| Internet-Draft | RDAP Reliability Assessment | September 2026 |
| Bertoldi & Romano | Expires 10 March 2027 | [Page] |
This document proposes an extension to the Registration Data Access Protocol (RDAP) that enables the representation and exchange of structured reliability assessment metadata for registrars and domain names. The extension defines a structured assessment envelope through which an RDAP server can expose assessment results produced by a registry, registrar, or third-party assessor in a common, machine-readable format within RDAP responses.¶
The extension standardizes how assessment results are transported and referenced, not how they are computed. Scoring methodologies, thresholds, criteria, and governance frameworks are intentionally left to the operational and policy layer. This document does, however, place requirements on the specification of any scheme whose results are intended for publication through RDAP, because publishing an evaluative judgement about an identified party without safeguards for notification, remediation, and contestation is not a safe practice.¶
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 10 March 2027.¶
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.¶
The domain registration ecosystem relies on registrars as critical intermediaries between domain owners and the global DNS infrastructure. Research [DEEPSEC2025] has identified recurring systemic vulnerabilities in registrar processes, including credential recovery, identity verification, and email authentication configurations, that represent structural risks affecting large numbers of domains and their owners. These issues have in several cases remained unaddressed for extended periods despite responsible disclosure.¶
The Registration Data Access Protocol (RDAP), defined in [RFC7480], [RFC7481], [RFC9082], [RFC9083], and [RFC9224], was designed as the successor to WHOIS and introduces structured JSON responses, authentication and authorization support, and a well-defined extensibility model. These properties make RDAP a suitable foundation for exposing structured security and reliability metadata in a standardized, interoperable way.¶
This document defines an RDAP extension that provides an envelope for carrying assessment results related to the security posture and reliability of registrars and domain names. The extension standardizes the transport and referencing of such results; it does not define scoring methodologies, thresholds, or enforcement mechanisms. The protocol enables representation; the ecosystem decides how to populate and consume the exposed fields.¶
The proposal is intended as complementary to [I-D.loffredo-regext-rdap-verified-contacts], which establishes that contact data has been verified. The present extension adds a parallel layer: structured metadata about the security posture of the registrar and the domain itself.¶
Unlike most registration data fields, this extension carries evaluative metadata whose interpretation and consequences depend on a methodology and an issuer external to the RDAP server. This creates materially different considerations regarding provenance, contestation, staleness, and disclosure than those that apply to registration data held by the responding authority. The remainder of this document is shaped by that difference.¶
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 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
The following terms are used throughout this document:¶
The structured set of fields defined by this extension representing a single result of a security or reliability assessment of a registrar or domain, produced by one issuer under one scheme. A result carries the outcome and points to the methodology; it does not define the methodology itself.¶
The array of assessment results carried by the JSON member defined by this extension.¶
An identifier or URI that denotes the scoring methodology used to produce an assessment result. The scheme defines the semantics of the result, its range, and its criteria. Scheme definitions are maintained externally to this document.¶
A globally unique and stable identifier for the entity that performed the assessment and produced the result.¶
The registrar or domain that is the subject of an assessment result.¶
An RDAP service that publishes assessment results and is not authoritative for the registration data of the assessed party.¶
The RDAP extension string registered in the IANA RDAP Extensions Registry that identifies this extension.¶
Research [DEEPSEC2025] has identified recurring systemic vulnerabilities in domain registrar processes that cannot be addressed by conventional technical security controls. These vulnerabilities arise from weaknesses in the interface between digital systems and human processes, and include deficiencies in credential recovery, identity verification, and email authentication configurations.¶
A notable characteristic of these vulnerabilities is their persistence: in several documented cases, exploitable issues remained unresolved for extended periods following responsible disclosure, reflecting diffuse accountability and the absence of structured incentives for timely remediation.¶
These findings suggest that while technical standards such as SPF [RFC7208], DKIM [RFC6376], and DMARC [RFC9989] are sound, their adoption and correct configuration cannot be reliably ensured without structured, machine-readable signaling mechanisms. The full technical details of the underlying research are documented in [DEEPSEC2025].¶
