Section 5.3 apply unchanged, with "Entity Identifier" read as
          the same. This document adds one rule; a binding MAY tighten it
          but MUST NOT relax it:
        </t>
        <ol>
          <li><t>
            6. Entity Type does not alter the semantics of any other
            element, and absence of Entity Type means <tt>agent</tt>. A
            consumer MUST ignore an Entity Type value it does not
            recognize, treating the instance as an opaque entity; it MUST
            NOT fail to parse the remaining elements, and it MUST still
            apply the Security Considerations of <xref target="security"/>.
          </t></li>
        </ol>
      </section>
    </section>

    <section anchor="type-registry" title="Entity Type Registry and Initial Values">
      <t>
        Entity Type values are registered in a new registry, "DAWN Entity
        Types" (<xref target="iana"/>). Registration requires a short
        definition and, where applicable, a profile reference; it does not
        require changes to this document. Values MUST be 1-32 characters,
        lowercase letters, digits, and hyphens, beginning with a letter.
      </t>
      <t>
        This section defines the initial values. For each value it states
        the definition, the typical protocol values (the <tt>proto</tt>
        axis, <xref target="dns-binding"/>), and the native description
        document that the Capability Reference / descriptor reference
        typically locates.
      </t>

      <section anchor="type-agent" title="agent">
        <t>
          <strong>Definition:</strong> an individual named agent &#8212; an
          autonomous or semi-autonomous entity that interacts using an agent
          protocol. This is the default entity type and the baseline
          profile of <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>a2a</tt> (Agent2Agent),
          <tt>mcp</tt> (Model Context Protocol), and other registered agent
          protocols.
        </t>
        <t>
          <strong>Typical description document:</strong> the
          protocol-native agent description &#8212; for A2A, the AgentCard
          located by <tt>cardUrl</tt>; for MCP, the server metadata
          document.
        </t>
      </section>

      <section anchor="type-catalog" title="catalog">
        <t>
          <strong>Definition:</strong> a discovery-furthering service: a
          service whose description enumerates or indexes further entities,
          so that discovering it advances the discovery process beyond the
          initial browse. A catalog is a service, specialized by its role
          in the discovery process.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>https</tt> (or the
          catalog's native query protocol).
        </t>
        <t>
          <strong>Typical description document:</strong> a catalog document
          &#8212; a list whose entries are MDI instances or references to
          them. Resolution of a catalog is recursive discovery
          (<xref target="two-stage"/>); implementations SHOULD bound
          recursion and apply the dereferencing protections of
          <xref target="security"/> to every step.
        </t>
      </section>

      <section anchor="type-service" title="service">
        <t>
          <strong>Definition:</strong> a network service that provides
          AI-related capability but is not itself a named agent &#8212;
          e.g., an LLM API service, an embedding endpoint, a retrieval
          service, or an inference gateway. The LLM API service of
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/> is an entity
          of this type.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>openresponses</tt> (Open
          Responses API profile), and other registered API profiles.
        </t>
        <t>
          <strong>Typical description document:</strong> the API profile
          definition and service metadata; the endpoint path is carried
          directly in the instance.
        </t>
      </section>

      <section anchor="type-tool" title="tool">
        <t>
          <strong>Definition:</strong> a callable tool or capability server
          &#8212; an entity exposing a set of invocable tools rather than
          an agent persona. An MCP server advertising tool capabilities is
          the canonical example; it may equally be advertised as an agent
          when its role in a deployment is agent-like.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>mcp</tt>, and registered
          tool protocols.
        </t>
        <t>
          <strong>Typical description document:</strong> the
          tool/capability manifest listing the exposed tools.
        </t>
      </section>

      <section anchor="type-task" title="task">
        <t>
          <strong>Definition:</strong> a discoverable unit of work &#8212;
          a task definition, job template, or task-queue endpoint that a
          discovering party may inspect, claim, or schedule according to
          semantics defined outside MDI.
        </t>
        <t>
          <strong>Typical protocols:</strong> the task protocol registered
          for the deployment.
        </t>
        <t>
          <strong>Typical description document:</strong> the task
          specification or queue descriptor.
        </t>
      </section>

      <section anchor="type-data-source" title="data-source">
        <t>
          <strong>Definition:</strong> a source of data usable by entities
          or discovering parties &#8212; a dataset, stream, knowledge base,
          or data-serving endpoint.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>https</tt>, and
          registered data protocols.
        </t>
        <t>
          <strong>Typical description document:</strong> the
          dataset/data-service description (a data card or catalog entry).
        </t>
      </section>

