Network Working Group B. Zhang, Ed.
Internet-Draft Pengcheng Laboratory
Intended status: Informational L. Dunbar
Expires: 23 April 2027 Futurewei
Y. Yang, Ed.
Sun Yat-sen University
20 October 2026
A Framework for Agent Discovery in DAWN
draft-zhang-dawn-agent-discovery-framework-09
Abstract
This document describes a two-layer reference architecture for
discovering the minimum public information needed to reach AI agents.
The Local Discovery Plane obtains discovery information within an
administrative domain through mDNS/DNS-SD or infrastructure-assisted
discovery. For cross-domain discovery, the Federation Plane uses
publication of policy-approved information in authoritative DNS zones
and on-demand resolution as its baseline; Federation Gateways apply
export policy and validate and cache accepted results. The document
also identifies optional functions that may complement this baseline
where DNS publication and request-driven resolution do not meet
deployment needs: recipient-specific sharing, proactive or multi-hop
dissemination, and trusted aggregated directory queries. In mixed-
mode deployments, DNS may bootstrap an endpoint that provides one of
these additional functions.
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 23 April 2027.
Zhang, et al. Expires 23 April 2027 [Page 1]
Internet-Draft A Framework for Agent Discovery in DAWN October 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 . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1. Document Scope . . . . . . . . . . . . . . . . . . . . . 4
1.2. Out of Scope . . . . . . . . . . . . . . . . . . . . . . 5
1.3. Document Organization . . . . . . . . . . . . . . . . . . 6
2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 6
3. Architectural Requirements . . . . . . . . . . . . . . . . . 9
3.1. Functional Requirements . . . . . . . . . . . . . . . . . 9
3.2. Non-Functional Requirements . . . . . . . . . . . . . . . 11
4. Relationship to DAWN Family Protocols . . . . . . . . . . . . 11
4.1. ACAP (Agent Capability Advertisement Protocol) . . . . . 12
4.2. Agent Directory . . . . . . . . . . . . . . . . . . . . . 12
4.3. AGNTCY Agent Directory Service (ADS) . . . . . . . . . . 12
4.4. DNS-Based Naming Proposals (DNS-AID, DN-ANR) . . . . . . 13
4.5. MDI (Minimum Discoverable Information) . . . . . . . . . 13
5. State-of-the-Art and Gap Analysis . . . . . . . . . . . . . . 14
5.1. Centralized Global Directory . . . . . . . . . . . . . . 14
5.2. Link-Local Multicast Discovery (mDNS/DNS-SD) . . . . . . 14
5.3. Gap Summary . . . . . . . . . . . . . . . . . . . . . . . 14
6. Two-Layer Reference Architecture . . . . . . . . . . . . . . 15
6.1. Layer 1: Local Discovery Plane . . . . . . . . . . . . . 17
6.1.1. Locally Advertised or Discovered Resources . . . . . 17
6.1.2. Infrastructure-Maintained Managed Resources . . . . . 18
6.1.3. Common Local Discovery Processing . . . . . . . . . . 19
6.2. Layer 2: Federation Plane . . . . . . . . . . . . . . . . 19
6.3. Federation Operation Model . . . . . . . . . . . . . . . 20
6.3.1. Federation Roles and Trust Relationships . . . . . . 20
6.3.2. Federation Workflow . . . . . . . . . . . . . . . . . 21
6.3.3. FAD Scope and Separation from Agent Communication . . 23
6.3.4. Attachment Context and Endpoint Resolution . . . . . 23
6.4. Functional Components . . . . . . . . . . . . . . . . . . 23
7. Additional Discovery Functions Beyond the DNS Baseline . . . 24
7.1. Recipient-Specific Sharing . . . . . . . . . . . . . . . 25
Zhang, et al. Expires 23 April 2027 [Page 2]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
7.2. Proactive or Multi-Hop Dissemination . . . . . . . . . . 25
7.3. Trusted Directory Publication . . . . . . . . . . . . . . 26
7.4. Mixed-Mode Federation . . . . . . . . . . . . . . . . . . 26
8. Security Reference Model . . . . . . . . . . . . . . . . . . 27
8.1. Layer 1: Transport Security . . . . . . . . . . . . . . . 27
8.2. Layer 2: FMR Origin Authentication and Integrity . . . . 27
8.3. Layer 3: Access Control on Discovery Information . . . . 28
8.4. Layer 4: Federation Admission Control . . . . . . . . . . 28
8.5. Denial-of-Service Considerations . . . . . . . . . . . . 28
9. Resilience and Operational Considerations . . . . . . . . . . 28
9.1. Network Partition Handling . . . . . . . . . . . . . . . 28
9.2. Gateway Failure and Recovery . . . . . . . . . . . . . . 29
9.3. Graceful Shutdown . . . . . . . . . . . . . . . . . . . . 30
9.4. TTL-Based Stale Record Aging . . . . . . . . . . . . . . 31
9.5. Version Evolution and Forward Compatibility . . . . . . . 32
9.6. Operational Monitoring . . . . . . . . . . . . . . . . . 32
10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 33
11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 33
12. Informative References . . . . . . . . . . . . . . . . . . . 33
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 36
1. Introduction
The IETF DAWN working group
[I-D.akhavain-moussa-dawn-problem-statement] is chartered to enable a
client to discover the minimum public discovery information about AI
resources before interacting with them, covering the local network,
an organization, and cooperating parties with trust relationships
over the public Internet. Its problem statement identifies that no
single existing mechanism satisfies all discovery scenarios,
particularly when crossing trust and administrative boundaries
[I-D.moussa-dawn-gap-analysis]. DAWN requirements
[I-D.king-dawn-requirements] and use cases [I-D.kay-dawn-use-cases]
scope discovery to obtaining the minimum public discovery information
needed to reach and communicate with a resource: type, reachability
information, communication protocol, and services.
A growing number of individual submissions propose concrete
mechanisms within this space: DNS-based naming extensions (DNS-AID
[I-D.mozleywilliams-dnsop-dnsaid], DN-ANR
[I-D.cui-dns-native-agent-naming-resolution]); DNS-like hierarchical
architectures (Agent Root, Agent Registry, Agent Resolver
[I-D.yao-dawn-agent-discovery-architect]); mDNS/DNS-SD-based two-
phase discovery [I-D.jakab-dawn-agent-discovery-mdns]; host-level
self-description (A2A Agent Cards, ACAP [I-D.zahed-acap], ANP agent-
descriptions, api-catalog [RFC9727]); and registry protocols (Agent
Directory [I-D.jimenez-agent-directory], AGNTCY ADS
[I-D.mp-agntcy-ads]). A survey by Jimenez et al.
Zhang, et al. Expires 23 April 2027 [Page 3]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
[I-D.jimenez-dawn-discovery-landscape] compares these mechanisms and
identifies gaps, including the lack of interoperable federation
across administrative boundaries.
Cui [I-D.cui-dawn-mdi-model] proposes an encoding-neutral information
model for Minimum Discoverable Information (MDI) -- a common header
that any discovery response may carry, independent of transport or
format. MDI defines three mandatory elements (Entity Identifier,
Entity Type, Endpoint) and a short set of recommended elements
(Capability Summary, Authentication Hint, Trust Reference,
Freshness).
This document complements the above work by defining a two-layer
reference architecture for initial discovery within and across
administrative boundaries. The Federation Gateway (FGW) connects
local discovery to DNS publication and resolution. Its Export Policy
Engine selects the minimum information that may leave the site. A
Federation Metadata Record (FMR) represents that information, while
the Federated Agent Directory (FAD) caches accepted results on
demand. Discovery may continue through a referral to another
discovery endpoint.
Local discovery uses mDNS/DNS-SD or infrastructure-assisted
discovery. Across administrative domains, DNS publication and on-
demand resolution provide the baseline: each site publishes permitted
information in its authoritative DNS zone, and remote sites resolve,
validate, and cache it when needed. Section 7 identifies optional
additional functions for deployments whose requirements are not met
by that baseline: recipient-specific sharing, proactive or multi-hop
dissemination, and trusted aggregated directory queries. A mixed-
mode deployment may use DNS to discover an endpoint that provides an
additional function.
1.1. Document Scope
This document defines a two-layer reference architecture for
discovering the minimum public information needed to reach AI
resources within a local network, an organization, or cooperating
administrative domains. DNS and mDNS provide the baseline
mechanisms. Non-DNS mechanisms and mixed-mode deployments may
complement the baseline by supporting additional functions include
recipient-specific sharing, proactive or multi-hop dissemination, and
trusted aggregated directory queries.. Discovery returns an endpoint,
communication protocol, and any applicable reference needed to
continue discovery.
