Internet-Draft Credential to OIDC Bridge August 2026
Svensson, et al. Expires 8 February 2027 [Page]
Workgroup:
TBD
Internet-Draft:
draft-svensson-credential-oidc-bridge-01
Published:
Intended Status:
Informational
Expires:
Authors:
M. Svensson
SUNET
L. Johansson
Siros Foundation
J. Rangsmo
Siros Foundation

Credential Presentation to OIDC Claims Bridge

Abstract

This document defines a mechanism for conveying digital credential claims via OpenID Connect (OIDC). It specifies how an OpenID Provider (OP) that collects credentials from a wallet can expose those claims to Relying Parties as standard OIDC claims, enabling existing OIDC deployments to consume digital credentials without implementing any wallet-facing presentation protocol.

Discussion Venues

This note is to be removed before publishing as an RFC.

Source for this draft and an issue tracker can be found at https://github.com/masv3971/rfc_credential_oidc_bridge.

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 8 February 2027.

Table of Contents

1. Introduction

Digital credential wallets can present verified claims to verifiers using protocols such as OpenID for Verifiable Presentations (OpenID4VP) [OpenID4VP] or DIDComm. However, many existing services rely on OpenID Connect (OIDC) [OpenID.Core] for authentication and attribute retrieval, and implementing a wallet-facing presentation protocol represents a significant barrier for these Relying Parties.

This document specifies how an OP, acting as a bridge, can collect credentials from a wallet using any suitable presentation protocol and expose the resulting claims in standard OIDC ID Tokens and UserInfo responses. The Relying Party interacts with a normal OIDC flow and receives credential data in the "presented_credentials" claim without needing any knowledge of the underlying presentation protocol.

This specification is intended to complement, not compete with, work produced by the OpenID Foundation. Implementers are encouraged to follow developments in the OpenID Foundation's Digital Credentials Protocols working group.

2. Architecture Overview

The following diagram illustrates the high-level interaction between the participants:

+--------+                +------------------+              +----------+
|        | 1. OIDC AuthN  |                  | 2. Present   |          |
|   RP   | -------------> |    OP (Bridge)   | -----------> |  Wallet  |
|        |                |                  | <----------- |          |
|        |                |                  | 3. Response  |          |
|        |                |                  |              +----------+
|        |                | 4. Verify &      |
|        | 5. ID Token    |    extract claims|
|        | <------------- |                  |
+--------+                +------------------+

The presentation protocol (steps 2-3) is out of scope for this specification. The OP MAY use OpenID4VP, DIDComm, or any other suitable mechanism.

The steps are:

  1. The RP sends an OIDC Authentication Request to the OP, including credential type scopes.

  2. The OP initiates a credential presentation request to the user's wallet.

  3. The wallet responds with the disclosed credentials.

  4. The OP verifies the credentials and extracts claims.

  5. The OP returns an ID Token and/or UserInfo response containing the "presented_credentials" claim.

3. Conventions and Definitions

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:

Relying Party (RP)

An OIDC client that consumes claims from the OP. In this specification, the RP receives credential claims without directly interacting with the wallet.

OpenID Provider (OP)

The authorization server that acts as a bridge between the wallet and the RP. The OP collects credentials from the wallet using a presentation protocol and exposes the resulting claims via standard OIDC mechanisms.

Wallet

A user-controlled application that holds digital credentials and can present them to a verifier upon request.

Credential Set

A JSON object within the "presented_credentials" array where each key is a credential type scope and each value is an array of Credential Entry objects.

Credential Entry

A JSON object representing a single credential presented during the presentation flow, containing metadata and disclosed claims.

4. Relying Party Requirements

4.1. Requesting Credential Claims

An RP that wishes to receive credential data via this bridge MUST include credential type scopes in the OIDC Authentication Request (e.g., scope=openid ehic pda1). Scopes are always required because they tell the OP which credential types to collect. The RP MAY additionally include a "requested_credential_sets" claims parameter for fine-grained control.

  • Scope-based (simple): Include only credential type scopes. The OP requests all available claims for those credential types and returns them in the response. This approach requires no additional parameters but offers no selective disclosure or value constraints from the RP side.

  • Scope + claims-based (detailed): Include credential type scopes AND a "requested_credential_sets" member inside the OIDC "claims" request parameter (within the "id_token" or "userinfo" entry). This gives the RP fine-grained control: which specific claims to request, whether each is essential or optional, value constraints, and trusted issuer requirements.

Scopes identify which credentials to collect, and the "requested_credential_sets" claims parameter (when present) specifies how to collect them. If only scopes are present without a "requested_credential_sets" claims parameter, the OP requests all available claims for the given credential types.