      <section anchor="type-model" title="model">
        <t>
          <strong>Definition:</strong> a machine-learning model exposed as
          a discoverable entity &#8212; a model endpoint, a model-serving
          front end, or a model card pointer. Where the model is reachable
          through an API, the endpoint and protocol are carried as for any
          entity; the model card is located by the Capability Reference.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>openresponses</tt>, and
          registered serving protocols.
        </t>
        <t>
          <strong>Typical description document:</strong> the model card.
        </t>
      </section>

      <section anchor="type-workload" title="workload">
        <t>
          <strong>Definition:</strong> a schedulable or running workload
          &#8212; a computation, training job, inference batch, or similar
          unit of execution advertised for discovery, monitoring, or
          handoff.
        </t>
        <t>
          <strong>Typical protocols:</strong> the workload-management
          protocol registered for the deployment.
        </t>
        <t>
          <strong>Typical description document:</strong> the workload
          descriptor or status document.
        </t>
      </section>

      <section anchor="type-broker" title="broker">
        <t>
          <strong>Definition:</strong> an intermediary that mediates
          interaction among entities &#8212; e.g., a message broker, an
          agent broker, or an orchestrator that routes tasks or messages.
          Unlike a catalog, a broker primarily furthers interaction after
          discovery.
        </t>
        <t>
          <strong>Typical protocols:</strong> the broker's native protocol.
        </t>
        <t>
          <strong>Typical description document:</strong> the broker
          metadata document.
        </t>
      </section>

      <section anchor="type-app-endpoint" title="application-endpoint">
        <t>
          <strong>Definition:</strong> an application endpoint through which
          users or other applications interact with entities &#8212; e.g.,
          a UI endpoint, a webhook receiver, or an application-facing API
          exposed for integration.
        </t>
        <t>
          <strong>Typical protocols:</strong> <tt>https</tt>, and
          registered application protocols.
        </t>
        <t>
          <strong>Typical description document:</strong> the application
          endpoint description.
        </t>
      </section>
    </section>

    <section anchor="profiles" title="Profiles">
      <t>
        A profile binds the generalized model to an entity type and
        protocol values. Profiles are the extension mechanism: adding an
        entity type or a protocol defines a profile (or amends one); the
        core model does not change. A profile MUST specify:
      </t>
      <ul>
        <li><t>the Entity Type value (or values) it applies to;</t></li>
        <li><t>the protocol values (<tt>proto</tt> axis) used for that type;</t></li>
        <li><t>the native description document that the Capability Reference (descriptor reference) locates, and how it is retrieved;</t></li>
        <li><t>the value space of capability-summary tags for the type, if a shared tag space is needed;</t></li>
        <li><t>any type- or protocol-specific TXT/SVCB keys, named with a registered prefix.</t></li>
      </ul>
      <t>
        A profile MUST NOT relax the MUST elements or the constraints of
        <xref target="constraints"/>.
      </t>

      <section anchor="profile-agent" title="Agent Profile (Baseline, Reused)">
        <t>
          The agent profile is defined by
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> with the
          common parameter set of
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/>, and
          it is reused here unchanged. No element of this profile is
          modified by the generalization. A discovering party that
          implements only the agent profile interoperates with every
          advertisement in this document.
        </t>
      </section>

      <section anchor="profile-llm" title="LLM API Service Profile">
        <t>
          The LLM API service profile aligns
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/> with the
          common model, as anticipated by the observation that an LLM API
          profile is one more protocol under the same service type:
        </t>
        <ul>
          <li><t><strong>Entity Type:</strong> <tt>service</tt>.</t></li>
          <li><t><strong>Protocol value:</strong> <tt>openresponses</tt>, identified by <tt>proto=openresponses</tt> (the <tt>api</tt> key of <xref target="I-D.cui-dawn-local-llm-api-discovery"/> corresponds to the common <tt>proto</tt> key); the protocol version is carried in <tt>proto-ver</tt> (the <tt>protovers</tt> key corresponds to <tt>proto-ver</tt>), e.g., <tt>proto-ver=2026-04-24</tt>.</t></li>
          <li><t><strong>Endpoint:</strong> the SRV/SVCB target and port give the host and port; the base path is carried in the common <tt>path</tt> key (the <tt>base</tt> key corresponds to <tt>path</tt>); the API base URL is constructed from scheme (https unless a binding states otherwise), host, port, and path.</t></li>
          <li><t><strong>Authentication hint:</strong> the common authentication hint, simplified as the <tt>auth</tt> key with values <tt>required</tt>/<tt>none</tt>; <tt>auth=none</tt> does not constitute authorization.</t></li>
        </ul>
        <t>
          One instance describes one API profile at one base path on one
          listener, exactly as in
          <xref target="I-D.cui-dawn-local-llm-api-discovery"/>; a listener
          offering several API profiles publishes one instance per profile.
        </t>
      </section>