A published assessment is not only exposure for a party that is managed poorly. It is equally a means by which a well-run operator can demonstrate its posture in a form that a third party can verify and consume automatically. The absence of such a mechanism today means that operators who invest in security have no interoperable way to make that investment legible, and that operators who do not have no structured incentive to change. Both effects are addressed by the same mechanism.¶
Certification and operational assessment address different needs and are complementary rather than alternative.¶
A certification such as ISO/IEC 27001 [ISO27001] or ISO/IEC 27701 [ISO27701] attests conformity against a defined scope at a point in time, through a formal audit process with its own accreditation chain. An operational assessment can instead represent finer-grained technical and procedural properties that change over time, such as how credential recovery is handled or how email authentication is configured on operator-controlled domains.¶
The intent of this document is not to displace certification, but to complement it with a representation capable of expressing a more frequent and more current view of security posture than a periodic certification cycle can provide. Nothing in this document should be read as suggesting that an assessment result is equivalent to, or a substitute for, a certification.¶
Structured representation of assessment metadata at the protocol level offers several complementary benefits relative to existing operational approaches.¶
First, machine-readable assessment data, where its publication is appropriate, creates reputational and commercial incentives for registrars and domain owners to adopt and maintain security best practices.¶
Second, a standardized RDAP extension enables any registry, registrar, browser, or security tool to consume assessment metadata through a common interface, eliminating dependency on proprietary or registry-specific systems. The alternative, in practice, is each assessor publishing its own HTTP interface with its own object model and query syntax.¶
Third, protocol-level representation does not replace operational scoring or enforcement programs. Rather, it provides a standardized channel through which the outputs of such programs can be expressed and consumed by the broader ecosystem.¶
The following principles guide the design of this extension.¶
This document defines the structured assessment envelope and extension points. The governance of who computes results, the specific methodology, any thresholds applied, and any enforcement actions based on results belong to the operational and policy layer and are outside the scope of this document. The requirements in Section 8 are the single deliberate exception, and are requirements on a scheme specification rather than on an RDAP server.¶
The extension standardizes how assessment results are transported and referenced within RDAP. It does not standardize how assessments are performed, what criteria are evaluated, or what thresholds apply.¶
This document defines no new query, no new path segment, and no new query parameter. It extends responses only.¶
The extension is designed to accommodate additional fields without breaking backward compatibility, consistent with RDAP's existing extensibility model.¶
The extension is not tied to any specific registry, registrar, or policy framework. Any conformant RDAP server may implement it independently.¶
The extension is designed to coexist with and complement [I-D.loffredo-regext-rdap-verified-contacts], rather than replace or duplicate it.¶
The -01 revision of this document described no deployment model. Discussion at IETF 126 made clear that this was the principal gap, and this section responds to it.¶
Three configurations are anticipated.¶
An RDAP service operated by the assessor, distinct from any authority for the registration data of the assessed party. This is the primary configuration anticipated by this document.¶
A registry, registrar, or other RDAP server that publishes an assessment result produced by a third party. This configuration introduces an intermediary and is discussed in Section 11.1.¶
An operator publishing an assessment of itself. This remains possible, but the identity of the issuer conveyed in "scoreIssuer" allows a consumer to recognise the configuration and to weigh the result accordingly.¶
An assessment service MUST NOT be treated as authoritative for the registration data of the assessed object.¶
An assessment service SHOULD return only the members required to bind the response to the assessed object, together with the member defined by this extension. For a domain object this is "objectClassName" and "ldhName"; for an entity object it is "objectClassName", "handle", and, where applicable, "publicIds" and "roles". The "rdapConformance" member is required by [RFC9083].¶
An assessment service SHOULD NOT mirror the registration data returned by the authoritative server and enrich it with assessment metadata. Such a pattern would place the assessment service in the position of restating registration data for which it has no authority and no update path, and it is not a requirement of this document.¶
The bootstrap mechanism defined in [RFC9224] determines the authoritative server for registration data. An assessment service is not authoritative for registration data, and this document therefore does not place assessment services in that bootstrap and does not define a parallel one.¶