Zhang, et al. Expires 23 April 2027 [Page 4]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
1.2. Out of Scope
The following items are explicitly out of scope for this
informational document, consistent with the DAWN charter:
* Normative protocol message formats, encodings, or state machines.
These are addressed in separate protocol specifications that may
be developed based on this architecture.
* Mandatory cryptographic algorithms or cipher suites. The
architecture requires transport security and metadata integrity,
but specific algorithm selection is deferred to protocol
documents.
* A global root trust authority or certificate hierarchy. The
architecture supports federation-local trust models, consistent
with DAWN's initial scope [CSA-DAWN-Note].
* Exchange of capability information beyond the minimum public
discovery information; in particular, catalogs of capabilities are
out of scope. The FMR carries only the minimum public discovery
information (type, reachability, communication protocol,
services).
* All communication beyond the initial discovery process, including
any interaction, invocation, or selection that follows discovery.
* Open-ended search, semantic matchmaking, ranking, or the selection
of an AI resource based on capabilities or user intent.
* Identity management of AI agents, tools, and skills.
* Trust management and trust evaluation methods beyond secure
content exchange between DAWN parties. The architecture provides
integrity for exchanged discovery information but does not define
a general trust framework.
* Agent runtime execution, task scheduling, or inter-agent
communication protocols beyond discovery.
* Business-level service level agreements or commercial federation
governance rules.
Zhang, et al. Expires 23 April 2027 [Page 5]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* Open-Internet-scale agent indexing -- the "AI resource indexing
across the wider Internet" case excluded by the DAWN charter -- is
out of scope. Trusted directory publication (Section 7.2.3) is
limited to publication to trusted or semi-public directories under
explicit operator policy, and does not constitute open-Internet
indexing.
1.3. Document Organization
Section 2 defines terminology, aligned with DAWN
[I-D.farrel-dawn-terminology]. Section 3 states architectural
requirements and maps them to DAWN requirements. Section 4 positions
this architecture relative to existing DAWN family protocols.
Section 5 provides a state-of-the-art review and gap analysis.
Section 6 presents the two-layer reference architecture, federation
workflow, and functional components. Section 7 identifies optional
discovery functions beyond the DNS baseline and describes mixed-mode
DNS bootstrap. Section 8 identifies existing security mechanisms and
disclosure boundaries. Section 9 covers resilience, publication
updates, and caching.
2. Terminology
This document uses terms defined in [I-D.farrel-dawn-terminology],
including Entity, Discovery, Discovery Information, Discoverable
Object, Minimum Discoverable Information (MDI), Capability Card, and
Trust Indicator. The following additional terms are defined for the
purpose of this architecture:
* *Agent*: An autonomous network-resident entity that can execute
tasks, expose service interfaces, and advertise capability
metadata. Aligns with the DAWN definition of Entity when the
entity type is "agent".
* *Administrative Domain*: A network or set of networks under a
single organizational authority, with its own security and
operational policies. Equivalent to the DAWN scope concept.
* *Federation Metadata Record (FMR)*: A lightweight, sanitized
representation of the minimum public discovery information about
an AI resource that is made available through the Federation
Plane. An FMR may include the Entity Identifier, Entity Type,
reachability or locator information, communication protocol,
services, status, and freshness information.
* *Capability Card*: A fuller description of an agent's
capabilities. Exchange of capability information beyond the
minimum public discovery information is out of scope for DAWN's
Zhang, et al. Expires 23 April 2027 [Page 6]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
initial phase (see Section 1.2); this architecture does not
distribute or retrieve Capability Cards as part of discovery.
ACAP's Agent Capability Document (ACD) [I-D.zahed-acap] and A2A
Agent Cards are examples of such fuller descriptions.
* *Site / Local Domain*: An independent administrative area,
typically one local-area network or a single tenant namespace,
where agents run.
* *Federation Gateway (FGW)*: A functional component that obtains
local agent discovery information, applies export policy, and
arranges publication of permitted information in the domain's
authoritative DNS service. For cross-domain requests, it obtains
remote discovery information through DNS resolution, validates the
results, and caches accepted information. DNS publication and
resolution services MAY operate separately from the FGW.
* *Federation*: A trust group formed by multiple federation gateways
from different administrative domains, for exchanging agent
discovery metadata.
* *Federation Peer*: The DNS-based baseline for the Federation
Plane. Each domain publishes permitted discovery information in
its authoritative DNS zone. A requesting FGW resolves and
validates that information when no valid matching FAD entry
exists, then caches accepted results for their remaining permitted
lifetime. This model does not require persistent FGW-to-FGW
sessions or complete replication of remote records. Deployment
considerations are described in Section 7.1.
* *Additional Discovery Functions*: Optional functions that
complement DNS publication and on-demand resolution where
deployment requirements call for recipient-specific sharing,
proactive or multi-hop dissemination, or trusted aggregated
directory queries. DNS may bootstrap an endpoint that provides
such a function in a mixed-mode deployment. Section 7 describes
these functions without specifying their protocols.
* *Minimum Public Discovery Information*: The set of public
properties a client needs before interacting with an AI resource:
its type, reachability information, communication protocol, and
services. This information MAY be represented using the DAWN MDI
model [I-D.cui-dawn-mdi-model] and MAY be obtained through agent
advertisement or the Resource Publication Hook
[I-D.dunbar-dawn-infra-assisted-discovery]. An FMR carries this
information together with the identifier, status, and time-to-live
described in its definition; capability information beyond this
minimum remains out of scope.
Zhang, et al. Expires 23 April 2027 [Page 7]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* *Iterative Discovery / Referral*: A discovery process in which a
returned record refers the client to a catalog or to another
discovery endpoint (referral, delegation, or indirection),
continuing until the client obtains the minimum public discovery
information it needs.
* *Federated Agent Directory (FAD)*: The discovery-information store
maintained by an FGW. Remote discovery information is obtained on
demand and cached after validation. Cached information is
retained only for its remaining permitted lifetime. Locally
originated discovery information MAY be maintained within the same
store or separately.
* *Export Policy Engine*: A functional module on the FGW that
filters, redacts, and controls which agent metadata may be shared
with which federation peers. This is the core mechanism for data
sovereignty.
* *Local Discovery Plane*: The first architectural layer,
responsible for obtaining and maintaining lightweight AI-resource
discovery information through existing local discovery mechanisms
or infrastructure-assisted discovery. The architecture does not
mandate a specific local discovery protocol.
* *Infrastructure-Assisted Discovery (IAD)*: An infrastructure-
assisted approach to discovery described in
[I-D.dunbar-dawn-infra-assisted-discovery]. In this architecture,
IAD provides deployment-maintained agent discovery information to
the FGW, primarily for organization-level discovery in managed
environments.
* *Local Infrastructure Gateway*: An authorized orchestration,
placement, hosting, or management system that collects and
maintains discovery information on behalf of agents deployed
within a managed organizational environment. It MAY be
implemented separately from, or integrated with, the FGW. When
implemented separately, it advertises the collected agent
discovery information to the FGW and provides updates as
deployment state changes. When integrated, the FGW primarily
obtains agent discovery information from registration records
maintained within the managed infrastructure. Its local discovery
role does not itself imply participation in cross-domain
federation.
Zhang, et al. Expires 23 April 2027 [Page 8]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* *Resource Publication Hook*: The infrastructure-assisted mechanism
described in [I-D.dunbar-dawn-infra-assisted-discovery] through
which deployment infrastructure supplies resource discovery
information for local registration and maintenance. In this
architecture, the FGW MAY use the supplied information to
construct FMRs after applying export policy.
* *Attachment Context Hook*: An infrastructure integration point
through which trusted network infrastructure provides the
requester's current Attachment Scope to the discovery mechanism,
as described in [I-D.dunbar-dawn-infra-assisted-discovery]. This
context complements published resource information when
identifying applicable discovery domains and reachable agent
endpoints. Its detailed realization is outside the scope of this
document.
* *Federation Plane*: The second architectural layer, operating
across administrative domains, responsible for FMR exchange among
FGWs.
3. Architectural Requirements
This section defines high-level architectural requirements for the
two-layer agent-discovery architecture. These requirements are
derived from the DAWN working group's problem statement
[I-D.akhavain-moussa-dawn-problem-statement], requirements
[I-D.king-dawn-requirements], and use cases [I-D.kay-dawn-use-cases].