      <section anchor="profile-other" title="Other Type Profiles (Template)">
        <t>
          Profiles for <tt>catalog</tt>, <tt>tool</tt>, <tt>task</tt>,
          <tt>data-source</tt>, <tt>model</tt>, <tt>workload</tt>,
          <tt>broker</tt>, and <tt>application-endpoint</tt> follow the
          same template and are expected as separate documents or future
          revisions of this one. Until a profile is defined,
          advertisements of that type carry the core elements only: Entity
          Type, Entity Identifier, Endpoint (locator and protocols), and,
          SHOULD, a Capability Reference to the type's native description
          document. This is sufficient for enumeration, triage, and
          recursive discovery (for catalogs), and it is the sense in which
          the model is elastic: a new entity type is discoverable from the
          day its registry value is assigned, with full profile detail
          added later without breaking early implementations.
        </t>
      </section>
    </section>

    <section anchor="dns-binding" title="Discovery Procedure and DNS-SD Binding">
      <section anchor="single-service-type" title="Single Service Type, Single Browse">
        <t>
          All entity types are advertised under one generic DNS-SD service
          type, following the settled architecture of
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> and
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/>. The
          service type name is <tt>agents</tt> with transport
          <tt>tcp</tt> (<tt>_agents._tcp</tt>) per
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/>; the
          earlier spelling <tt>_agent._tcp</tt> of
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> denotes the
          same model and the two spellings are to be reconciled in IANA
          registration (<xref target="iana-service"/>). A separate service
          type per entity type (e.g., <tt>_llm-api._tcp</tt>,
          <tt>_tool._tcp</tt>) is explicitly NOT used, for the reasons
          documented in <xref target="I-D.blwm-dawn-common-discovery-parameters"/>,
          Appendix B.5: each additional service type requires its own
          registration, reconfiguration of mDNS gateways and proxies,
          splits instance-name uniqueness, and fragments what clients must
          browse.
        </t>
        <t>
          A discovering party browses <tt>_agents._tcp.local.</tt> once and
          learns, per instance: the Entity Type from the <tt>rtype</tt> key
          (absent means <tt>agent</tt>), the protocol from the
          <tt>proto</tt> key, the entity identifier, the endpoint
          (SRV/SVCB), lightweight selection hints, and the descriptor
          reference.
        </t>
      </section>

      <section anchor="protocol-values" title="Protocol Values">
        <t>
          Protocol values (the <tt>proto</tt> key) are registered in the
          "DAWN Entity Protocols" registry (<xref target="iana-protocols"/>).
          Values are lowercase, digits, and hyphens (e.g., <tt>a2a</tt>,
          <tt>mcp</tt>, <tt>openresponses</tt>). A protocol value names the
          application protocol spoken at the advertised endpoint; it is
          registered together with the profile that defines its use. Note
          that an MCP server is discovered by <tt>proto=mcp</tt> with
          <tt>rtype=tool</tt> or <tt>rtype=agent</tt> according to its
          role, and an LLM API service by <tt>proto=openresponses</tt> with
          <tt>rtype=service</tt>: entity type and protocol vary
          independently.
        </t>
      </section>

      <section anchor="two-stage" title="Two-Stage Discovery and Catalog Recursion">
        <t>
          Discovery proceeds in two stages, unchanged from
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>: Stage 1
          browses and resolves DNS-SD records for enumeration and
          lightweight selection; Stage 2 retrieves the protocol-native
          description (AgentCard, MCP manifest, model card, catalog
          document, ...) from the descriptor reference and proceeds
          entirely in the entity's own protocol.
        </t>
        <t>
          For a catalog entity, Stage 2 retrieval yields further entities
          (MDI instances or references), and the discovering party applies
          Stage 2 to each selected entry. Recursion SHOULD be bounded (a
          small default maximum depth, e.g., 3), each hop is subject to
          the dereferencing protections of <xref target="security"/>, and
          cached entries from a catalog are governed by their own
          Freshness and by the carrier lifetime, whichever expires earlier.
        </t>
      </section>