Where a consumer has established trust in an issuer out of band, the endpoint of that issuer's assessment service is provisioned as part of the same trust configuration, and no separate discovery mechanism is required. A consumer that has not established such trust has no basis on which to act on a result in any case, as discussed in Section 11.1.¶
An assessment result is only meaningful if the object it is attached to can be bound unambiguously to the assessed party.¶
For domain objects, "ldhName" provides the standard lookup key and the query
form /domain/<ldhName> defined in [RFC9082] works unchanged. No additional
mechanism is required.¶
For entity objects the situation is less straightforward. [RFC9082] defines
entity lookup only as /entity/<handle>, and the syntax of a handle is
specific to the registration provider. The "publicIds" member is a response
structure, not a lookup key.¶
An assessment service SHOULD use its own service-local handle for
/entity/<handle> lookups, and SHOULD include a "publicIds" member binding
that local object to the assessed registrar, for example through the
applicable IANA Registrar ID.¶
This document deliberately does not define a lookup by IANA Registrar ID. Doing so would convert this extension from a response extension into a query extension, with the consequences described in Section 2.3 of [I-D.ietf-regext-rdap-extensions], and the authors do not consider it warranted at this stage.¶
Two assessors will use different local handles for the same registrar, and "publicIds" supplies the binding only after the object has been retrieved. A client that holds an IANA Registrar ID and wishes to locate the corresponding object at a known assessment service therefore has no in-band means of constructing the query.¶
Object tagging [RFC8521] can bootstrap an entity query where a client already holds a tagged handle whose suffix identifies the RDAP service provider. One conceivable profile would use the IANA Registrar ID as the local portion of the handle and the assessor's registered provider tag as its suffix, making the full handle derivable once the assessor is known out of band. The authors raise this as an open question rather than proposing it: it is not clear whether this is a sound use of [RFC8521] given that the assessor would be authoritative for the assessment representation but not for the underlying registration data. Feedback is explicitly sought.¶
The binding for registrars operating under ccTLD registries, and for registrars with no IANA Registrar ID, is likewise unresolved. The "publicIds" member accommodates other identifier types, but this document does not prescribe one.¶
This extension is identified by the string "reliabilityAssessment", to be registered in the IANA RDAP Extensions Registry (see Section 13). RDAP responses that include this extension MUST include the extension identifier in the "rdapConformance" array.¶
Consistent with Section 2.2 of [I-D.ietf-regext-rdap-extensions], the extension identifier contains no underscore character. The JSON member defined by this extension is prefixed with the extension identifier, avoiding the bare extension identifier pattern discussed in that section.¶
Section 5.4 of [I-D.ietf-regext-rdap-extensions] states that RDAP extension identifiers are opaque and that extensions without an explicitly defined versioning scheme are opaquely versioned. This extension defines no such scheme, and is therefore opaquely versioned.¶
Section 6.1 of [I-D.ietf-regext-rdap-versioning] provides that all extensions support Opaque Versioning by default and that an extension document should identify the versioning types it supports together with the associated extension version identifier. Accordingly:¶
Opaque Versioning, as described in Section 5.1 of [I-D.ietf-regext-rdap-versioning].¶
reliabilityAssessment¶
This document does not use Point Versioning. It defines a single version of a single extension, and the machinery for negotiating concurrent versions is not warranted at this stage. Nothing in this choice precludes the later adoption of Point Versioning: an extension supporting another versioning type may still be referred to using Opaque Versioning.¶
Implementation of [I-D.ietf-regext-rdap-versioning] is not required in order to implement this extension, and this document places no such requirement on clients or servers.¶
Consistent with Section 5.4 of [I-D.ietf-regext-rdap-extensions], any change to this extension that is not backward compatible will require the registration of a new extension identifier and a new specification.¶
This extension groups all its fields under a single top-level JSON member named "reliabilityAssessment_results". The member name is prefixed with the extension identifier, while the members of the objects it contains (scoreScheme, scoreValue, and so on) are not individually prefixed. This follows the guidance in Section 2.4.2 of [I-D.ietf-regext-rdap-extensions], which states that prefixing is not required for children of a prefixed JSON object defined by an RDAP extension.¶
The "reliabilityAssessment_results" member contains a JSON array of assessment result objects. This member MAY appear within entity objects (objectClassName: "entity") and domain objects (objectClassName: "domain"). An entity is considered to represent a registrar when its "roles" array includes the value "registrar" as defined in Section 10.2.4 of [RFC9083].¶