The following is a non-normative example of a scope-based request (no claims parameter). The RP requests EHIC and PDA1 credentials with all available claims:

GET /authorize?
  response_type=code
  &scope=openid ehic pda1
  &client_id=https://rp.example.org
  &redirect_uri=https://rp.example.org/cb
  &nonce=n-0S6_WzA2Mj HTTP/1.1

The requested scopes directly correlate to the keys in credential sets returned by the OP. For example, if the RP requests scopes "ehic" and "pda1", the OP will return a credential set containing "ehic" and "pda1" entries.

The RP MAY additionally use the "claims" request parameter to specify which individual claims within a credential type are desired, enabling selective disclosure.

The "claims" member within a credential type entry is a JSON array of claim query objects. Each claim query object contains the following members:

path

REQUIRED. A non-empty JSON array of strings representing the path to the claim within the credential, following the Claims Path Pointer syntax defined in Section 7 of [OpenID4VP]. For top-level claims the array contains a single string (e.g., ["name"]). For nested claims the array contains one element per level (e.g., ["address", "street_address"]).

essential

OPTIONAL. A boolean indicating whether the claim MUST be disclosed. Defaults to true if omitted.

value

OPTIONAL. A JSON value that the disclosed claim MUST exactly equal. If present, the OP MUST verify that the disclosed claim value matches and MUST treat the credential as not satisfying the request if it does not. If omitted, any disclosed value is accepted.

The "requested_credential_sets" member is placed inside the OIDC "claims" request parameter as permitted by Section 5.5.1 of [OpenID.Core]. Its value is a JSON array of credential set objects, each representing one combination of credentials to request.

Each credential type entry within a credential set MAY also include the following optional members to express trust requirements:

trusted_issuers

A JSON array of strings, where each string is an issuer identifier. When present, the OP MUST only accept credentials issued by one of the listed issuers. If the wallet presents a credential from an issuer not in this list, the OP MUST treat it as not satisfying the request. If omitted, the OP applies its own issuer policy.

trusted_issuer_lists

A JSON array of strings, where each string is an HTTPS URI identifying a published trust list. The trust list is a JSON document containing an "issuers" member whose value is a JSON array of issuer identifier strings in the same format as "trusted_issuers". The OP MUST fetch and cache each referenced trust list and MUST accept credentials from any issuer appearing in at least one of the referenced lists. When both "trusted_issuers" and "trusted_issuer_lists" are present, the effective set of trusted issuers is the union of the explicitly listed issuers and the issuers from all referenced lists. If both members are omitted, the OP applies its own issuer policy.

The following is a non-normative example requesting a PID from a specific set of trusted issuers:

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]},
            {"path": ["birth_date"]}
          ],
          "trusted_issuers": [
            "https://pid.example.gov.se",
            "https://pid.example.gov.no"
          ]
        }
      }
    ]
  }
}

The following is a non-normative example using trust lists to accept PID and EHIC credentials from all issuers recognized by the EU trust list, without enumerating each issuer individually:

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]},
            {"path": ["birth_date"]}
          ],
          "trusted_issuer_lists": [
            "https://trust.eu.example.org/pid-issuers.json"
          ]
        },
        "ehic": {
          "essential": true,
          "claims": [
            {"path": ["ehic_number"]}
          ],
          "trusted_issuer_lists": [
            "https://trust.eu.example.org/ehic-issuers.json"
          ]
        }
      }
    ]
  }
}

The following is a non-normative example of a trust list document served at the URI referenced above:

{
  "issuers": [
    "https://pid.example.gov.se",
    "https://pid.example.gov.no",
    "https://pid.example.gov.de",
    "https://pid.example.gov.fr"
  ]
}

The "requested_credential_sets" array MUST contain at least one credential set entry. The structure expresses combinatorial logic over credential types:

  • AND (within a set): Credentials listed in the same credential set with "essential": true are all required for that set to be considered satisfied. If any essential credential in the set cannot be obtained, the set is not satisfied and the OP proceeds to the next alternative set. The following is a non-normative example requesting PID AND EHIC together (both are required):

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]}
          ]
        },
        "ehic": {
          "essential": true,
          "claims": [
            {"path": ["ehic_number"]}
          ]
        }
      }
    ]
  }
}
  • OR (between sets): Multiple credential sets represent alternatives. The OP MUST attempt to satisfy the sets in order and use the first set that can be fully satisfied. Only one credential set is returned in the response. If no set can be fully satisfied, the OP MUST return an OIDC error response (e.g., "access_denied"). The following is a non-normative example requesting PID OR EHIC (either one satisfies the RP). The OP tries the first set; if the wallet cannot provide a PID, it falls back to the second set:

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]}
          ]
        }
      },
      {
        "ehic": {
          "essential": true,
          "claims": [
            {"path": ["ehic_number"]}
          ]
        }
      }
    ]
  }
}

This structure is equivalent to Disjunctive Normal Form (DNF): a flat list of AND-groups joined by OR. It cannot directly express an OR nested inside an AND. For example, the requirement "(PID OR EHIC) AND PDA1" must be manually expanded into two credential sets: {PID, PDA1} OR {EHIC, PDA1}. Complex boolean combinations may therefore require a number of credential sets that grows multiplicatively with the number of OR-branches.

The following is a non-normative example requesting PID (required) with EHIC as optional (nice-to-have):

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]}
          ]
        },
        "ehic": {
          "essential": false,
          "claims": [
            {"path": ["ehic_number"]}
          ]
        }
      }
    ]
  }
}

The following is a non-normative example requesting a PID with a value constraint and an optional claim:

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"], "essential": true},
            {"path": ["age_over_18"], "essential": true, "value": true},
            {"path": ["email"], "essential": false}
          ]
        }
      }
    ]
  }
}

In the above example the OP MUST ensure the wallet discloses "name" and "age_over_18", and that "age_over_18" equals true. The "email" claim is requested but not required; the OP SHOULD request it from the wallet but MUST NOT fail if the wallet does not disclose it.

The following is a non-normative example requesting nested claims from a PID credential:

{
  "id_token": {
    "requested_credential_sets": [
      {
        "pid": {
          "essential": true,
          "claims": [
            {"path": ["name"]},
            {"path": ["address", "street_address"]},
            {"path": ["address", "country"], "value": "SE"}
          ]
        }
      }
    ]
  }
}

If the wallet presents multiple credentials of the same type (e.g., two EHICs for different family members), the OP returns all of them in the array. The RP is responsible for selecting the appropriate credential by inspecting the returned claims (for example, matching on name or date of birth against the authenticated user's identity).

4.2. Consuming Credential Claims

The "presented_credentials" claim structure is defined in Section 5.1. The RP MUST parse the claim according to that definition. Specifically, the RP MUST:

  1. Check for the presence of the "presented_credentials" claim. If the claim was requested as essential and is absent, the RP SHOULD treat the authentication as failed.

  2. Parse the "presented_credentials" array and extract the Credential Set objects relevant to its use case.

  3. Validate that the expected claims are present in each Credential Entry's "claims" object. If the RP specified a "value" constraint for a claim, verify that the returned value matches.

The RP MUST NOT assume that all requested scopes will be present in the response; the user may have declined to present certain credentials. The RP MUST ignore unrecognised credential keys and unrecognised members within Credential Entry objects.

4.3. Trust Model

The RP places trust in the OP to have correctly verified the presented credentials. The RP does not interact with the wallet or credential issuer directly. The trust relationship between the RP and the OP is established through standard OIDC mechanisms (client registration, token validation, TLS).

4.4. Liability and Accountability

The bridge architecture shifts credential verification responsibility from the RP to the OP. This has implications for liability that deployments MUST consider.

The OP is the sole party that interacts with the wallet and verifies credential authenticity, revocation status, and holder binding. The RP relies entirely on the OP's assertion that the credential claims are valid. If the OP incorrectly accepts a forged, expired, or revoked credential, the RP has no independent means of detecting this.

Deployments SHOULD establish clear agreements between the OP operator and RPs that address:

  • The OP's obligations regarding credential verification (e.g., which trust frameworks it enforces, whether it checks revocation).

  • Liability allocation when the OP accepts a credential that turns out to be invalid or fraudulent.

  • The OP's obligations to communicate changes to its verification policy that may affect RP authorization decisions.

  • Audit and logging requirements that allow after-the-fact review of verification decisions.

The "verification" metadata in the Credential Entry (Section 5.1.2) provides a technical mechanism for the OP to communicate verification details to the RP. However, the "verification" object does not constitute a legal guarantee. RPs operating in regulated environments (e.g., healthcare, finance) SHOULD require contractual assurances from the OP in addition to the technical signals provided by this specification.