      <section anchor="dns-keys" title="TXT and SVCB Keys">
        <t>
          The common key set of
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/> is
          reused unchanged: <tt>txtvers</tt> (MUST), <tt>proto</tt>
          (SHOULD), <tt>desc-ref</tt>, <tt>desc-ref-digest</tt>,
          <tt>proto-ver</tt>, <tt>path</tt>, <tt>well-known</tt>,
          <tt>title</tt> (SHOULD), and <tt>alpn</tt> (SVCB).
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> adds the
          agent descriptor keys <tt>agentId</tt>, <tt>cardUrl</tt>,
          <tt>caps</tt>, <tt>ver</tt>, and protocol-specific keys. This
          document adds exactly one common key:
        </t>
        <dl>
          <dt><tt>rtype</tt>:</dt>
          <dd>
            the Entity Type value (<xref target="type-registry"/>). SHOULD
            be present; absence means <tt>agent</tt>. A consumer MUST
            ignore an unrecognized <tt>rtype</tt> value (Constraint 6,
            <xref target="constraints"/>). Values are lowercase, digits,
            and hyphens.
          </dd>
        </dl>
        <t>
          Key names follow the common conventions: lowercase, digits, and
          hyphens, at most nine characters where possible, so one
          vocabulary serves both TXT and SVCB encodings. Type- or
          protocol-specific keys use the registered value as a prefix
          followed by a hyphen (e.g., <tt>a2a-</tt> binding pattern), so
          independently registered profiles cannot collide. Exactly one TXT
          record is used per service instance, and one service instance is
          advertised per (entity, protocol) pair; instances of the same
          logical entity carry the same descriptor reference so a client
          can tell they belong together.
        </t>
      </section>

      <section anchor="instance-naming" title="Instance Naming">
        <t>
          The Entity Identifier is used as the initial candidate for the
          "Instance" portion of the DNS-SD service instance name, encoded
          as a single DNS label per <xref target="RFC6763"/>, Section
          4.1.1, with literal dots and backslashes escaped per Section 4.3
          of that document. If the label would exceed 63 octets or cannot
          be represented, the advertising implementation does not silently
          truncate; it advertises under an explicitly configured alternate
          instance name or does not advertise. mDNS conflict resolution of
          <xref target="RFC6762"/> applies; the TXT identifier key (e.g.,
          <tt>agentId</tt>) continues to carry the original Entity
          Identifier after any renaming, exactly as for agents.
        </t>
      </section>

      <section anchor="transition" title="Transition and Coexistence">
        <t>
          Deployments and implementations of the earlier, per-type service
          types (<tt>_agent._tcp</tt> spellings, <tt>_llm-api._tcp</tt>)
          can dual-advertise during transition, as already contemplated
          for other coexisting agent discovery approaches. New entity types
          register only <tt>rtype</tt> values and publish under the single
          generic service type; no new service type is needed for a new
          entity type, ever.
        </t>
      </section>
    </section>

    <section anchor="deployment" title="Deployment Considerations">
      <t>
        The deployment considerations of
        <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>, Section 7
        apply unchanged: link-local scope of mDNS, separation of discovery
        from reachability and trust, and compatibility with multicast
        filtering and local policy. Two additional notes:
      </t>
      <ul>
        <li><t>
          <strong>Enumeration exposure:</strong> because all entity types
          share one service type, a single browse exposes the presence of
          every type on the link. Deployments that consider tool, model, or
          workload presence more sensitive than agent presence SHOULD
          minimize the advertised hints for those types (small TXT
          records, no capability tags) and rely on Stage 2 access control;
          the model supports this because the descriptor reference can
          require authorization that discovery does not perform.
        </t></li>
        <li><t>
          <strong>Subtype filtering:</strong> clients that care about one
          type or protocol browse the corresponding DNS-SD subtype, e.g.,
          <tt>_openresponses._sub._agents._tcp</tt> for LLM API services,
          <tt>_tool._sub._agents._tcp</tt> for tools, without any change to
          the core model or to generic clients.
        </t></li>
      </ul>
    </section>

    <section anchor="extensibility" title="Extensibility Guidelines">
      <t>
        Adding support for a new kind of entity is a three-step procedure;
        none of the steps touches the core model:
      </t>
      <ol>
        <li><t>
          Register the Entity Type value in the "DAWN Entity Types"
          registry (<xref target="iana-types"/>), with a definition and,
          when available, a profile reference.
        </t></li>
        <li><t>
          If the type uses protocols not yet registered, register the
          protocol values (<xref target="iana-protocols"/>) with their
          profiles.
        </t></li>
        <li><t>
          Publish the profile (<xref target="profiles"/>), specifying
          value spaces, the native description document, and type-specific
          keys. Until the profile exists, advertisements carry the core
          elements only (<xref target="profile-other"/>).
        </t></li>
      </ol>
      <t>
        Clients written to this document require no update to enumerate and
        Stage-1-triage the new type (they ignore the unknown
        <tt>rtype</tt> value per Constraint 6 but still parse the
        instance), and require an update only to interpret the new type's
        native description in Stage 2. This is the elasticity property:
        extension is registration plus profile, never core change.
      </t>
    </section>