If the member is present, the array MUST contain at least one element. The order of elements is not significant, and clients MUST NOT infer precedence, authority, or recency from position. Clients MUST NOT assume that the array contains exactly one element.¶
A server SHOULD publish at most one current result for any given combination of "scoreScheme" and "scoreIssuer".¶
Each result carries the outcome of an external assessment and points to the methodology used. It does not define the methodology, the criteria, or the thresholds.¶
The members of each assessment result object are described in the following table.¶
| Field | Type | Description |
|---|---|---|
| scoreScheme | string (URI or identifier) | REQUIRED. Identifies the methodology used to produce the result. When expressed as a URI, it MUST conform to [RFC3986]. The scheme defines the semantics, range, and criteria of the result. Scheme definitions are maintained externally. |
| scoreIssuer | string (URI) | REQUIRED. Identifies the entity that performed the assessment and issued the result. It MUST be a URI conforming to [RFC3986], globally unique and stable, and SHOULD be under the control of the issuer. |
| scoreValue | number | OPTIONAL. The non-negative numeric result of the assessment, as defined by the scoreScheme. If scoreMaxValue is present, scoreValue SHOULD NOT exceed scoreMaxValue unless the scoreScheme explicitly defines otherwise. |
| scoreMaxValue | number | OPTIONAL. The non-negative maximum possible value under the scoreScheme. Together with scoreValue, helps consumers interpret the numeric result. |
| scoreDate | string (date-time) | OPTIONAL. The date and time at which the assessment was performed, in the format defined in [RFC3339]. |
| validUntil | string (date-time) | OPTIONAL. The date and time after which the issuer no longer asserts the result as current, in the format defined in [RFC3339]. Its expiry does not render the result invalid; it indicates that the issuer does not vouch for its currency beyond that point. |
| assessmentId | string | OPTIONAL. An identifier for this specific result, unique within the scope of the combination of scoreScheme and scoreIssuer. It allows a specific result to be referenced, in particular in a contestation or correction process. |
| status | string | OPTIONAL. The lifecycle state of the result as asserted by the issuer. Permitted values are defined in Section 7.6. |
| evidenceUri | string (URI) | OPTIONAL. A URI pointing to supporting documentation, a detailed report, or the full assessment record maintained by the scoreIssuer. |
A member for which no value is available is omitted rather than being included with a null value.¶
If "scoreValue" is present, "scoreDate" SHOULD also be present, so that the result can be situated in time. If "scoreMaxValue" is present, "scoreValue" SHOULD also be present.¶
Implementers MUST NOT infer normative meaning from the illustrative field values used in the examples in this document or in Appendix C.¶
The "status" member, when present, takes one of the following values:¶
The issuer asserts the result as current.¶
The result is the subject of a review or contestation process, and the issuer indicates that it may change. Consumers SHOULD treat a result in this state with additional caution.¶
The issuer no longer asserts the result. A withdrawn result MAY continue to be published so that consumers holding a cached copy can observe the withdrawal.¶
The absence of the "status" member means that the issuer does not convey a lifecycle state. Consumers MUST NOT assume a value of "active" in that case.¶
This document does not define a supersession mechanism. A server that publishes only current results does not require one, and the publication of assessment history is outside the scope of this document.¶
If all semantics of a result are defined outside this protocol, the identity of the scheme must be stable for those semantics to be resolvable.¶
The semantics associated with a given "scoreScheme" identifier MUST NOT change in a manner that is not backward compatible. A revision of a scheme that is not backward compatible MUST be assigned a new scheme identifier.¶
The absence of the "reliabilityAssessment_results" member MUST NOT be interpreted as evidence that the object has not been assessed, nor as a negative assessment of the object.¶
The member may be absent because the object has not been assessed, because it has been assessed but the result is not published, because publication is withheld under server policy, because the data is subject to access control and the client is not authorized, or because the server does not implement this extension.¶
This document deliberately defines no signal distinguishing "not assessed" from "assessed but not disclosed". Such a signal would defeat the purpose of withholding a result, since it would itself indicate that there is something to withhold, and would reintroduce the disclosure risk described in Section 11.¶
The following example illustrates how the assessment envelope MAY appear within an entity object representing a registrar.¶
{
"rdapConformance": [
"rdap_level_0",
"reliabilityAssessment"
],
"objectClassName": "entity",
"handle": "REGISTRAR-EXAMPLE",
"roles": ["registrar"],
"publicIds": [
{
"type": "IANA Registrar ID",
"identifier": "9999"
}
],
"reliabilityAssessment_results": [
{
"scoreScheme": "https://assessor.example/schemes/registrar/v1",
"scoreIssuer": "https://assessor.example",
"scoreValue": 8,
"scoreMaxValue": 10,
"scoreDate": "2026-01-15T10:30:00Z",
"validUntil": "2026-07-15T00:00:00Z",