5. OpenID Provider Requirements

5.1. The presented_credentials Claim

The following is a non-normative example of an OIDC ID Token containing the "presented_credentials" claim with a credential set containing two credentials:

{
  "iss": "https://issuer.example.org",
  "sub": "user@example.org",
  "iat": 1722772800,
  "exp": 1722859200,
  "presented_credentials": [
    {
      "ehic": [
        {
          "type": "https://credential.example.org/ehic/1.0",
          "issuer": "https://svs.example.se",
          "valid_from": 1709251200,
          "valid_until": 1740787200,
          "verified_at": 1722772700,
          "verification": {
            "holder_binding": "key_binding"
          },
          "claims": {
            "name": "John Doe",
            "dob": "1990-01-01",
            "ehic_number": "1234567890"
          }
        }
      ],
      "pda1": [
        {
          "type": "https://credential.example.org/pda1/1.0",
          "issuer": "https://tax.example.se",
          "valid_from": 1709251200,
          "valid_until": 1740787200,
          "verified_at": 1722772700,
          "verification": {
            "holder_binding": "key_binding"
          },
          "claims": {
            "name": "John Doe",
            "dob": "1990-01-01",
            "pda1_number": "0987654321",
            "employer": {
              "name": "Example Corp AB",
              "country": "SE"
            }
          }
        }
      ]
    }
  ]
}

5.1.1. The presented_credentials Array

The "presented_credentials" claim is a top-level JSON array included in the OIDC ID Token or UserInfo response. It contains credential data obtained via a credential presentation flow (e.g., OpenID4VP, DIDComm), re-packaged for consumption by OIDC Relying Parties.

The array contains one or more Credential Set objects. Each Credential Set is a JSON object where each key is a scope value corresponding to a credential type and each value is a JSON array of Credential Entry objects as defined in Section 5.1.2.

Within a Credential Set:

  • Each key MUST be unique.

  • Each value (array) MUST contain at least one Credential Entry. Multiple entries indicate the wallet presented more than one credential of that type.

The outer array MUST contain at least one Credential Set. In most deployments a single set is returned; multiple sets are possible when the RP requested alternatives (see Section 4.1).

Additional members within a Credential Set MAY be present. Implementations that do not recognise additional members MUST ignore them.

5.1.2. Credential Entry Object

Each Credential Entry object represents a single credential presented during the presentation flow. It MUST contain the following members:

type

A string identifying the credential type. The value is protocol- specific: for SD-JWT VC credentials it is the Verifiable Credential Type (vct), for mdoc credentials it is the docType, and for other formats it is whatever type identifier the credential format defines. The OP MUST set this field based on the presented credential.

claims

A JSON object [RFC8259] containing the disclosed claims from the credential. Each key is a claim name and each value is the claim value. Claim names are determined by the credential type and MUST be strings. Claim values MAY be any valid JSON type.

It MAY contain the following additional members:

issuer

A string identifying the entity that issued the credential. For SD-JWT VC credentials this is the "iss" claim value, for mdoc credentials it is the issuing authority identifier. The OP SHOULD populate this field to allow the RP to make issuer-aware authorization decisions.

valid_from

A NumericDate (as defined in [RFC7519]) indicating when the credential became valid (i.e., the issuance or activation date).

valid_until

A NumericDate indicating when the credential expires. The OP MUST NOT include credentials that have already expired at the time of presentation.

verified_at

A NumericDate indicating when the OP verified the credential during the presentation flow. This allows the RP to assess the freshness of the verification relative to its own requirements.

verification

A JSON object providing metadata about the verification the OP performed on the credential. This object MAY contain the following members:

holder_binding

A string describing the mechanism used to verify that the presenter is the legitimate holder of the credential. Values are taken from the "Credential Holder Binding Methods" registry defined in Section 8.3.

Additional members within the "verification" object MAY be present. Implementations that do not recognise additional members MUST ignore them.

The following is a non-normative example of a Credential Entry with nested claims, as might appear in a PID credential:

{
  "type": "urn:eu.europa.ec.eudi:pid:1",
  "issuer": "https://pid.example.gov.se",
  "valid_from": 1709251200,
  "valid_until": 1740787200,
  "verified_at": 1722772700,
  "verification": {
    "holder_binding": "key_binding"
  },
  "claims": {
    "family_name": "Doe",
    "given_name": "John",
    "birth_date": "1990-01-01",
    "address": {
      "street_address": "123 Main St",
      "locality": "Stockholm",
      "postal_code": "11122",
      "country": "SE"
    },
    "age_over_18": true,
    "nationalities": [
      "SE",
      "NO"
    ]
  }
}

Additional members MAY be present. Implementations that do not recognise additional members MUST ignore them.

5.2. Discovery

An OP that supports this bridge mechanism MUST include "presented_credentials" in the "claims_supported" list in its OpenID Connect Discovery [OpenID.Discovery] metadata document.