    <section anchor="security" title="Security Considerations">
      <t>
        The Security Considerations of
        <xref target="I-D.cui-dawn-mdi-model"/> and
        <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> apply
        unchanged and are not restated. In brief: MDI is advertised,
        unauthenticated information; any element can be forged;
        identifiers are names, not identities; digests give integrity of
        referenced bytes, not trust; trust is established in Stage 2 by
        the entity protocol's own mechanisms; reference locators are
        untrusted input and the usual dereferencing protections apply; and
        implementations MUST NOT draw conclusions from the absence of an
        element.
      </t>
      <t>
        Generalization adds no new attack surface beyond enumeration
        exposure (<xref target="deployment"/>): an advertiser forges an
        <tt>rtype</tt> value as easily as any other TXT key, and a
        discovering party MUST NOT treat a recognized entity type as more
        trustworthy than an unrecognized one. Catalog recursion
        (<xref target="two-stage"/>) multiplies untrusted dereferences and
        MUST be bounded and policy-filtered like any other reference
        following.
      </t>
    </section>

    <section anchor="privacy" title="Privacy Considerations">
      <t>
        The Privacy Considerations of
        <xref target="I-D.cui-dawn-mdi-model"/> apply unchanged: publishing
        MDI can reveal entity names, endpoints, organizational structure,
        capability hints, and scope information. Publishers SHOULD minimize
        public MDI to what discovery requires and place sensitive detail
        behind access-controlled references. Because one browse now
        enumerates every entity type, minimization applies across types,
        not only to agents.
      </t>
    </section>

    <section anchor="iana" title="IANA Considerations">
      <section anchor="iana-types" title="DAWN Entity Types Registry">
        <t>
          IANA is requested to create the "DAWN Entity Types" registry,
          with registration procedure "Specification Required"
          <xref target="RFC8126"/> and values of the lexical form of
          <xref target="type-registry"/>. The initial contents are the
          values defined in <xref target="type-registry"/>:
          <tt>agent</tt>, <tt>catalog</tt>, <tt>service</tt>,
          <tt>tool</tt>, <tt>task</tt>, <tt>data-source</tt>,
          <tt>model</tt>, <tt>workload</tt>, <tt>broker</tt>,
          <tt>application-endpoint</tt>.
        </t>
      </section>

      <section anchor="iana-service" title="Service Name Registration">
        <t>
          This document uses the generic service type
          <tt>_agents._tcp</tt> of
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/>.
          Should this usage be progressed, registration of the service
          name <tt>agents</tt> for TCP in the Service Name and Transport
          Protocol Port Number Registry <xref target="RFC6335"/> is
          requested, following the underscored-service-name conventions of
          <xref target="RFC8552"/>, with no fixed port number (the port is
          conveyed per instance in SRV/SVCB). The relationship between the
          spellings <tt>_agent._tcp</tt>
          <xref target="I-D.jakab-dawn-agent-discovery-mdns"/> and
          <tt>_agents._tcp</tt>
          <xref target="I-D.blwm-dawn-common-discovery-parameters"/> is to
          be reconciled at registration time; this document makes no
          independent request. No per-type service names
          (<tt>llm-api</tt>, <tt>tool</tt>, etc.) are requested now or
          anticipated.
        </t>
      </section>

      <section anchor="iana-protocols" title="DAWN Entity Protocols Registry">
        <t>
          IANA is requested to create the "DAWN Entity Protocols" registry,
          with registration procedure "Specification Required". The initial
          contents are: <tt>a2a</tt> (Agent2Agent protocol), <tt>mcp</tt>
          (Model Context Protocol), <tt>openresponses</tt> (Open Responses
          API profile), <tt>https</tt> (generic HTTPS endpoint). This
          document defines the common key <tt>rtype</tt>
          (<xref target="dns-keys"/>).
        </t>
      </section>
    </section>
  </middle>