They are non-normative for this informational document.
3.1. Functional Requirements
1. *F-1: Cross-Domain Discovery* -- The architecture MUST enable
discovery of the minimum public discovery information about
resources that reside in administrative domains different from
the requester's domain, i.e., between cooperating parties with
trust relationships over the public Internet. Satisfies DAWN use
case "Administrative Scope Extensions" [I-D.kay-dawn-use-cases].
2. *F-2: Zero-Configuration Local Discovery* -- Agents MUST be
capable of advertising themselves on the local network without
pre-provisioned discovery server addresses. Satisfies DAWN
requirements for local-context discovery
[I-D.king-dawn-requirements].
Zhang, et al. Expires 23 April 2027 [Page 9]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
3. *F-3: Data Sovereignty* -- Each site administrator MUST retain
full control over which agent metadata records can be shared
outside its local domain, including per-record filtering, field
redaction, and per-peer access policy. Addresses the gap
identified in [I-D.moussa-dawn-gap-analysis] regarding lack of
interoperable federation with per-domain policy control.
4. *F-4: Decentralized Federation* -- The architecture MUST NOT
require a mandatory global centralized directory. Federation
participants MUST be able to operate without submitting data to a
third-party central authority. Aligns with DAWN's emphasis on
avoiding single points of failure [I-D.king-dawn-requirements].
5. *F-5: Minimum Public Discovery Information Only* -- The
architecture MUST exchange only the minimum public discovery
information (type, reachability, communication protocol,
services) in the federation plane, and MUST NOT distribute or
retrieve capability information beyond this minimum. This
instantiates the DAWN MDI principle of a thin, encoding-neutral
record [I-D.cui-dawn-mdi-model].
6. *F-6: Dynamic Agent Lifecycle* -- The architecture MUST support
changes in agent availability, deployment state, reachable
endpoints, and export permission. After detecting a relevant
change, the responsible publisher MUST promptly update or remove
the affected authoritative DNS records. Previously cached copies
remain subject to their established validity limits. These
procedures support dynamic updates and freshness
[I-D.king-dawn-requirements].
7. *F-7: DNS-Based Federation* -- The architecture MUST support
independently administered DNS publication and request-driven
retrieval. A requesting FGW checks for a valid matching cached
result before resolving remote discovery information. Accepted
results retain their provenance, applicable disclosure
conditions, and remaining permitted lifetime.
8. *F-8: Iterative Discovery and Referral* -- The architecture MUST
support iterative discovery in which a returned record may refer
the client to a catalog or to another discovery endpoint
(referral, delegation, or indirection). The AI resource to be
discovered may itself be a catalog or service that furthers the
discovery process.
9. *F-9: Integrity of Discovery Information* -- The architecture
MUST provide integrity for all discovery information exchanged
between DAWN parties, allowing a recipient to detect alteration
of, or unauthorized injection into, the exchanged records.
Zhang, et al. Expires 23 April 2027 [Page 10]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
3.2. Non-Functional Requirements
1. *NF-1: Security Against Forged Advertisements* -- The
architecture MUST provide a mechanism to verify the origin and
integrity of FMRs, preventing malicious federation participants
from injecting forged agent advertisements.
2. *NF-2: Fault Tolerance* -- The architecture MUST tolerate
individual gateway failures, network partitions, and transient
connectivity loss without catastrophic loss of discovery
capability.
3. *NF-3: Interoperability* -- The architecture reuses existing DNS/
mDNS procedures and the information models and bindings defined
by the applicable discovery specifications. Additional
mechanisms follow their own existing specifications and preserve
the publication, validation, and validity conditions of the
information they handle. Interoperability does not require a new
common federation wire format or gateway session protocol.
4. *NF-4: Privacy Protection* -- The architecture MUST prevent
leakage of internal network topology, private IP addresses, and
sensitive agent details to unauthorized external parties.
Satisfies DAWN privacy considerations
[I-D.king-dawn-requirements].
5. *NF-5: Scalability* -- The architecture MUST scale from small
federations of a few sites to large federations of hundreds or
thousands of sites, without requiring every site to maintain
separate manual relationships with all other sites.
6. *NF-6: Operational Simplicity* -- The architecture MUST minimize
operational overhead for site administrators, particularly for
local-site deployment where zero-configuration is a primary goal.
4. Relationship to DAWN Family Protocols
This architecture is designed to complement, not replace, existing
protocols being discussed in the DAWN community and adjacent efforts.
The following positioning statements clarify the relationship.
Zhang, et al. Expires 23 April 2027 [Page 11]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
4.1. ACAP (Agent Capability Advertisement Protocol)
ACAP [I-D.zahed-acap] defines a well-known URI scheme, a rich Agent
Capability Document (ACD) format, and HTTP/3 operations (GET, PUT,
POST) for retrieval, registration, and search within a site. ACAP
operates primarily at the Description Plane: it specifies how a host
publishes its capabilities and how a domain-level query endpoint
answers discovery queries.
This architecture treats ACAP as a valid implementation of the
Description Plane within a site. An FGW may collect ACDs via ACAP
PUT/GET from local agents, extract the minimum public discovery
information (type, reachability, communication protocol, services)
into an FMR, and distribute the FMR in the Federation Plane; the
richer capability content of the ACD is not carried in the FMR,
consistent with the charter. Cross-domain discovery queries may be
answered from valid FAD entries. In DNS-based discovery, a cache
miss triggers a DNS query for the remote domain, followed by
validation and caching of the result. A global ACAP query endpoint
is not required. Thus, ACAP and this architecture are composable:
ACAP handles intra-domain description, while this architecture
handles inter-domain federation and export policy.
4.2. Agent Directory
The Agent Directory [I-D.jimenez-agent-directory] adapts the CoRE
Resource Directory [RFC9176] to software agents, providing an HTTP/
JSON service for registration and lookup with soft-state lifetimes.
It defines a well-known URI (/.well-known/ad) and supports capability
filtering.
This architecture is complementary: the Agent Directory may serve as
the Local Discovery Plane mechanism within a site, collecting agent
registrations and answering local queries. The FGW may subscribe to
the local Agent Directory, extract MDI-compliant records, apply
export policy, and generate FMRs for federation. The FAD maintained
by the FGW is conceptually a distributed extension of the Agent
Directory across administrative boundaries. Federation across
multiple Agent Directory instances is acknowledged but not specified
in [I-D.jimenez-agent-directory]; this architecture provides one
possible federation model.
4.3. AGNTCY Agent Directory Service (ADS)
AGNTCY ADS [I-D.mp-agntcy-ads] is a distributed directory using
libp2p Kad-DHT for content routing and a hierarchical skill taxonomy.
It supports open, Internet-scale discovery, which is outside DAWN's
initial scope.
Zhang, et al. Expires 23 April 2027 [Page 12]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
This architecture differs in approach: ADS is a global or wide-area
distributed hash table, while this architecture is a federated
gateway model with explicit per-peer trust and policy control. ADS
targets open, Internet-scale discovery, which is excluded by the DAWN
charter; this architecture targets closed or semi-closed federations
where data sovereignty and per-peer export policy are paramount. A
hybrid deployment is possible: an FGW may publish a subset of its
FMRs to a trusted directory such as AGNTCY ADS for broader
discoverability, while retaining sensitive records for private
federation peers only (see trusted directory publication,
Section 7.2.3).
4.4. DNS-Based Naming Proposals (DNS-AID, DN-ANR)
DNS-AID [I-D.mozleywilliams-dnsop-dnsaid] and DN-ANR
[I-D.cui-dns-native-agent-naming-resolution] propose DNS-based
mechanisms for agent naming and location resolution. These operate
at the DAWN Naming Plane, providing decentralized name-to-location
mapping.
This architecture treats DNS-based discovery as one valid mechanism
for the Local Discovery Plane (e.g., resolving a known partner's
domain to locate its FGW or agents). The FGW may itself be
advertised via DNS records, and FMRs may contain DNS-based locators.
The two approaches are orthogonal and composable.
4.5. MDI (Minimum Discoverable Information)
The MDI information model [I-D.cui-dawn-mdi-model] defines an
abstract, encoding-neutral common header for discovery responses.