The OP MUST include a "credential_presentations_supported" member in its discovery metadata. This is a JSON object where each key is a scope value that the RP can use in the authorization request, and each value is an object describing the credential type configuration. Each configuration object MUST contain at minimum:

format

A string identifying the credential format (e.g., "dc+sd-jwt", "mso_mdoc").

type

A string identifying the credential type. For SD-JWT VC credentials this is the vct value; for mdoc credentials this is the docType.

The OP uses this metadata to translate the RP's scope request into the correct credential query (e.g., a DCQL query with the appropriate "vct_values" or "doctype_value") toward the wallet. The Credential Entry "type" field in the response MUST be populated from the presented credential itself and MUST match the "type" value declared in this mapping.

The following is a non-normative example of an RP discovering the OP's supported credential types:

GET /.well-known/openid-configuration HTTP/1.1
Host: op.example.org
HTTP/1.1 200 OK
Content-Type: application/json

{
  "issuer": "https://op.example.org",
  "authorization_endpoint": "https://op.example.org/authorize",
  "token_endpoint": "https://op.example.org/token",
  "userinfo_endpoint": "https://op.example.org/userinfo",
  "jwks_uri": "https://op.example.org/jwks.json",
  "credential_presentations_supported": {
    "ehic": {
      "format": "dc+sd-jwt",
      "type": "urn:eu.europa.ec.eudi:ehic:1"
    },
    "pda1": {
      "format": "dc+sd-jwt",
      "type": "urn:eu.europa.ec.eudi:pda1:1"
    }
  }
}

5.3. Authentication Flow

When the OP receives an OIDC Authentication Request that includes credential type scopes (and optionally a "requested_credential_sets" claims parameter for fine-grained control), it MUST:

  1. Validate that each credential type scope corresponds to a key in the OP's "credential_presentations_supported" metadata. The OP MUST ignore any credential type scope that is not present in its metadata. If none of the requested credential scopes are supported, the OP MUST return an OIDC error response.

  2. Initiate a credential presentation request to the user's wallet for the supported credential types, using the presentation protocol supported by the deployment.

  3. Verify the presented credentials according to the applicable trust framework.

  4. Extract the disclosed claims from each verified credential.

  5. Construct the "presented_credentials" object as defined in Section 5.1.1.

  6. Include the "presented_credentials" claim in the ID Token, the UserInfo response, or both, depending on the OP's policy and the size considerations described in Section 6.1.

Common presentation protocols include OpenID4VP [OpenID4VP] and DIDComm Present Proof. The choice of protocol is a deployment decision and does not affect the "presented_credentials" format returned to the RP.

5.4. Credential Mapping

The OP uses the "credential_presentations_supported" discovery metadata to translate scopes into credential queries. For each credential type scope in the authorization request, the OP looks up the corresponding key in "credential_presentations_supported" and uses the "format" and "type" values to construct the presentation request. The same key is used as the DCQL "id" and as the key in the "presented_credentials" response.

The binding is:

  • The key in "credential_presentations_supported" (e.g., "ehic") is the scope value the RP includes in the authorization request.

  • The OP MUST use this same value as the "id" in the DCQL Credential Query.

  • The OP MUST use the "format" and "type" from the configuration entry to populate the DCQL "format" and "meta" fields.

  • The OP MUST use this same value as the key in the "presented_credentials" response.

For example, given the following discovery metadata:

{
  "credential_presentations_supported": {
    "ehic": {
      "format": "dc+sd-jwt",
      "type": "urn:eu.europa.ec.eudi:ehic:1"
    },
    "pda1": {
      "format": "dc+sd-jwt",
      "type": "urn:eu.europa.ec.eudi:pda1:1"
    }
  }
}

and an RP request with scope=openid ehic pda1, the OP constructs:

{
  "credentials": [
    {
      "id": "ehic",
      "format": "dc+sd-jwt",
      "meta": { "vct_values": ["urn:eu.europa.ec.eudi:ehic:1"] },
      "claims": []
    },
    {
      "id": "pda1",
      "format": "dc+sd-jwt",
      "meta": { "vct_values": ["urn:eu.europa.ec.eudi:pda1:1"] },
      "claims": []
    }
  ]
}

The wallet returns a VP Token keyed by these same "id" values, allowing the OP to map results back to the corresponding scope.

The OP MUST use the scope value as the key within the "presented_credentials" response object. For example, if the RP requested scope "ehic", the resulting entry MUST be keyed as "ehic". This ensures a predictable, stable mapping between the RP's request and the response.

The OP MUST NOT include claims that were not disclosed by the wallet. The OP MUST NOT modify claim values during the mapping.