"assessmentId": "a-2026-01-15-0042",
"status": "active",
"evidenceUri": "https://assessor.example/rep/a-2026-01-15-0042"
}
]
}
¶
The following example illustrates how the assessment envelope MAY appear within a domain object.¶
{
"rdapConformance": [
"rdap_level_0",
"reliabilityAssessment"
],
"objectClassName": "domain",
"handle": "example.com",
"ldhName": "example.com",
"reliabilityAssessment_results": [
{
"scoreScheme": "https://assessor.example/schemes/domain/v2",
"scoreIssuer": "https://assessor.example",
"scoreValue": 7,
"scoreMaxValue": 10,
"scoreDate": "2026-02-01T08:00:00Z",
"status": "active"
}
]
}
¶
This document does not define assessment methodologies. It does, however, place requirements on the specification of a scheme whose results are intended for publication through RDAP, because publishing an evaluative judgement about an identified party without safeguards is not a safe practice, and because this document creates the channel through which such publication would occur.¶
A specification defining a "scoreScheme" whose results are intended for publication through RDAP:¶
MUST document a disclosure threat model, describing what an adversary could learn from the published result and how that risk is weighed against the benefit of publication;¶
MUST define a notification to the assessed party prior to publication of a result;¶
MUST define a period, following that notification and preceding publication, during which the assessed party may remediate the findings or contest the result;¶
SHOULD define whether a reassessment performed within that period replaces the pending result, so that a party that remediates promptly is not exposed by a result that no longer reflects its posture;¶
MUST describe how data minimization is applied, stating whether aggregation, coarse granularity, access control, or delayed publication are used, and why the chosen granularity is appropriate;¶
MUST state whether its results are intended for publication in respect of domain objects and, if so, MUST justify that choice specifically against the targeting risk described in Section 11.¶
This document deliberately does not prescribe the duration of the remediation and contestation period, the governance of the process, the body responsible for adjudicating a contestation, or the consequences of a successful one. Those belong to the scheme and its operational model.¶
There is a tension that this document cannot resolve on its own: the more actionable a signal is for defenders, the more useful it is to attackers. The requirements above are intended to force each scheme to confront that tension explicitly rather than to leave it unexamined.¶
The [I-D.loffredo-regext-rdap-verified-contacts] extension establishes that contact data associated with a domain or registrar has been verified. The present extension adds a complementary and orthogonal layer: structured metadata about the security posture of the registrar and the domain itself.¶
The two extensions answer different questions. The verified-contacts extension addresses which contact fields have been verified and how the verification was performed. The reliability assessment extension addresses what the assessed security posture of the entity managing that domain is. Both are expressible within the RDAP framework and are designed to coexist within the same RDAP response.¶
PIR's Abuse Intervention Program (AIP) and Quality Performance Index (QPI) [PIR-AIP] represent an example of a registry-specific operational assessment program. This document does not standardize such a program; it defines a generic RDAP representation that could carry outputs produced by programs such as PIR/QPI or other assessment frameworks. The authors note that PIR has not, to their knowledge, undertaken RDAP work in connection with these programs.¶
An operational scoring system such as PIR/QPI could, in principle, expose its outputs through the envelope defined in this document, enabling broader interoperability without modifying its internal methodology.¶
This extension is designed to be fully conformant with the RDAP core specifications [RFC7480] [RFC7481] [RFC9082] [RFC9083] [RFC9224]. It uses the JSON response format defined in [RFC9083], the extensibility model provided in [RFC7480], and the security framework of [RFC7481]. Query formats follow [RFC9082] and are not extended by this document.¶
The authors are aware of no implementation of this extension at the time of writing, and state this explicitly rather than leave it to be inferred.¶
The configurations in which the extension is expected to be useful are those described in Section 5: an assessment service operated by a research group, a security vendor, or a sector body; a registry or registrar republishing a third-party result; and, more marginally, self-assessment. The consumers anticipated are RDAP clients operated by registries, registrars, enterprise security teams, and researchers.¶
The authors would welcome statements of interest from any party that operates or plans to operate an assessment program, implements RDAP clients, or would consume this kind of data. A statement that a party would not use the extension, together with the reason, is equally valuable.¶
The fields defined in this extension are informational. They do not constitute enforcement mechanisms and MUST NOT be interpreted as authoritative security certifications.¶