This architecture directly adopts MDI as the basis for FMRs: an FMR
is a concrete, protocol-specific binding of MDI elements (Entity
Identifier, Entity Type, Endpoint, Capability Summary, Freshness,
Provenance) for the purpose of cross-gateway exchange. By grounding
FMRs in MDI, this architecture ensures that federated metadata can be
translated to and from other DAWN discovery formats (ACAP ACDs, Agent
Directory records, A2A Agent Cards) without loss of the minimal
necessary information.
The FMR is a thin description of one entity rather than a capability
catalog. Its identifier is interpreted within the originating
publisher's scope; reachability and communication protocol
information are associated with the endpoint. Publication and
caching lifetimes follow the applicable discovery mechanism. This
framework does not define an FMR wire format or extension encoding;
those details belong to the specifications for the mechanisms that
carry the information.
Zhang, et al. Expires 23 April 2027 [Page 13]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
5. State-of-the-Art and Gap Analysis
This section reviews existing approaches to cross-domain service and
agent discovery, and identifies gaps that motivate the proposed two-
layer architecture. It extends the gap analysis in
[I-D.moussa-dawn-gap-analysis] and
[I-D.jimenez-dawn-discovery-landscape].
5.1. Centralized Global Directory
One approach is to deploy a hierarchical, DNS-like or WebFinger-like
global registration system. Sites register their agent metadata with
regional or global directory servers; requesters query the directory
to discover agents.
* *Strengths*: Familiar operational model; query latency is
controllable; well-understood caching and delegation mechanisms.
* *Gaps*: (1) Data sovereignty -- sites must submit agent metadata
to a third-party directory, which many organizations resist for
industrial and enterprise deployments. (2) Trust and governance
-- operating a global root directory requires a governance body
that all participants accept. (3) Single-point-of-failure and
scaling concerns at the root level. (4) Granular per-peer access
control is difficult in a public query model. These gaps are
noted in [I-D.moussa-dawn-gap-analysis] as "no interoperable
federation" across administrative boundaries.
5.2. Link-Local Multicast Discovery (mDNS/DNS-SD)
mDNS/DNS-SD [RFC6762] [RFC6763] provides excellent zero-configuration
discovery within a single broadcast domain.
* *Strengths*: Zero configuration; widely deployed; no
infrastructure required; excellent for local-area networks.
* *Gaps*: (1) Link-local multicast does not cross IP routers, so
discovery is confined to a single LAN. (2) No built-in mechanism
for cross-domain metadata exchange. (3) Large LANs can experience
multicast storm issues without a proxy or gateway. This satisfies
only the "local context" use case [I-D.kay-dawn-use-cases].
5.3. Gap Summary
No single existing approach satisfies all requirements
simultaneously:
Zhang, et al. Expires 23 April 2027 [Page 14]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* Centralized directories fail F-3 (data sovereignty) and F-4
(decentralized federation).
* Link-local multicast fails F-1 (cross-domain discovery).
* Maintaining a separate manual relationship with every other site
does not scale well as a federation grows (NF-5).
* Proactive or multi-hop dissemination can complicate NF-1 (security
against forged advertisements) and F-6 (prompt lifecycle
invalidation) when updates propagate through intermediaries.
The two-layer federated architecture addresses these gaps by: (1)
keeping local discovery zero-configuration and decentralized, without
mandating a specific local protocol; (2) introducing a federation
gateway layer that enforces data sovereignty via export policy; (3)
using DNS publication and on-demand resolution as the cross-domain
baseline, with optional additional functions where needed; and (4)
exchanging only the minimum public discovery information in FMRs to
reduce cross-domain data exposure.
6. Two-Layer Reference Architecture
As illustrated in the following figure, the architecture consists of
the Local Discovery Plane (Layer 1) and the Federation Plane (Layer
2). The FGW connects local discovery information to policy-
controlled DNS publication and request-driven resolution. Existing
DNS publication and resolver functions may be deployed separately.
Agent communication is independent of the discovery metadata path.
Zhang, et al. Expires 23 April 2027 [Page 15]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Domain A Domain B
+----------------------+ +----------------------+
| Agents / AI resources| | Agents / AI resources|
+----------------------+ +----------------------+
| | | |
v v v v
+---------+ +----------+ +---------+ +----------+
|Local | |Infra- | |Local | |Infra- |
|discovery| |assisted | |discovery| |assisted |
|mechanism| |discovery | |mechanism| |discovery |
+---------+ +----------+ +---------+ +----------+
| | | |
+-----+-----+ +-----+-----+
| |
v v
+----------------------+ +----------------------+
| Federation Gateway A | | Federation Gateway B |
| | | |
| Export Policy Engine | | Export Policy Engine |
| | | | | |
| v | | v |
| Construct local FMRs | | Construct local FMRs |
| | | |
| FAD | | FAD |
| On-demand DNS cache | | On-demand DNS cache |
| Validated results | | Validated results |
+----------------------+ +----------------------+
| |
| Layer 1: Local Discovery Plane |
-----------|--------------------------------|------------
| Layer 2: Federation Plane |
v v
+--------------------------------------------------------+
| DNS baseline: authoritative publication |
| Request -> cache miss -> DNS query -> validate -> cache|
| |
| DNS service may be separate from the FGW |
| Remote FADs populate after discovery requests |
+--------------------------------------------------------+
Local discovery mechanisms and Infra-assisted discovery may coexist.
DNS services may be separate from FGWs.
Agent data traffic uses the discovered endpoints.
FGWs need not forward agent data traffic.
Figure 1: Two-Layer Agent Discovery Architecture
Zhang, et al. Expires 23 April 2027 [Page 16]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
6.1. Layer 1: Local Discovery Plane
The Local Discovery Plane identifies AI-agent resources available
within an administrative domain and makes the resulting discovery
information available to the Federation Gateway (FGW), subject to
local policy.
Two complementary mechanisms may be used:
* locally advertised or discovered resources; and
* infrastructure-maintained resources in managed environments.
Both mechanisms can coexist within the same administrative domain.
The Local Discovery Plane does not itself determine what information
is exported to other domains; that is controlled by the FGW and its
export policy.
6.1.1. Locally Advertised or Discovered Resources
In a local network, an agent MAY advertise its lightweight discovery
information after startup through a site-local discovery mechanism,
such as link-local multicast mDNS/DNS-SD (for example, service type
"_agent._tcp.local"). An authorized entity, such as an agent
orchestration system or local gateway, MAY advertise this information
on the agent's behalf.
Other agents and local discovery components can obtain the advertised
information through the same mechanism without requiring a
preconfigured discovery server. Alternative local mechanisms MAY
also be used, including local API registration with an Agent
Directory [I-D.jimenez-agent-directory], edge-platform-driven
discovery such as Kubernetes service discovery, or registration
through ACAP [I-D.zahed-acap].
The architecture therefore supports advertisement by either the agent
or an authorized entity and does not mandate a specific local
discovery protocol.
The FGW MAY collect discovery information learned through these
mechanisms. For example, it MAY listen to local mDNS/DNS-SD
advertisements, subscribe to an Agent Directory, or retrieve
information through ACAP.
Zhang, et al. Expires 23 April 2027 [Page 17]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Locally advertised records carry the minimum public discovery
information, such as Entity Identifier, Entity Type, reachability or
locator information, communication protocol, and services.
Capability information beyond this minimum is not required to be
transmitted through multicast or local broadcast mechanisms.
Local peers may discover and communicate directly within the site.
The FGW is NOT required to forward agent data traffic; its role in
this architecture is limited to discovery information collection and
federation functions.
6.1.2. Infrastructure-Maintained Managed Resources
In managed organizational environments, deployment infrastructure may
already maintain current information about available AI-agent
services or instances, their reachable endpoints, communication
protocols, and associated deployment state.
The FGW MAY obtain this discovery information through the Resource
Publication Hook described in [I-D.dunbar-dawn-infra-assisted-
discovery]. The FGW uses this information to maintain its local
discovery view and, after applying export policy, to construct
discovery information for external publication. Infrastructure-
provided changes can be used to update or remove corresponding
published information as deployment state changes.
Infrastructure-assisted discovery may be provided through a Local
Infrastructure Gateway that performs local collection and
registration functions. The Local Infrastructure Gateway MAY be
implemented separately from, or integrated with, the FGW.
When implemented separately, the Local Infrastructure Gateway
advertises the collected agent discovery information to the FGW
through an appropriate local discovery mechanism and provides updates
as deployment state changes. When integrated, the FGW obtains this
information internally, for example from deployment or registration
records maintained within the managed infrastructure.