When the RP uses the claims-based request mechanism (with explicit "path" entries), the OP MUST include only the requested claims in the Credential Entry's "claims" object. Even if the wallet discloses additional claims (for example, because the credential format does not support selective disclosure), the OP MUST NOT relay unrequested claims to the RP. This ensures data minimization regardless of the underlying credential format's selective disclosure capabilities.

When the RP uses scope-only (no "requested_credential_sets" claims parameter), the OP includes all claims disclosed by the wallet, since the RP did not express a preference for specific claims.

The OP MUST enforce claim-level constraints specified in the "claims" array:

  • If a claim query has "essential": true (or the default applies) and the wallet does not disclose the claim identified by "path", the credential MUST be treated as not satisfying the request.

  • If a claim query specifies a "value" member and the disclosed value does not exactly match (using JSON value equality as defined in [RFC8259]), the credential MUST be treated as not satisfying the request.

  • Claim queries with "essential": false that are not disclosed by the wallet MAY be omitted from the Credential Entry without failing the request.

The OP translates each "path" array directly into the corresponding DCQL Claims Path Pointer when constructing the presentation query toward the wallet.

The OP is responsible for translating the RP's OIDC-level request (scopes, claims parameter) into a protocol-specific credential query directed at the wallet. For example, when using OpenID4VP the OP would construct a DCQL (Digital Credentials Query Language) query matching the requested credential types; when using DIDComm the OP would build a Present Proof request. This translation is deployment- specific and outside the scope of this specification. The OP MUST document the mapping between OIDC scopes and the credential types they resolve to, either in its discovery metadata or in out-of-band documentation.

When the RP's request includes a "value" constraint on a claim, the OP SHOULD propagate this constraint to the wallet where the presentation protocol supports it. The RP's "path" arrays map directly to DCQL Claims Path Pointers, and the "value" member maps to the DCQL "values" array. For example, an RP request containing:

"claims": [
  { "path": ["name"] },
  { "path": ["address", "country"], "value": "SE" }
]

would translate to the following DCQL Claims Query entries:

{
  "credentials": [
    {
      "id": "ehic",
      "format": "dc+sd-jwt",
      "meta": {
        "vct_values": [
          "urn:credential:ehic"
        ]
      },
      "claims": [
        {"path": ["name"]},
        {"path": ["address", "country"], "values": ["SE"]}
      ]
    }
  ]
}

However, DCQL value matching is defined as best-effort: the wallet SHOULD filter on the constraint but is not required to do so (see Section 6.4.1 of [OpenID4VP]). Consequently, the OP MUST NOT rely on the wallet to enforce value constraints and MUST always validate disclosed claim values against the RP's "value" requirements after receiving the presentation. Propagating the constraint to the wallet remains useful as a privacy optimisation, because it allows the wallet to avoid disclosing credentials that would not satisfy the request.

The OIDC request model (scopes and the "claims" parameter) is intentionally simpler than the query languages available in presentation protocols (e.g., DCQL in OpenID4VP). This means that certain constraints expressible in a presentation query -- such as issuer restrictions, issuance date filters, or compound field requirements -- cannot be communicated by the RP. The OP MUST apply its own policy to resolve these edge cases (for example, by selecting the most recent matching credential or by restricting accepted issuers via configuration). The OP SHOULD document any such policies so that RPs can anticipate the resulting behaviour.

The structure of the "presented_credentials" claim MUST be independent of the credential presentation protocol used between the OP and the wallet. Whether the OP collected credentials via OpenID4VP, DIDComm, or any other mechanism, the resulting claim format MUST conform to this specification. The RP MUST NOT need to be aware of which presentation protocol was used.

6. Limitations and Considerations

6.1. Claim Set Size Limits

OIDC ID Tokens are typically passed as JWTs [RFC7519] in HTTP headers or URL fragments, which impose practical size limits. Browser URL length limits are commonly around 2048 bytes, and many HTTP servers reject headers exceeding 8192 bytes. When multiple credentials with many disclosed claims are included in the "presented_credentials" object, the resulting token may exceed these limits.

Implementations SHOULD consider the following mitigations:

  • Return credential claims via the UserInfo endpoint rather than embedding them directly in the ID Token.

  • Limit the number of disclosed claims to those requested by the Relying Party via the OIDC "claims" parameter.

  • Use token introspection or reference tokens where supported by the deployment.

6.2. Error Handling

The OP MUST handle the following failure scenarios gracefully:

Wallet rejection

The user declines to present credentials. The OP MAY either omit the "presented_credentials" claim entirely or return an OIDC error response (e.g., "access_denied") depending on whether the credential presentation was essential to the authentication.