A published result indicating a weak security posture is itself a signal that may be used to identify and prioritise targets. The risk increases where the result is accompanied by the methodology, the assessment date, evidence, or the specific dimensions that contributed to it. The risk is materially greater for domain objects than for registrar objects, since a domain-level result may identify an individual poorly defended target, whereas a registrar-level result concerns an accountable operator of scale. Section 8 requires each scheme to address this explicitly.¶
The Network and Information Security Directive [NIS2] does not explicitly prohibit the publication of a security or reliability result concerning a domain name or a registrar. Article 21(1) does, however, require essential and important entities to take appropriate and proportionate technical, operational, and organisational measures to manage the risks posed to the security of network and information systems, and Article 21(2) enumerates areas that such measures should address, including risk analysis and information system security and vulnerability handling and disclosure. Systematic and unauthenticated public disclosure of assessment results may be difficult to reconcile with those obligations where the disclosure substantially increases the exposure of the assessed entity. Implementers operating under that framework, or an equivalent one, should assess the publication of results accordingly.¶
An assessment result may be read as a rating of the assessed party, with consequent reputational, commercial, and legal exposure, irrespective of the intent of the scheme that produced it. Publication is therefore not a neutral act, and Section 8 places requirements on schemes accordingly.¶
A result may change quickly. A value published today may be higher or lower tomorrow, and a consumer acting on a result may be acting on a state of affairs that no longer holds. The "validUntil" and "status" members provide partial mitigation but do not eliminate this.¶
Publication of assessment metadata is not mandatory. A server MAY expose this extension only to authenticated and authorized clients, and MAY apply different policies to registrar and domain objects. Nothing in this document requires that assessment metadata be publicly accessible.¶
See Section 7.8. The absence of assessment metadata is not evidence of anything, and clients MUST NOT treat it as such.¶
This document does not define an interface for creating or modifying assessment metadata. Any administrative or provisioning interface used for that purpose MUST authenticate and authorize the submitting entity and MUST provide appropriate integrity and confidentiality protection. Without authentication of the issuing entity, assessment metadata is susceptible to manipulation.¶
Any visible scoring system creates incentives for gaming. The governance framework, which is outside the scope of this document, is expected to address verification, audit, and revocation mechanisms.¶
Consumers of assessment metadata MUST NOT treat results as equivalent to security certifications. A high value does not guarantee the absence of vulnerabilities. Consumers SHOULD treat results as one signal among many and SHOULD NOT make high-stakes trust decisions based solely on RDAP assessment fields.¶
Results reflect the state at the time indicated by "scoreDate". Consumers SHOULD check freshness and SHOULD treat results that have not been updated recently, or that are past their "validUntil", with appropriate skepticism.¶
The resource identified by "evidenceUri" is maintained by the issuer and is outside the control of the RDAP server. Its content may change over time, may be subject to access control, and may become unavailable. Consumers SHOULD NOT assume that the evidence resource is publicly accessible or immutable.¶
Three distinct properties are frequently conflated when reasoning about assessment metadata, and this document separates them explicitly.¶
To whom a result is attributed. The "scoreIssuer" member conveys this, and the requirement that it be a stable globally unique URI ensures that the attribution is unambiguous.¶
Whether the consumer has grounds to rely on the named issuer. This is established out of band and is not conveyed by the protocol.¶
Whether the assertion in fact originated with the named issuer and was relayed unaltered. This is not established by the protocol.¶
The "scoreIssuer" member records to whom a result is attributed. It does not demonstrate that the named issuer produced it, nor that the publishing server relayed it unaltered.¶
Where the assessor operates its own assessment service, trust in the endpoint and trust in the issuer can reasonably coincide, because the consumer's out of band trust configuration names both. Where a registry, registrar, or other server republishes an assessor's result, the consumer must additionally trust the publisher as an intermediary, and this document provides no means by which a consumer can distinguish the two cases from the response alone.¶
This is a known limitation of the present revision, stated here rather than left implicit. Consumers SHOULD establish out of band trust in both the issuer and the publishing server before acting on assessment data. The definition of a signed assessment assertion, which would permit authenticity to be established in band and would allow republication without additional trust in the intermediary, is left to future work and is not specified here.¶