The FGW MAY also use requester attachment context supplied through
the Attachment Context Hook described in [I-D.dunbar-dawn-infra-
assisted-discovery] to determine which locally known discovery
domains or reachable agent endpoints are applicable to the requester.
The detailed realization of Attachment Scope is outside the scope of
this document.
Zhang, et al. Expires 23 April 2027 [Page 18]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Infrastructure-maintained discovery is primarily applicable to
organization-level discovery in managed enterprise, cloud, edge, and
communication-network environments. It MAY coexist with locally
advertised discovery within the same administrative domain.
6.1.3. Common Local Discovery Processing
The FGW MAY collect agent discovery information from locally
advertised or discovered resources, infrastructure-maintained managed
resources, or both. For locally advertised or discovered resources,
the FGW MAY obtain information through mechanisms such as mDNS/DNS-
SD, an Agent Directory, ACAP, or another local discovery mechanism.
For infrastructure-maintained managed resources, the FGW receives
information from a separate Local Infrastructure Gateway or, when
integrated with that gateway, obtains the information directly from
records maintained within the managed infrastructure. Discovery
information is updated or withdrawn as the corresponding local
resource or deployment state changes.
The Export Policy Engine on the FGW applies local sharing policy
before information is made available through the Federation Plane.
It determines which locally known resources may be advertised
externally, redacts sensitive fields, and constructs federation-ready
FMRs. Different external peers or publication mechanisms MAY receive
different policy-approved subsets of the locally available discovery
information.
Local peers may communicate directly using the discovered endpoints.
The FGW is NOT required to forward agent application traffic; its
role in this architecture is limited to discovery-information
collection and federation functions.
6.2. Layer 2: Federation Plane
Federation Gateways from separate administrative domains support
cross-domain discovery. DNS publication and resolution form the
baseline: each domain publishes permitted discovery information in
its authoritative zone, and remote FGWs resolve and validate the
records when needed. Publication does not populate remote FADs or
require persistent FGW-to-FGW sessions. Optional functions beyond
the DNS baseline are described in Section 7.
1. Each FGW publishes only the sanitized discovery information
permitted by its export policy. Publication makes records
available in the originating domain's authoritative zone and does
not push them into remote FADs.
Zhang, et al. Expires 23 April 2027 [Page 19]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
2. When an agent's discovery information changes, its FGW or an
authorized DNS controller updates or removes the corresponding
authoritative records. Remote FADs are not proactively
synchronized. A later request obtains information through DNS
resolution when no valid matching cached result remains.
3. An FGW does not maintain a complete local copy of remote agent
metadata by default. Its FAD contains accepted records
previously retrieved in response to discovery requests and
retained within their permitted cache lifetimes.
4. When a client requests cross-domain agent information, the local
FGW first checks for a valid matching FAD cache entry. On a
miss, it uses a validating resolver to query the remote domain,
using a discovery name or a known domain and service type
supplied or obtained by the application. It validates the
response, applies import policy, caches accepted information, and
returns the result. Later requests may reuse a valid entry. An
explicit application-level referral may continue discovery at
another endpoint.
6.3. Federation Operation Model
Federation connects independently administered discovery domains
while each domain controls the information it publishes. The
originating FGW applies export policy and makes permitted FMRs
available; the requesting FGW retrieves or receives eligible records
according to the selected mode and applies validation and import
policy. Publication in one domain does not by itself create a record
in another domain's FAD.
In DNS-based discovery, the originating domain's authoritative DNS
service holds the published agent discovery information. Remote
records are not synchronized into other FGWs' FADs by default. The
FAD serves as a request-driven cache rather than a replica of remote
DNS zones.
6.3.1. Federation Roles and Trust Relationships
The originating domain is responsible for the discovery information
describing its local agents. Its FGW obtains that information
through existing local discovery mechanisms or infrastructure-
assisted discovery and applies export policy before constructing FMRs
for external publication. The federation process operates on these
FMRs independently of the local collection mechanism.
Zhang, et al. Expires 23 April 2027 [Page 20]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
The receiving domain decides which external sources and records it
accepts. Its FGW validates obtained FMRs and applies local import
policy before making them available through its FAD. A federation
relationship therefore combines the originating domain's permission
to disclose information with the receiving domain's permission to
accept and use that information.
Participating domains use existing DNS validation and deployment
security policies. Discovering an endpoint does not authorize access
to restricted metadata or the agent itself.
6.3.2. Federation Workflow
The Federation Plane uses DNS publication and on-demand resolution as
its baseline. Each administrative domain controls its authoritative
discovery information. Publication makes that information available
for queries without pre-populating remote FADs or requiring
persistent FGW-to-FGW sessions. The workflow consists of the
following steps:
1. *Identify the discovery domain:* The requester supplies a
discovery name or obtains the relevant domain through
configuration, an existing discovery mechanism, or a referral.
DNS delegation locates the authoritative source. DNS-SD
[RFC6763] can browse service instances of a known type within a
known domain.
2. *Construct and publish an export view:* The originating FGW
applies its Export Policy Engine to local discovery information
and selects the resources and fields permitted for disclosure.
It publishes the corresponding FMR information through the
domain's authoritative DNS service using the applicable discovery
binding. Public DNS contains only information approved for
public disclosure. The authoritative DNS service may operate
separately from the FGW.
3. *Receive a request and check the FAD:* When a user requests
cross-domain agent information, the local FGW checks its FAD for
a valid matching entry. If that entry satisfies the request and
applicable policy, the FGW may answer directly. Otherwise, the
request triggers DNS resolution for the relevant discovery name
or domain.
Zhang, et al. Expires 23 April 2027 [Page 21]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
4. *Resolve, validate, and cache:* The requesting FGW uses a
validating resolver to obtain the remote discovery records,
applies DNSSEC validation [RFC4035] and local import policy, and
caches accepted information in its FAD. The cache lifetime does
not exceed the remaining DNS TTL or any shorter applicable
validity limit. A recursive resolver may supply a previously
cached response with its remaining TTL.
5. *Return the discovery result:* The FGW returns the accepted
discovery information to the requester. The result provides an
endpoint or an explicit referral for further discovery.
Subsequent requests may reuse a valid matching FAD entry, but
cache hits do not restart its lifetime. Agent access and
communication use the discovered service's existing protocols and
policies.
6. *Maintain authoritative publication:* Changes in agent
availability, endpoints, communication protocols, or export
policy cause the responsible publisher to update or remove the
affected DNS records. An integrated FGW obtains infrastructure
changes from existing registration records; a separate
infrastructure gateway supplies the changes to the FGW.
Publication uses existing mechanisms such as DNS UPDATE [RFC2136]
with the applicable update security [RFC3007], or an authorized
DNS management interface.
Authoritative updates do not immediately invalidate copies already
cached by recursive resolvers or remote FADs. Those copies remain
subject to their established expiration times. An expired FAD entry
is no longer eligible for discovery responses, and a subsequent
request triggers another lookup. DNS TTL limits cached copies rather
than authoritative record lifetime; the publisher reconciles
authoritative records with current deployment state to remove
obsolete information.
Where detailed discovery information requires a different disclosure
policy, DNS may publish only a pointer to an existing access-
controlled discovery endpoint. The requester then uses that
endpoint's established retrieval and security mechanisms. DNSSEC
validation of the pointer does not validate the subsequently
retrieved information or authorize access to the agent.
Zhang, et al. Expires 23 April 2027 [Page 22]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
6.3.3. FAD Scope and Separation from Agent Communication
Each FAD represents the discovery information currently available and
acceptable to its local domain, rather than a complete replica of all
agent information in the federation. Its contents depend on local
storage choices, export and import policies, the selected federation
mechanisms, previous discovery requests, and record validity.
When a cross-domain request has no valid matching cached entry, the
FGW retrieves the relevant information through DNS resolution,
validates it, and caches the accepted result. The cache lifetime
does not exceed the remaining DNS TTL or any shorter applicable
validity limit. Cache hits do not restart that lifetime, and expired
entries are no longer eligible for discovery responses.
Locally originated FMRs may be maintained within the FAD or in a
separate store. A deployment may use additional directory or
distribution sources. Cached entries retain their provenance,
validity, and applicable disclosure conditions.
Once discovery provides a suitable agent endpoint, the requester
communicates with that agent using the indicated communication
protocol and the agent's access-control requirements. The FGWs
support discovery and metadata federation; they are not required to
forward agent business data traffic.