  <back>
    <references title="Normative References">
      <reference anchor="RFC2119" target="https://www.rfc-editor.org/rfc/rfc2119">
        <front><title>Key words for use in RFCs to Indicate Requirement Levels</title><author initials="S." surname="Bradner" fullname="Scott Bradner"/><date year="1997" month="March"/></front>
        <seriesInfo name="BCP" value="14"/><seriesInfo name="RFC" value="2119"/>
      </reference>
      <reference anchor="RFC8174" target="https://www.rfc-editor.org/rfc/rfc8174">
        <front><title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title><author initials="B." surname="Leiba" fullname="Barry Leiba"/><date year="2017" month="May"/></front>
        <seriesInfo name="BCP" value="14"/><seriesInfo name="RFC" value="8174"/>
      </reference>
      <reference anchor="RFC6762" target="https://www.rfc-editor.org/rfc/rfc6762">
        <front><title>Multicast DNS</title><author initials="S." surname="Cheshire" fullname="Stuart Cheshire"/><author initials="M." surname="Krochmal" fullname="Marc Krochmal"/><date year="2013" month="February"/></front>
        <seriesInfo name="RFC" value="6762"/>
      </reference>
      <reference anchor="RFC6763" target="https://www.rfc-editor.org/rfc/rfc6763">
        <front><title>DNS-Based Service Discovery</title><author initials="S." surname="Cheshire" fullname="Stuart Cheshire"/><author initials="M." surname="Krochmal" fullname="Marc Krochmal"/><date year="2013" month="February"/></front>
        <seriesInfo name="RFC" value="6763"/>
      </reference>
      <reference anchor="RFC6335" target="https://www.rfc-editor.org/rfc/rfc6335">
        <front><title>Internet Assigned Numbers Authority (IANA) Procedures for the Management of the Service Name and Transport Protocol Port Number Registry</title><author initials="M." surname="Cotton" fullname="Michelle Cotton"/><author initials="L." surname="Eggert" fullname="Lars Eggert"/><author initials="J." surname="Touch" fullname="Joe Touch"/><author initials="M." surname="Westerlund" fullname="Magnus Westerlund"/><author initials="S." surname="Cheshire" fullname="Stuart Cheshire"/><date year="2011" month="August"/></front>
        <seriesInfo name="BCP" value="165"/><seriesInfo name="RFC" value="6335"/>
      </reference>
      <reference anchor="RFC8126" target="https://www.rfc-editor.org/rfc/rfc8126">
        <front><title>Guidelines for Writing an IANA Considerations Section in RFCs</title><author initials="M." surname="Cotton" fullname="Michelle Cotton"/><author initials="B." surname="Leiba" fullname="Barry Leiba"/><author initials="T." surname="Narten" fullname="Thomas Narten"/><date year="2017" month="June"/></front>
        <seriesInfo name="BCP" value="26"/><seriesInfo name="RFC" value="8126"/>
      </reference>
      <reference anchor="RFC8552" target="https://www.rfc-editor.org/rfc/rfc8552">
        <front><title>Scoped Interpretation of DNS Resource Records through "Underscored" Naming of Attribute Leaves</title><author initials="D." surname="Crocker" fullname="Dave Crocker"/><date year="2019" month="March"/></front>
        <seriesInfo name="BCP" value="222"/><seriesInfo name="RFC" value="8552"/>
      </reference>
      <reference anchor="I-D.jakab-dawn-agent-discovery-mdns" target="https://datatracker.ietf.org/doc/draft-jakab-dawn-agent-discovery-mdns/">
        <front><title>Agent Discovery Using mDNS</title><author initials="L." surname="Jakab" fullname="Lorand Jakab"/><date year="2026" month="October"/></front>
        <seriesInfo name="Internet-Draft" value="draft-jakab-dawn-agent-discovery-mdns"/>
      </reference>
      <reference anchor="I-D.blwm-dawn-common-discovery-parameters" target="https://datatracker.ietf.org/doc/draft-blwm-dawn-common-discovery-parameters/">
        <front><title>Common Agent DNS Discovery Parameters</title><author initials="F." surname="Brockners" fullname="Frank Brockners"/><author initials="L." surname="Jakab" fullname="Lorand Jakab"/><author initials="N." surname="Williams" fullname="Nic Williams"/><author initials="J." surname="Mozley" fullname="Jim Mozley"/><date year="2026" month="October"/></front>
        <seriesInfo name="Internet-Draft" value="draft-blwm-dawn-common-discovery-parameters"/>
      </reference>
      <reference anchor="I-D.cui-dawn-mdi-model" target="https://datatracker.ietf.org/doc/draft-cui-dawn-mdi-model/">
        <front><title>Minimum Public Discovery Information: Requirements and Common Semantics</title><author initials="Y." surname="Cui" fullname="Yong Cui"/><author initials="K." surname="Yao" fullname="Kehan Yao"/><date year="2026" month="November"/></front>
        <seriesInfo name="Internet-Draft" value="draft-cui-dawn-mdi-model"/>
      </reference>
      <reference anchor="I-D.farrel-dawn-terminology" target="https://datatracker.ietf.org/doc/draft-farrel-dawn-terminology/">
        <front><title>Terminology for the Discovery of Agents, Workloads, and Named Entities (DAWN)</title><author initials="A." surname="Farrel" fullname="Adrian Farrel"/><author initials="K." surname="Yao" fullname="Kehan Yao"/><author initials="R." surname="Schott" fullname="Roland Schott"/><author initials="N." surname="Williams" fullname="Nic Williams"/><date year="2026" month="November"/></front>
        <seriesInfo name="Internet-Draft" value="draft-farrel-dawn-terminology"/>
      </reference>
    </references>

