Normative text on VP-within-VP

[rmlearney-digicatapult]

Hello,

Reading through the specification I'd like to ask if VC DM 2.0/2.1 allows VP-within-VP (e.g. EnvelopedVerifiablePresentation) - a sub-graph within the graph.

VC DM 2.1’s does not specifically mention this, and the verifiableCredential property text still only names credentials/enveloped credentials as supported.

However VC-JOSE-COSE permits VP-within-VP:

§ 3.1.2 Securing JSON-LD Verifiable Presentations with JOSE

... Verifiable Presentations in verifiable presentations MUST use the Enveloped Verifiable Presentation type defined by the VC Data Model. ...

§ 3.2.2 Securing JSON-LD Verifiable Presentations with SD-JWT

... Verifiable Presentations in verifiable presentations MUST use the Enveloped Verifiable Presentation type defined by the VC Data Model. ...

If so does the entire VerifiablePresentation have to be wrapped as an EnvelopedVerifiablePresentation to conform with VC-JOSE-COSE or just the sub-graph? i.e. the entire VP could be signed with a Data Integrity Proof but the subgraphs are secured as EnvelopedVerifiablePresentations, and the total graph can contain a mixture of nodes which are Enveloped VC, Enveloped VP, DI secured VC etc.

[iherman]

This was discussed during the vcwg meeting on 24 June 2026.

View the transcript

w3c/vc-data-model#1633

Brent: We will look at them and determine whether we're going to address them here.
… Normative text on VP-within-VP
… Do folks have any questions or need more clarity?
… I will note that the poster is asking questions, so it may be sufficient for the group to answer the question.
… The answers may affect the specification.

Ivan: My feeling is that the answer is no, that the model is such that the VP can have several VCs but not another VP. As far as I remember, this is the model.
… I would prefer having someone deeper in the technology give a definitive answer.

Phila: I'm not qualified to answer directly, but it looks to me like this commenter wants links to other credentials. If the answer is no, could the answer be to look at recognized entities?

Brent: The VC Data Model says the a presentation contains a set of VCs. The JOSE-COSE documentation says that it can contain VPs as well.

Ivan: I think you're right, Brent, but that means for me that the JOSE-COSE document is incorrect in terms of the model. I personally do not see how recognized entities help.
… I would like to see an example that would be solved by this.

Brent: The description of the VP is that it contains multiple VCs and can contain arbitrary data, and that the VCs may be from multiple issuers.
… Conceptually, there wouldn't be a problem with including another VP, but it doesn't explicitly say that that can be done.
… I think that the VC JOSE-COSE recognizes that need from the supply chain space. Often folks would be receiving VPs and would often want to pass them along directly rather than extract the VCs.
… What I'm reading in the core data model section of our spec is that this is possible but in order to make it explicitly possible it would require normative and substantive changes to the VC data model.
… These changes would be beyond what we're currently chartered to do.
… I'm adding the class 4 and future labels.

---