Wallet timeout

The wallet does not respond within a reasonable time. The OP SHOULD treat this as equivalent to a rejection.

Invalid credentials

The wallet presents credentials that fail verification (expired, revoked, untrusted issuer). The OP MUST NOT include unverified credential claims in the "presented_credentials" object. The OP MAY proceed without those credentials or fail the authentication.

Partial presentation

The wallet presents only a subset of the requested credentials. The OP MUST include only the successfully verified credentials and MUST NOT fabricate entries for missing credentials.

6.3. Scope Mapping

An OP MAY define custom OIDC scopes that map to specific credential presentation requests. For example, the scope "ehic" might trigger a request for an EHIC credential with a predefined set of claims.

When using scope-based mapping, the OP SHOULD document the mapping between scopes and credential types in its discovery metadata or out- of-band documentation. The OP MUST ensure that the scope semantics are stable and do not change unexpectedly for registered RPs.

6.4. Fresh Presentation Requirement

Each OIDC Authentication Request that includes credential scopes MUST result in a new credential presentation from the wallet. The OP MUST initiate a fresh presentation protocol transaction (e.g., OpenID4VP, DIDComm) for every authentication request and MUST NOT reuse credentials from a previous presentation.

The OP MUST NOT store, cache, or persist credential data beyond the scope of the current authentication transaction. Once the OP has constructed the ID Token or UserInfo response and delivered it to the RP, it MUST discard the credential claims. This ensures that credential freshness is guaranteed and that the OP does not become an unnecessary repository of sensitive personal data.

6.5. Scope of This Specification

This specification defines the data format of the "presented_credentials" claim, the mechanism for requesting and returning credential claims via OIDC, and the responsibilities of the OP and RP in that exchange. The following aspects are explicitly out of scope:

  • Credential validation policy. This specification does not define which issuers to trust, which revocation mechanisms to check, or what trust frameworks to apply. These decisions are deployment- specific and determined by the OP operator.

  • Business logic and authorization decisions. How the RP interprets the received claims -- for example, whether an EHIC credential grants access to a healthcare service, or whether a PDA1 is sufficient for a given transaction -- is entirely the RP's responsibility and outside this specification.

  • Credential issuance. This specification deals only with presentation of existing credentials, not with the issuance of new credentials (which is covered by OpenID4VCI).

  • Wallet implementation. The interaction between the OP and the wallet uses a presentation protocol such as OpenID4VP or DIDComm. This specification does not mandate a specific protocol and does not impose additional requirements on wallet implementations.

  • User consent and identity matching. How the OP associates a credential presentation with an OIDC subject identifier, and how user consent is obtained, are implementation details left to the OP.

In summary, this specification provides the plumbing for transporting verified credential claims through OIDC. Everything above that layer -- trust decisions, access control, and business rules -- is the domain of the deploying parties.

7. Security Considerations

7.1. Credential Replay

The OP MUST ensure that credential presentations are bound to the current authentication session. The OP SHOULD use nonces in the OpenID4VP request to prevent replay of previously captured VP Tokens.

The RP MUST validate standard JWT claims ("iat", "exp", "nonce") in the ID Token to ensure freshness of the "presented_credentials" claim.

7.2. Token Leakage

ID Tokens containing "presented_credentials" may carry sensitive personal data (e.g., national ID numbers, health information). Implementations MUST use TLS for all token transmissions. The OP SHOULD prefer delivering credential claims via the UserInfo endpoint (which uses a back-channel request) rather than embedding them in the ID Token (which may be exposed in browser history or logs).

7.3. Claim Injection

The OP MUST NOT allow external parties to inject or modify claims within the "presented_credentials" object. The OP MUST populate this claim exclusively from verified credential presentations. The ID Token MUST be signed by the OP to protect integrity.

7.4. Trust List Fetching

When the OP dereferences "trusted_issuer_lists" URIs provided by an RP, it MUST enforce the following safeguards:

  • The OP MUST only fetch trust lists over HTTPS.

  • The OP MUST NOT dereference arbitrary URIs provided by an RP. The OP MUST restrict which trust list URIs it is willing to fetch to prevent Server-Side Request Forgery (SSRF). The mechanism for this restriction (e.g., an allowlist, domain policy) is a deployment decision and outside the scope of this specification.

  • The OP MUST impose size limits on fetched trust list documents to prevent resource exhaustion.

  • The OP SHOULD cache trust lists and enforce a minimum refresh interval to limit the impact of a compromised or unavailable trust list endpoint.

  • If a trust list cannot be fetched or parsed, the OP MUST treat it as if no issuers were listed in that list. The OP MUST NOT fall back to accepting all issuers.