    <references title="Informative References">
      <reference anchor="I-D.cui-dawn-local-llm-api-discovery" target="https://datatracker.ietf.org/doc/draft-cui-dawn-local-llm-api-discovery/">
        <front><title>Discovery of LLM API Services in Local Networks</title><author initials="Y." surname="Cui" fullname="Yong Cui"/><date year="2026" month="November"/></front>
        <seriesInfo name="Internet-Draft" value="draft-cui-dawn-local-llm-api-discovery"/>
      </reference>
      <reference anchor="I-D.mozleywilliams-dawn-dnsaid" target="https://datatracker.ietf.org/doc/draft-mozleywilliams-dawn-dnsaid/">
        <front><title>DNS-Based Agent Discovery (DNSAID)</title><author initials="J." surname="Mozley" fullname="Jim Mozley"/><author initials="N." surname="Williams" fullname="Nic Williams"/><date year="2026" month="October"/></front>
        <seriesInfo name="Internet-Draft" value="draft-mozleywilliams-dawn-dnsaid"/>
      </reference>
      <reference anchor="RFC3444" target="https://www.rfc-editor.org/rfc/rfc3444">
        <front><title>On the Difference between Information Models and Data Models</title><author initials="A." surname="Pras" fullname="Aiko Pras"/><author initials="J." surname="Schoenwaelder" fullname="Juergen Schoenwaelder"/><date year="2003" month="January"/></front>
        <seriesInfo name="RFC" value="3444"/>
      </reference>
    </references>

    <section anchor="appendix-json" title="Illustrative JSON Examples">
      <t>
        Illustrative only; no encoding is normative. The generalized MDI
        model is encoding-neutral per <xref target="RFC3444"/>; these
        examples show one possible JSON binding.
      </t>

      <section anchor="appendix-json-agent" title="Named Agent (A2A)">
        <figure>
          <artwork type="json" name="agent-example.json"><![CDATA[
{
  "entity_id": "urn:example:agent:invoice-bot",
  "entity_type": "agent",
  "endpoint": [
    { "locator_type": "uri",
      "locator": "https://example.com/agent",
      "protocols": [ "a2a" ] }
  ],
  "capability_ref": [
    { "ref_type": "uri",
      "locator": "https://example.com/.well-known/agent-card.json" }
  ],
  "freshness": { "valid_until": "2027-03-01T00:00:00Z" }
}
]]></artwork>
        </figure>
      </section>

      <section anchor="appendix-json-llm" title="LLM API Service (Open Responses)">
        <figure>
          <artwork type="json" name="llm-example.json"><![CDATA[
{
  "entity_id": "office-llm",
  "entity_type": "service",
  "entity_summary": [ "Office LLM API (Open Responses)" ],
  "endpoint": [
    { "locator_type": "uri",
      "locator": "https://llm-node.local:8443/v1",
      "protocols": [ "openresponses" ] }
  ],
  "authentication_hint": [
    { "scheme": "bearer", "metadata": "required" }
  ],
  "freshness": { "version": "2026-04-24" }
}
]]></artwork>
        </figure>
      </section>