6.3.4. Attachment Context and Endpoint Resolution
The FGW MAY use resource discovery information supplied through the
Resource Publication Hook [I-D.dunbar-dawn-infra-assisted-discovery]
to maintain its local discovery view and construct or update FMRs
after applying export policy.
The FGW MAY use requester attachment context supplied through the
Attachment Context Hook [I-D.dunbar-dawn-infra-assisted-discovery] to
identify applicable discovery domains or reachable agent endpoints.
Context-dependent cached results are reused only for requests
satisfying the same applicability conditions.
6.4. Functional Components
* *Local Agent Discovery Module* (on FGW): Captures local agent
advertisements via passive mDNS listening, Agent Directory
subscription, or other local mechanisms; maintains the site-local
agent table.
Zhang, et al. Expires 23 April 2027 [Page 23]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* *Export Policy Engine* (on FGW): Executes data filtering, privacy
redaction, and per-peer export access policy. This is the core
data sovereignty mechanism.
* *DNS Publication Service* (on FGW): Arranges publication, updates,
and removal of permitted discovery information in the domain's
authoritative DNS service.
* *DNS Resolution and Validation Service* (on FGW): Uses a
validating resolver to obtain remote discovery information,
applies DNSSEC validation and local import policy, and supplies
accepted results to the FAD.
* *Discovery and Referral Response Service* (on FGW): Checks for
valid matching FAD entries, triggers DNS retrieval when necessary,
and returns accepted discovery information or referrals.
* *Metadata Integrity Module* (on FGW): Uses the DNSSEC validation
result for DNS-published discovery information and relies on the
selected application protocol for response protection and endpoint
authentication when DNS points to an application endpoint. It
does not require separate FMR signing or signature verification.
* *FAD Aging Module* (on FGW): Expires records according to their
applicable validity rules. The cache lifetime does not exceed the
remaining DNS TTL or any shorter applicable validity limit. Cache
hits do not restart that lifetime; an expired entry requires fresh
retrieval on a subsequent request.
7. Additional Discovery Functions Beyond the DNS Baseline
DNS publication and on-demand resolution provide the baseline
federation mechanism. The alternatives in this section address
deployment requirements that are not directly provided by this
baseline, including recipient-specific sharing with explicit peer
relationships, proactive or multi-hop dissemination among changing
participants, and application-level queries across an aggregated
collection of resources from multiple domains. These mechanisms
complement DNS where authoritative publication and request-driven
resolution alone do not meet operational or disclosure requirements.
DNS extensions, including DNS Push Notifications [RFC8765], should be
considered before introducing a separate synchronization mechanism.
Zhang, et al. Expires 23 April 2027 [Page 24]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
7.1. Recipient-Specific Sharing
Some deployments may need to distribute different discovery
information to different authorized peers. Where this cannot be
achieved through the deployed DNS publication model, an additional
peer-to-peer mechanism may be used. A structured peering mechanism
with route-policy-like export control is one possible realization.
The specific protocol is outside the scope of this document.
A structured peering mechanism may support incremental exchange of
discovery information, per-peer export policy, and delegation or
referral between discovery domains. The specific synchronization
procedures, peering topology, convergence behavior, and scaling
characteristics are outside the scope of this document.
An originating FGW applies its Export Policy Engine to determine the
discovery information permitted for each authorized peer. The
selected peer-to-peer mechanism conveys that information and any
subsequent updates or withdrawals to the receiving FGW, which
validates the information and applies local import policy before
making accepted records available for discovery. Procedures for peer
establishment, synchronization, recovery, and reconciliation are
properties of the selected mechanism and are outside the scope of
this document.
7.2. Proactive or Multi-Hop Dissemination
Some deployments may require discovery information to be distributed
proactively among participating domains or propagated through
authorized intermediaries. This can be useful when participants
change dynamically, when connectivity is intermittent, or when
request-driven retrieval alone does not meet operational requirements
A dissemination mechanism may propagate policy-approved discovery
information, including updates and withdrawals, across multiple
participating sites. The originating administrative domain remains
responsible for determining which information may be exported.
The specific dissemination protocol, peer-selection algorithm,
propagation procedure, reconciliation method, convergence behavior,
and scaling characteristics are outside the scope of this document.
Zhang, et al. Expires 23 April 2027 [Page 25]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
7.3. Trusted Directory Publication
Some deployments may need to support application-level queries across
discovery information contributed by multiple administrative domains.
DNS and DNS-SD can resolve or browse information within known naming
or discovery scopes, but they do not by themselves provide an
aggregated query service across an authorized collection of resources
from multiple domains.
A trusted directory may provide this function. Participating domains
publish a policy-approved subset of their discovery information to
the directory, which can index the information and answer queries on
behalf of authorized requesters.
Each originating administrative domain remains responsible for
determining what information may be exported to the directory. The
directory does not become authoritative for the underlying resources;
it provides an aggregated discovery view based on information
supplied by participating domains.
The directory protocol, query language, indexing method,
synchronization procedure, and internal storage model are outside the
scope of this document.
7.4. Mixed-Mode Federation
DNS-based discovery and an additional federation mechanism may be
used together.
For example, DNS may be used to discover the relevant federation
endpoint, after which another mechanism provides a function not
available through the DNS baseline, such as recipient-specific
sharing, proactive dissemination, or an aggregated directory query.
Conceptually:
Zhang, et al. Expires 23 April 2027 [Page 26]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Conceptually:Requester / FGW
|
v
DNS-based discovery or bootstrap
|
v
Federation endpoint
|
+--> recipient-specific peer exchange
|
+--> proactive dissemination
|
+--> trusted directory query
The use of an additional mechanism does not replace authoritative DNS
publication where DNS is used. Each administrative domain remains
responsible for controlling the discovery information it exports, and
received information remains subject to validation and local import
policy.
The specific protocol interactions between DNS and the additional
federation mechanism are outside the scope of this document.
8. Security Reference Model
Cross-domain discovery needs integrity, privacy, and appropriate
disclosure boundaries. The following subsections identify relevant
existing mechanisms and deployment responsibilities. They do not
define identities, credentials, authentication exchanges, trust
evaluation, or admission procedures.
8.1. Layer 1: Transport Security
Discovery uses the security mechanisms of the underlying protocols.
DNS over TLS [RFC7858] or DNS over HTTPS [RFC8484] can protect
resolver transport where applicable. Existing HTTPS endpoints use
TLS [RFC8446] and their established application security. Transport
protection does not by itself validate the original DNS data or grant
permission to use an agent.
8.2. Layer 2: FMR Origin Authentication and Integrity
In the DNS baseline, DNSSEC [RFC4035] provides origin authentication
and integrity for signed discovery records under the applicable
validation policy. The FGW does not accept a bogus DNSSEC result as
validated information. This framework does not require a separate
FMR signature envelope or define key management.
Zhang, et al. Expires 23 April 2027 [Page 27]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Where DNS advertises an application endpoint, DNSSEC protects the
pointer, not information subsequently obtained from that endpoint.
The application protocol supplies the appropriate response protection
and endpoint authentication. DANE [RFC6698] may support endpoint
authentication where applicable; its procedures are not redefined
here.
8.3. Layer 3: Access Control on Discovery Information
The Export Policy Engine determines which discovery information may
leave the domain and for which audience. Records explicitly approved
for public disclosure may be published in public DNS. Restricted
information uses an appropriately controlled discovery domain or an
existing access-controlled endpoint located through DNS.
Public DNS caching does not provide arbitrary per-user disclosure
control. Restricted cached results are reused only under compatible
audience and applicability conditions. Identity provisioning and
authorization exchanges use the selected existing mechanisms and
remain outside this document. Agent invocation authorization is
separate from discovery.
8.4. Layer 4: Federation Admission Control
The DNS baseline does not introduce a federation membership or
admission protocol. Operators use existing domain administration,
DNS validation policy, and any access-control arrangements required
by an external endpoint. Establishing agent identities, evaluating
participant trust, and defining admission procedures are outside
scope.
8.5. Denial-of-Service Considerations
Existing DNS and endpoint operational protections apply.
Implementations can bound cache storage and request processing
according to local policy. This framework does not introduce
additional peer keepalives, hold timers, connection admission
messages, or a federation-wide security state machine.
9. Resilience and Operational Considerations
9.1. Network Partition Handling
When a network partition occurs, FGWs on each side lose connectivity
to peers on the other side. The architecture handles this as
follows:
Zhang, et al. Expires 23 April 2027 [Page 28]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
* Previously accepted remote FMRs remain eligible in the local FAD
only until their permitted validity expires. This prevents
transient partitions from causing mass agent disappearance.