7.5. Privacy Considerations

The bridge architecture introduces the OP as a party that observes all credential claims presented by the user. This creates a correlation point. Deployments SHOULD consider the following:

  • The OP SHOULD request only the minimum claims necessary to satisfy the RP's request (selective disclosure).

  • The OP SHOULD NOT log or persist credential claims beyond what is necessary for the authentication session.

  • RPs SHOULD be aware that repeated presentations of the same credential claims across sessions may enable correlation by the OP.

  • Where possible, deployments SHOULD use pairwise subject identifiers to limit cross-RP correlation.

8. IANA Considerations

The claim name "presented_credentials" was chosen to be protocol- agnostic, clearly describing the content (credentials that were presented) without implying a dependency on any particular presentation protocol or namespace.

8.1. JSON Web Token Claims Registration

This specification requests registration of the following claim in the IANA "JSON Web Token Claims" registry:

Claim Name

"presented_credentials"

Claim Description

Digital credential claims obtained via a credential presentation flow, structured for consumption by OIDC Relying Parties.

Change Controller

IETF

Specification Document(s)

Section 5.1.1 of this document

8.2. OpenID Connect Discovery Metadata Registration

This specification requests registration of the following metadata parameter:

Metadata Name

"credential_presentations_supported"

Metadata Description

A JSON object describing the credential types the OP can collect via credential presentation and expose as OIDC claims.

Change Controller

IETF

Specification Document(s)

Section 5.2 of this document

8.3. Credential Holder Binding Methods Registry

This specification establishes the "Credential Holder Binding Methods" registry. The registration policy is "Specification Required" as defined in Section 4.6 of [RFC8126].

Each entry in the registry contains the following fields:

Method Name

A short string identifying the holder binding method.

Description

A brief description of the method.

Change Controller

The entity responsible for the registration.

Specification Document(s)

Reference to the specification defining the method.

The initial contents of the registry are:

Table 1
Method Name Description Change Controller Specification
key_binding Cryptographic proof of possession of a private key bound to the credential IETF Section 5.1.2 of this document
biometric Biometric verification of the presenter against data bound to the credential IETF Section 5.1.2 of this document
pin Verification of a PIN or passcode known to the credential holder IETF Section 5.1.2 of this document

9. References

9.1. Normative References

[OpenID4VP]
Terbu, O., Lodderstedt, T., Yasuda, K., and T. Looker, "OpenID for Verifiable Presentations (OpenID4VP) 1.0", , <https://openid.net/specs/openid-4-verifiable-presentations-1_0.html>.
[OpenID.Core]
Sakimura, N., Bradley, J., Jones, M., Medeiros, B. de., and C. Mortimore, "OpenID Connect Core 1.0", , <https://openid.net/specs/openid-connect-core-1_0.html>.
[OpenID.Discovery]
Sakimura, N., Bradley, J., Jones, M., and E. Jay, "OpenID Connect Discovery 1.0", , <https://openid.net/specs/openid-connect-discovery-1_0.html>.
[RFC2119]
Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, , <https://www.rfc-editor.org/rfc/rfc2119>.
[RFC7519]
Jones, M., Bradley, J., and N. Sakimura, "JSON Web Token (JWT)", RFC 7519, DOI 10.17487/RFC7519, , <https://www.rfc-editor.org/rfc/rfc7519>.
[RFC8126]
Cotton, M., Leiba, B., and T. Narten, "Guidelines for Writing an IANA Considerations Section in RFCs", BCP 26, RFC 8126, DOI 10.17487/RFC8126, , <https://www.rfc-editor.org/rfc/rfc8126>.
[RFC8174]
Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, , <https://www.rfc-editor.org/rfc/rfc8174>.
[RFC8259]
Bray, T., Ed., "The JavaScript Object Notation (JSON) Data Interchange Format", STD 90, RFC 8259, DOI 10.17487/RFC8259, , <https://www.rfc-editor.org/rfc/rfc8259>.

9.2. Informative References

[ISO.18013-5]
ISO/IEC, "Personal identification -- ISO-compliant driving licence -- Part 5: Mobile driving licence (mDL) application", , <https://www.iso.org/standard/69084.html>.
[SD-JWT.VC]
Terbu, O. and D. Fett, "SD-JWT-based Verifiable Credentials (SD-JWT VC)", , <https://www.ietf.org/archive/id/draft-ietf-oauth-sd-jwt-vc-05.html>.

Acknowledgments

TODO acknowledge.

Authors' Addresses

Magnus Svensson
SUNET
Leif Johansson
Siros Foundation
Joel Rangsmo
Siros Foundation