      <section anchor="appendix-json-catalog" title="Catalog (Discovery-Furthering Service)">
        <figure>
          <artwork type="json" name="catalog-example.json"><![CDATA[
{
  "entity_id": "lab-catalog",
  "entity_type": "catalog",
  "entity_summary": [ "Lab entity catalog" ],
  "endpoint": [
    { "locator_type": "uri",
      "locator": "https://catalog.example.com/mdi",
      "protocols": [ "https" ] }
  ],
  "capability_ref": [
    { "ref_type": "uri",
      "locator": "https://catalog.example.com/mdi",
      "digest": "sha-256=<base64url digest of the catalog document>" }
  ]
}
]]></artwork>
        </figure>
        <t>
          The catalog document located by <tt>capability_ref</tt> is a list
          whose entries are MDI instances (<xref target="structure"/>) or
          references to them; its own entries need not repeat the
          catalog's own scope hints.
        </t>
      </section>
    </section>

    <section anchor="appendix-mapping" title="Mapping Across the Converging Documents">
      <t>
        The table sketches how the generalized MDI elements surface in the
        converging discovery documents. Field names are indicative; full
        mappings belong to binding documents.
      </t>
      <table>
        <name>MDI Element Mapping Across Documents</name>
        <thead>
          <tr>
            <th>MDI element</th>
            <th>MDI model (cui)</th>
            <th>jakab (_agent)</th>
            <th>blwm (_agents)</th>
            <th>local llm-api</th>
          </tr>
        </thead>
        <tbody>
          <tr><td>Entity Identifier</td><td>entity-identifier</td><td>agentId / cardUrl</td><td>instance name</td><td>instance name</td></tr>
          <tr><td>Entity Type</td><td>entity-type</td><td>(default: agent)</td><td>rtype (this doc)</td><td>(fixed: service)</td></tr>
          <tr><td>Endpoint locator</td><td>locator</td><td>SRV target+port</td><td>SVCB target+port</td><td>SRV target</td></tr>
          <tr><td>Endpoint protocols</td><td>protocols</td><td>proto</td><td>proto</td><td>api</td></tr>
          <tr><td>Capability Reference</td><td>capability-reference</td><td>cardUrl</td><td>desc-ref</td><td>(path/base)</td></tr>
          <tr><td>Capability Summary</td><td>capability-summary</td><td>caps</td><td>(profile)</td><td>n/a</td></tr>
          <tr><td>Freshness version</td><td>version</td><td>ver</td><td>proto-ver</td><td>protovers</td></tr>
          <tr><td>Authentication Hint</td><td>auth-hint</td><td>n/a</td><td>(profile)</td><td>auth</td></tr>
          <tr><td>Human-readable name</td><td>entity-summary</td><td>instance name</td><td>title</td><td>instance name</td></tr>
        </tbody>
      </table>
    </section>

    <section anchor="appendix-llm-absorption" title="How This Document Absorbs draft-cui-dawn-local-llm-api-discovery">
      <t>
        The LLM API discovery draft
        <xref target="I-D.cui-dawn-local-llm-api-discovery"/> proposes a
        separate <tt>_llm-api._tcp</tt> service type. Under this
        specification, the same capability is achieved as follows:
      </t>
      <table>
        <name>LLM API Draft to This Specification Mapping</name>
        <thead>
          <tr><th>LLM API draft</th><th>This specification</th></tr>
        </thead>
        <tbody>
          <tr><td><tt>_llm-api._tcp</tt> service type</td><td><tt>_agents._tcp</tt> (reused)</td></tr>
          <tr><td><tt>api=openresponses</tt></td><td><tt>proto=openresponses</tt> (common key)</td></tr>
          <tr><td><tt>protovers=2026-04-24</tt></td><td><tt>proto-ver=2026-04-24</tt> (common key)</td></tr>
          <tr><td><tt>scheme=https</tt></td><td>derived from <tt>alpn</tt> (h2 &#8594; https)</td></tr>
          <tr><td><tt>base=/v1</tt></td><td><tt>path=/v1</tt> (common key)</td></tr>
          <tr><td><tt>auth=required</tt></td><td><tt>auth=required</tt> (common key)</td></tr>
          <tr><td>(no type)</td><td><tt>rtype=service</tt> (new common key)</td></tr>
        </tbody>
      </table>
      <t>
        No information is lost. The client discovers LLM APIs with the same
        single browse that finds agents and MCP servers.
      </t>
    </section>

    <section anchor="acknowledgements" title="Acknowledgements">
      <t>
        Thanks to the DAWN WG participants for discussions on entity
        discovery, and to the authors of
        <xref target="I-D.jakab-dawn-agent-discovery-mdns"/>,
        <xref target="I-D.blwm-dawn-common-discovery-parameters"/>,
        <xref target="I-D.mozleywilliams-dawn-dnsaid"/>, and
        <xref target="I-D.cui-dawn-mdi-model"/> whose work this document
        builds upon.
      </t>
    </section>
  </back>
</rfc>