The fields defined in this extension describe security posture at the registrar and domain level and are not intended to expose personal data. They are designed to be compatible with the access control framework defined in [RFC7481].¶
Implementers MUST ensure that the presence or absence of assessment metadata does not inadvertently reveal information about the identity of natural persons. The separation between assessment metadata, which is designed to be accessible where server policy and applicable access controls permit, and contact data, which is subject to access control per [RFC7481], MUST be maintained in conformant implementations.¶
Assessment results published in respect of domain objects warrant particular care. A domain may be registered by a natural person, and a result describing the security posture of that domain may in some circumstances permit inferences about that person, independently of whether contact data is disclosed. The requirement in Section 8 that a scheme justify domain-level publication specifically is intended in part to address this.¶
This extension is intended to be compatible with applicable data protection regulations, including the General Data Protection Regulation [GDPR] and equivalent frameworks.¶
The "evidenceUri" field, if populated, SHOULD NOT point to resources that expose personal data or operationally sensitive details beyond what is necessary for the consumer to understand the result.¶
This document requests the registration of the following entry in the RDAP Extensions Registry, using the template defined in Section 8.1 of [RFC7480]:¶
reliabilityAssessment¶
Any¶
this document¶
Alessandro Bertoldi alessandro@bertoldicybersecurity.com¶
COMMON¶
Registration in this registry is subject to a Specification Required policy. Section 7.1.2 of [I-D.ietf-regext-rdap-extensions] additionally requires that an extension be documented in a stable, non-changing, and readily available reference that is not denoted as a work in progress. Registration is therefore expected to occur upon publication of this document as an RFC, and nothing in this document should be read as reserving the identifier in the interim.¶
This document requests the registration of a new value type in the RDAP JSON Values Registry, in accordance with Section 7.2 of [I-D.ietf-regext-rdap-extensions].¶
assessment status¶
the "status" member of an assessment result object, as defined in Section 7.6¶
The following values are requested for that type: "active", "under review", "withdrawn". Each is to be registered with this document as its published specification.¶
This appendix summarizes the changes made in this revision. It is intended to be removed prior to publication.¶
The extension identifier has been changed from "reliabilityScoring" to "reliabilityAssessment", aligning it with the title of the document and avoiding the reading of the extension as a rating mechanism. The document name is unchanged in order to preserve continuity in the datatracker.¶
The JSON member has been renamed from "reliabilityScoring_assessment" to "reliabilityAssessment_results" and now carries an array of result objects rather than a single object. Results from more than one issuer or scheme on the same object are a natural case, and moving from object to array in a later revision would itself have been a breaking change.¶
"scoreScheme" and "scoreIssuer" are now REQUIRED in each result object.¶
"scoreIssuer" MUST now be a URI. It was previously an opaque string whose format was defined by the scheme.¶
These changes are made together in a single revision, and at a point at which the authors are aware of no implementations.¶
Section 5, describing deployment configurations, minimal response, endpoint discovery, and the rejected referral mechanism.¶
Section 6, describing binding for domains and registrars and stating the open questions concerning [RFC8521] and registrars without an IANA Registrar ID.¶
Section 8, placing requirements on the specification of any scheme whose results are intended for publication through RDAP, including notification, a remediation and contestation period, and a disclosure threat model.¶
Section 11.1, separating identity, trust, and authenticity, and stating the limitations of the present revision.¶
The "validUntil", "assessmentId", and "status" members, and Section 7.6 defining permitted status values.¶
Section 10, stating that no implementations are known.¶
An "Extension Version Identifier" section declaring Opaque Versioning, following the guidance in Section 6.1 of [I-D.ietf-regext-rdap-versioning]. The -01 was silent on versioning.¶
Sections on scheme stability and on the absence of the member, the latter promoted from an appendix in -01 and expanded.¶
Expanded Security Considerations covering reconnaissance and exposure, risk-based regulatory obligations, perception as a rating, volatility, access control and server policy, and inference from absence.¶
A registration request for the RDAP JSON Values Registry.¶
The IANA Considerations section now uses the complete template from Section 8.1 of [RFC7480], including the "Intended usage" field omitted in -01.¶
Section references to [I-D.ietf-regext-rdap-extensions] have been updated to the numbering of the -15 revision.¶
[I-D.ietf-regext-rdap-extensions] has been moved from an informative to a normative reference. This document depends on it for identifier syntax, for the treatment of child JSON values, and for the procedures governing registration of a new value type, and cannot be implemented conformantly without it.¶
The reference to ISO/IEC 27701 has been corrected to the 2025 edition, which carries a revised title.¶