* When connectivity is restored, the FGW can resolve the needed
information again through DNS and validate the response before
updating its cache. Any additional mechanism handles its own
update or recovery procedures.
* Operators SHOULD set TTL values that balance prompt stale-record
removal against tolerance for transient partitions. A TTL of 300
seconds (5 minutes) is RECOMMENDED as a default.
9.2. Gateway Failure and Recovery
* An agent or FGW may stop unexpectedly without completing graceful
withdrawal. The architecture distinguishes resource failure,
discovery-component failure, and temporary loss of connectivity;
none of these conditions is inferred solely from a failed
federation session or an unexpired discovery record.
* - *Unexpected agent failure:* Authorized deployment or management
infrastructure detects the change and updates or withdraws the
agent's registration and corresponding published information.
Previously cached DNS results may remain usable until
expiration even after authoritative records are removed. A
requester may therefore encounter an unreachable endpoint
during this interval.
- *FGW failure with independent DNS service:* Published DNS
records may remain retrievable while the originating FGW is
unavailable. Remote FGWs can continue request-driven DNS
retrieval if the authoritative service remains reachable and
the records remain valid. An authorized controller or
replacement FGW is needed to maintain publication accuracy as
agent state changes; DNS availability alone does not establish
that the published agent information is current.
- *Failure of the authoritative DNS service:* A requesting FGW
MAY answer from a valid matching FAD entry within its permitted
lifetime. When no such entry exists, it cannot complete
discovery through that DNS source and reports a temporary
discovery failure or uses another authorized discovery path.
Failure to retrieve a record does not by itself prove that the
agent has stopped.
Zhang, et al. Expires 23 April 2027 [Page 29]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
- *Failure of a separate local infrastructure gateway:* The FGW
may stop receiving deployment updates while previously
collected information remains present. An authorized
monitoring or reconciliation mechanism determines whether that
information can still be treated as current and updates or
withdraws publications when necessary.
- *Recovery:* A recovered or replacement FGW reconciles local
discovery and infrastructure registration state with actual
deployments before restoring publication. It repairs
authoritative DNS records without requiring unsolicited
population of remote FADs; remote FGWs obtain the current
records through subsequent request-driven lookups.
9.3. Graceful Shutdown
Planned shutdown distinguishes withdrawal of an agent from shutdown
of an FGW. Agent withdrawal changes the availability of a
discoverable resource. FGW shutdown affects discovery publication or
federation functions but does not necessarily make the agents
represented by that FGW unavailable.
Before an agent is shut down, the responsible agent management or
deployment infrastructure SHOULD notify the local discovery component
and withdraw or update the agent's registration. A separate local
infrastructure gateway communicates the corresponding change to the
FGW; an integrated FGW obtains it from the infrastructure's
registration records. The FGW then updates the affected FMRs and
their publication state.
The FGW or an authorized publication controller removes or updates
the affected authoritative DNS records. This action prevents
subsequent authoritative queries from returning the withdrawn
information, but it does not immediately invalidate copies already
held by recursive resolvers or remote FADs. Those copies remain
subject to their previously established expiration times. Lowering
the authoritative TTL at shutdown does not shorten the lifetime of
copies already cached.
For a planned agent shutdown, an operator MAY lower the TTL
sufficiently far in advance, allow the previous larger cache
lifetimes to elapse, withdraw the affected records, and allow the
remaining cache lifetimes to expire before stopping the agent. Where
supported, the service MAY drain existing interactions separately
from discovery withdrawal. The detailed shutdown and draining
procedures are deployment-specific.
Zhang, et al. Expires 23 April 2027 [Page 30]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Before an FGW is shut down, its publication and federation
responsibilities SHOULD be transferred to a replacement component or
explicitly discontinued. If agents remain available and their
authoritative DNS records remain accurate and are maintained by
another authorized component, those records need not be withdrawn
solely because the FGW stops. Records advertising an FGW-hosted
discovery endpoint that will become unavailable SHOULD be updated or
removed.
9.4. TTL-Based Stale Record Aging
The FAD maintains expiration information for accepted remote records
according to the mechanism through which they were obtained.
Expiration limits the use of cached or synchronized information; it
does not by itself establish whether an agent is currently reachable.
Remote discovery information is not pre-populated in the FAD by
default. When a cross-domain request has no valid matching cache
entry, the FGW retrieves the relevant DNS records, validates them,
and caches the accepted information. Its expiration time is bounded
by the remaining TTL supplied by the DNS resolution process and any
shorter applicable record or validation lifetime. A response
obtained from an intermediate resolver does not receive a new full
authoritative TTL merely because it is copied into the FAD.
A cache hit does not restart the entry's lifetime. Once an entry
expires, it is no longer eligible to answer discovery requests. A
subsequent request triggers fresh DNS retrieval and validation. The
FGW does not extend an expired entry merely because the authoritative
service is temporarily unreachable. Background cleanup MAY remove
expired entries later, provided that request processing excludes them
immediately upon expiration.
A fresh validated DNS response MAY establish a new cache lifetime
according to its remaining TTL and applicable validity limits. DNS-
based discovery relies on request-driven retrieval rather than
periodic refreshes pushed to remote FADs. Retrieving an unchanged
record does not require a new FMR version.
DNS TTL bounds the lifetime of a cached copy, not the lifetime of a
record in the authoritative zone. If an obsolete record remains
published, subsequent queries can retrieve and cache it again.
Deployments therefore need an authorized mechanism to reconcile
published DNS records with current agent state and to remove obsolete
publications independently of cache aging.
Zhang, et al. Expires 23 April 2027 [Page 31]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Where discovery results depend on Attachment Scope, cached
information is reused only for requests satisfying the same
applicability conditions. A change in attachment context may require
a new lookup before the record's TTL expires.
9.5. Version Evolution and Forward Compatibility
* The information model may evolve through extensions defined by the
applicable discovery specifications. Each mechanism determines
how extensions are encoded, recognized, and handled by recipients.
* DNS-based discovery is the baseline. Where DNS-based mechanisms
and existing DNS extensions do not meet a deployment's discovery
requirements, additional mechanisms may complement them. An FGW
may support multiple discovery mechanisms concurrently, while
applying consistent publication, validation, and caching policies.
Each mechanism follows its own existing specification for
versioning and capability negotiation; this framework defines no
new negotiation procedure.
* Major architectural changes that break compatibility require a new
version of this framework document and SHOULD be discussed in the
DAWN working group before deployment.
9.6. Operational Monitoring
Operators of federation gateways SHOULD monitor the following
metrics:
* Number of active federation peers and peer session uptime.
* FAD size and growth rate, per origin gateway.
* FMR update rate and withdrawal rate.
* Signature verification failure rate (indicator of potential
attacks or misconfiguration).
* FAD cache-hit and cache-miss rates, entry expiration and eviction
rates, and DNS lookup latency. Cache expiration is normal and
does not by itself indicate publication failure.
* DNS resolution failures, DNSSEC validation failures, authoritative
publication update failures, and discrepancies detected during
publication reconciliation.
* Discovery query request rate and authorization failure rate.
Zhang, et al. Expires 23 April 2027 [Page 32]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
10. IANA Considerations
This informational document does not request any new IANA
registrations. The architecture itself does not define protocol
numbers, ports, or message types.
Future protocol specifications that realize this architecture may
request any IANA allocations they require. DNS-based realizations
may request allocation of DNS service names or types (e.g., under the
DNS-SD service-type registry) or new DNS resource record types for
agent discovery. Such requests are the responsibility of the
respective protocol documents.
11. Acknowledgements
The author thanks the DAWN working group participants for their
foundational work on agent discovery terminology, requirements, and
gap analysis. This architecture builds directly on the DAWN problem
statement, requirements, use cases, and the discovery mechanism
survey by Jimenez et al.
12. Informative References
[RFC6762] Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762,
February 2013, <https://www.rfc-editor.org/rfc/rfc6762>.
[RFC6763] Cheshire, S. and M. Krochmal, "DNS-Based Service
Discovery", RFC 6763, February 2013,
<https://www.rfc-editor.org/rfc/rfc6763>.
[RFC7858] Hu, Z., Zhu, L., Heidemann, J., Mankin, A., Wessels, D.,
and P. Hoffman, "Specification for DNS over Transport
Layer Security (TLS)", RFC 7858, May 2016,
<https://www.rfc-editor.org/rfc/rfc7858>.
[RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol
Version 1.3", RFC 8446, August 2018,
<https://www.rfc-editor.org/rfc/rfc8446>.
[RFC8484] Hoffman, P. and P. McManus, "DNS Queries over HTTPS
(DoH)", RFC 8484, October 2018,
<https://www.rfc-editor.org/rfc/rfc8484>.
[RFC9176] Amsuess, C., Ed., Shelby, Z., Koster, M., Bormann, C., and
P. van der Stok, "Constrained RESTful Environments (CoRE)
Resource Directory", RFC 9176, April 2022,
<https://www.rfc-editor.org/rfc/rfc9176>.
Zhang, et al. Expires 23 April 2027 [Page 33]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
[RFC9727] Smith, K., "api-catalog: A Well-Known URI and Link
Relation to Help Discovery of APIs", RFC 9727, June 2025,
<https://www.rfc-editor.org/rfc/rfc9727>.
[RFC4035] Arends, R., Austein, R., Larson, M., Massey, D., and S.
Rose, "Protocol Modifications for the DNS Security
Extensions", RFC 4035, March 2005,
<https://www.rfc-editor.org/rfc/rfc4035>.
[RFC6698] Hoffman, P. and J. Schlyter, "The DNS-Based Authentication
of Named Entities (DANE) Transport Layer Security (TLS)
Protocol: TLSA", RFC 6698, August 2012,
<https://www.rfc-editor.org/rfc/rfc6698>.
[CSA-DAWN-Note]
Cloud Security Alliance, "IETF's Race to Standardize AI
Agent Identity", September 2026,
<https://labs.cloudsecurityalliance.org/research/csa-
research-note-ietf-agent-identity-standards-
20260902-csa/>.
[I-D.akhavain-moussa-dawn-problem-statement]
Akhavain, A., Moussa, H., and D. King, "Problem Statement
for the Discovery of Agents, Workloads, and Named Entities
(DAWN)", Work in Progress, Internet-Draft, draft-akhavain-
moussa-dawn-problem-statement-05, July 2026,
<https://datatracker.ietf.org/doc/html/draft-akhavain-
moussa-dawn-problem-statement-05>.
[I-D.cui-dawn-mdi-model]
Cui, Y., "An Information Model for Minimum Discoverable
Information (MDI)", Work in Progress, Internet-Draft,
draft-cui-dawn-mdi-model-00, July 2026,
<https://datatracker.ietf.org/doc/html/draft-cui-dawn-mdi-
model-00>.
[I-D.cui-dns-native-agent-naming-resolution]
Cui, Y., "DNS-Native AI Agent Naming and Resolution", Work
in Progress, Internet-Draft, draft-cui-dns-native-agent-
naming-resolution-01, March 2026,
<https://datatracker.ietf.org/doc/html/draft-cui-dns-
native-agent-naming-resolution-01>.
Zhang, et al. Expires 23 April 2027 [Page 34]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
[I-D.farrel-dawn-terminology]
Farrel, A., Yao, K., Schott, R., and N. Williams,
"Terminology for the Discovery of Agents, Workloads, and
Named Entities (DAWN)", Work in Progress, Internet-Draft,
draft-farrel-dawn-terminology-02, June 2026,
<https://datatracker.ietf.org/doc/html/draft-farrel-dawn-
terminology-02>.
[I-D.jimenez-agent-directory]
Jimenez, J., "Agent Directory", Work in Progress,
Internet-Draft, draft-jimenez-agent-directory-01, May
2026, <https://datatracker.ietf.org/doc/html/draft-
jimenez-agent-directory-01>.
[I-D.jimenez-dawn-discovery-landscape]
Jimenez, J., Feng, J., Arkko, J., Kuehlewind, M., and R.
Kandoi, "A Survey of AI Agent Discovery Mechanisms", Work
in Progress, Internet-Draft, draft-jimenez-dawn-discovery-
landscape-00, July 2026,
<https://datatracker.ietf.org/doc/html/draft-jimenez-dawn-
discovery-landscape-00>.
[I-D.kay-dawn-use-cases]
King, D., Yao, K., and K. Adler, "Use Cases for the
Discovery of Agents, Workloads, and Named Entities", Work
in Progress, Internet-Draft, draft-kay-dawn-use-cases-00,
June 2026, <https://datatracker.ietf.org/doc/html/draft-
kay-dawn-use-cases-00>.
[I-D.king-dawn-requirements]
King, D. and A. Farrel, "Requirements for the Discovery of
Agents, Workloads, and Named Entities (DAWN)", Work in
Progress, Internet-Draft, draft-king-dawn-requirements-01,
April 2026, <https://datatracker.ietf.org/doc/html/draft-
king-dawn-requirements-01>.
[I-D.mp-agntcy-ads]
AGNTCY, "Agent Directory Service", Work in Progress,
Internet-Draft, draft-mp-agntcy-ads, 2025,
<https://spec.dir.agntcy.org>.
[I-D.moussa-dawn-gap-analysis]
Moussa, H. and A. Akhavain, "Gap Analysis and
Applicability Statement for Discovery Protocols of Agents,
Workloads, and Named Entities (DAWN)", Work in Progress,
Internet-Draft, draft-moussa-dawn-gap-analysis-01, June
2026, <https://datatracker.ietf.org/doc/html/draft-moussa-
dawn-gap-analysis-01>.
Zhang, et al. Expires 23 April 2027 [Page 35]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
[I-D.mozleywilliams-dnsop-dnsaid]
Mozley, J., Williams, N., Sarikaya, B., Schott, R., and J.
Damick, "DNS for AI Discovery", Work in Progress,
Internet-Draft, draft-mozleywilliams-dnsop-dnsaid-02, May
2026, <https://datatracker.ietf.org/doc/html/draft-
mozleywilliams-dnsop-dnsaid-02>.
[I-D.zahed-acap]
Sarker, Z. and T. Reddy, "Agent Capability Advertisement
Protocol (ACAP)", Work in Progress, Internet-Draft, draft-
zahed-acap-00, June 2026,
<https://datatracker.ietf.org/doc/html/draft-zahed-acap-
00>.
[I-D.yao-dawn-agent-discovery-architect]
Yao, J., Geng, G., Chen, M., and H. Li, "DNS-like Agent
Discovery Architecture", Work in Progress, Internet-Draft,
draft-yao-dawn-agent-discovery-architect-00, July 2026,
<https://datatracker.ietf.org/doc/html/draft-yao-dawn-
agent-discovery-architect-00>.
[I-D.jakab-dawn-agent-discovery-mdns]
"Agent Discovery Using mDNS", Work in Progress, Internet-
Draft, draft-jakab-dawn-agent-discovery-mdns-00, July
2026, <https://datatracker.ietf.org/doc/html/draft-jakab-
dawn-agent-discovery-mdns-00>.
[I-D.dunbar-dawn-infra-assisted-discovery]
Dunbar, L., "Infrastructure-Assisted Discovery for AI
Agents", Work in Progress, Internet-Draft, draft-dunbar-
dawn-infra-assisted-discovery-00, 13 October 2026,
<https://datatracker.ietf.org/doc/html/draft-dunbar-dawn-
infra-assisted-discovery-00>.
[RFC2136] Vixie, P., Ed., Thomson, S., Rekhter, Y., and J. Bound,
"Dynamic Updates in the Domain Name System (DNS UPDATE)",
RFC 2136, DOI 10.17487/RFC2136, April 1997,
<https://www.rfc-editor.org/rfc/rfc2136>.
[RFC3007] Wellington, B., "Secure Domain Name System (DNS) Dynamic
Update", RFC 3007, November 2000,
<https://www.rfc-editor.org/rfc/rfc3007>.
Authors' Addresses
Zhang, et al. Expires 23 April 2027 [Page 36]
Internet-Draft A Framework for Agent Discovery in DAWN October 2026
Bin Zhang (editor)
Pengcheng Laboratory
Sibilong Street
Shenzhen
China
Email: zhangb@pcl.ac.cn
Linda Dunbar
Futurewei
Dallas, TX,
United States of America
Email: ldunbar@futurewei.com
Yixuan Yang (editor)
Sun Yat-sen University
Gongchang Street
Shenzhen
Guangdong, 518055
China
Email: yangyx277@mail2.sysu.edu.cn
Zhang, et al. Expires 23 April 2027 [Page 37]