The authors considered and did not include a representation of confidence or uncertainty, a representation of categorical rather than numeric outcomes, and a supersession mechanism. In each case no requirement from an implementer or prospective consumer has been identified. The restructuring of the member into an array of result objects means that each could be added later as an additional member without a breaking change.¶
The authors thank Pawel Kowalik, Mario Loffredo, Maurizio Martinelli, James Gould, and James Galvin for their early engagement and feedback during IETF 125.¶
The authors further thank the participants in the REGEXT session at IETF 126, in particular Gavin Brown, James Galvin, James Gould, Jody Kolker, Mario Loffredo, Jim Reid, Werner Staub, Richard Wilhelm, and Maarten Wullink, whose questions and objections shaped this revision. The deployment model in Section 5, the treatment of contestation, and the expanded disclosure considerations are direct responses to points raised there and in subsequent discussion on the mailing list.¶
The examples in this appendix are provided for illustrative purposes only. Values, scheme identifiers, and issuer identifiers are fictional.¶
{
"rdapConformance": [
"rdap_level_0",
"reliabilityAssessment"
],
"objectClassName": "domain",
"handle": "example.com",
"ldhName": "example.com",
"reliabilityAssessment_results": [
{
"scoreScheme": "https://assessor.example/schemes/domain/v2",
"scoreIssuer": "https://assessor.example",
"scoreValue": 9,
"scoreMaxValue": 10,
"scoreDate": "2026-01-01T00:00:00Z",
"validUntil": "2026-04-01T00:00:00Z",
"assessmentId": "d-2026-01-01-7781",
"status": "active",
"evidenceUri": "https://assessor.example/rep/d-2026-01-01-7781"
}
],
"events": [
{
"eventAction": "registration",
"eventDate": "2024-01-01T00:00:00Z"
}
]
}
¶
This example illustrates the case for which the array representation is defined: two issuers, applying different schemes, have published results in respect of the same registrar. The second result is under review following a contestation by the assessed party.¶
{
"rdapConformance": [
"rdap_level_0",
"reliabilityAssessment"
],
"objectClassName": "entity",
"handle": "REGISTRAR-EXAMPLE",
"roles": ["registrar"],
"publicIds": [
{
"type": "IANA Registrar ID",
"identifier": "9999"
}
],
"reliabilityAssessment_results": [
{
"scoreScheme": "https://assessor.example/schemes/registrar/v1",
"scoreIssuer": "https://assessor.example",
"scoreValue": 10,
"scoreMaxValue": 10,
"scoreDate": "2026-01-01T00:00:00Z",
"assessmentId": "r-2026-01-01-0003",
"status": "active"
},
{
"scoreScheme": "https://observatory.example/scheme/2026",
"scoreIssuer": "https://observatory.example",
"scoreValue": 62,
"scoreMaxValue": 100,
"scoreDate": "2026-02-10T00:00:00Z",
"assessmentId": "2026-02-10-REG-9999",
"status": "under review"
}
]
}
¶
When no assessment metadata is conveyed, the "reliabilityAssessment_results" member is omitted. Its absence MUST NOT be interpreted as evidence that the object has not been assessed, nor as a negative assessment. See Section 7.8.¶
{
"rdapConformance": [
"rdap_level_0"
],
"objectClassName": "domain",
"handle": "plain.example",
"ldhName": "plain.example",
"events": [
{
"eventAction": "registration",
"eventDate": "2025-06-01T00:00:00Z"
}
]
}
¶
This appendix describes possible dimensions that an operational assessment program might evaluate when producing results to be carried by the envelope defined in this document. This content is entirely non-normative. Nothing in this appendix constrains implementers or defines mandatory evaluation criteria, and nothing in it should be read as a recommendation that any of these dimensions be published in the form described.¶
The intent is to demonstrate that the envelope model is expressive enough to carry results from real-world assessment programs, including those derived from existing empirical research on registrar security posture [DEEPSEC2025].¶
An operational program might evaluate registrars across dimensions such as: the strength of customer identity verification procedures; the adoption and enforcement of multi-factor authentication for customer-facing and internal systems; the possession of recognized information security certifications such as ISO/IEC 27001 [ISO27001] or ISO/IEC 27701 [ISO27701]; the correctness of email authentication configurations including SPF [RFC7208], DKIM [RFC6376], and DMARC [RFC9989] on registrar-operated domains; the existence of documented security policies; and the regularity of cybersecurity training programs for staff.¶
An operational program might evaluate individual domains across dimensions such as: the strength of owner identity verification at registration or renewal; the level of TLS certificate validation used; the correctness of SPF, DKIM, and DMARC configurations; the implementation of DNSSEC [RFC4033]; and the absence of the domain from monitored abuse blocklists over a defined observation period.¶
The considerations in Section 11 apply with particular force to dimensions of this kind, since several of them describe exploitable weaknesses directly.¶
Existing programs such as PIR's Quality Performance Index [PIR-AIP] focus on observable abuse outcomes and operational metrics within a single registry context. The assessment dimensions described above focus on structural security posture. The envelope defined in this document is agnostic to the methodology and could carry results from either type of program.¶