[MINUTES] VCWG VCALM 2026-07-07

Here's a summary of the VCWG VCALM meeting held on July 7, 2026:

The VCWG VCALM meeting focused on several key technical topics related to
verifiable credentials and their associated APIs. Discussions included the
progress and structure of the VC Test Suite, the review and potential
merging of the Threat Model PR, and proposed simplifications to the OpenAPI
Specification to better align with the Verifiable Credential Data Model.
The meeting also touched upon OAuth 2.0 authorization flows and the
inclusion of ZCAPS in the specification. Overall, the meeting saw good
progress on several fronts, with a clear path forward identified for
merging key documents and advancing the working group's goals.

*Topics Covered:*

   - *AI Participation Policy:* The group discussed and agreed on a
   preliminary policy for AI participation in meetings, requiring
   identification of the AI's owner and their legitimacy within the ecosystem.
   - *VC Test Suite Progress:* Patrick St-Louis presented the initial PR
   for the VCOM test suite, detailing its structure, approach to normative
   requirements, and how it will integrate with other test suites, focusing on
   API behavior and end-to-end workflows.
   - *Threat Model PR Review:* The group discussed the ongoing review of
   the Threat Model PR, with a plan to address controversial comments by
   creating new issues and merging the core document soon.
   - *OpenAPI Specification Simplification:* A PR was presented to simplify
   the OpenAPI Specification by removing detailed credential definitions and
   focusing on context and type, aligning it more closely with the VC Data
   Model.
   - *OAuth 2.0 Authorization Flow:* Kayode Ezike presented PRs related to
   OAuth 2.0, which aimed to add more detail to scope requirements and include
   a section for authorization capabilities.
   - *ZCAPS Support in Specification:* A PR was also discussed for adding
   ZCAPS support to the specification, including defining their format and
   providing examples.

*Action Items:*

   - Patrick St-Louis will close the current VCOM test suite PR, squash
   comments, and reopen it from a W3C branch, also sending a message on Slack
   about it.
   - Eric Schuh will work on the Threat Model PR, creating issues for any
   unresolved controversial comments by the end of Monday, and will ping
   Patrick on Slack to determine if it should be merged before the next call.
   - Kayode Ezike will re-review PRs 656 and 657 for the OAuth 2.0
   Authorization Flow and ZCAPS to ensure normative references are handled
   correctly based on the discussion.
   - The group will aim to merge the VCOM Test Suite PR, the Threat Model
   PR, and the OpenAPI Specification Simplification PR at the next meeting.

HTML: https://meet.w3c-ccg.org/archives/w3c-vcwg-vcalm-2026-07-07.html

Video: https://meet.w3c-ccg.org/archives/w3c-vcwg-vcalm-2026-07-07.mp4

[image: W3C] <https://www.w3.org/>
VCWG VCALM 7 July 2026 Attendees

Present

Dave Longley, Elaine Wooton, Eric Schuh, Joe Andrieu, John's Notetaker,
Kayode Ezike, Kevin Dean, Manu Sporny, Nate Otto, Parth Bhatt, Patrick
St-Louis, read.ai meeting notes, Rodrigo, Ted Thibodeau Jr

Regrets

-

Chair

-

Scribe

transcriber
Contents

   1. AI Participation Policy <#ac59>
   2. VC Test Suite Progress <#646d>
   3. Threat Model PR Review <#a8f0>
   4. OpenAPI Specification Simplification <#9955>
   5. OAuth 2.0 Authorization Flow <#30e0>
   6. ZCAPS Support in Specification <#dc67>

Meeting minutes

Patrick St-Louis: Welcome everyone. We'll get started in a couple minutes.

Patrick St-Louis: before we get started. So I can see in the message
individual name all this has added a AI bot to attend the meeting. while we
don't formally have policies for AI participation, we usually require that
we know who the person registering the AI does anyone know who all this E
is?

Patrick St-Louis: Okay, it looks like they have left the meeting.

Manu Sporny: I booted it.
AI Participation Policy

Patrick St-Louis: I was just going to ask in case somebody knew who they
were. let's adjust to the topic for today. how we should treat AI script
software. okay let's get started. it's past the hour. we have pretty good
attendance. as people come in they will be able to catch up. so today is
the 7th of July 2026.

Patrick St-Louis: This is the W3C verifiable credential working group VCOM
call during which we discuss verifiable credential API for life cycle
management. This is a W3C meeting so all W3C policies and contribution
agreements are into effect. before we get into the agenda, we're going to
leave room for introductions and community announcements as well as any
additional topics, users would like to cover.

Patrick St-Louis: I will add a small topic to start which is just to
refresh how we want to manage AI software that attends the meeting. It
should be a very quick discussion. So I will leave the floor open for
introductions or community announcement. Very good. yeah. So my current
understanding for AI participation is that it would be allowed if it's from
a known person that is known in the ecosystem and we have an agreed upon
understanding.

Patrick St-Louis: if there is a unknown participant which refused to
identify themselves we will unfortunately have to remove them. that is this
along with what everyone is thinking here are these some decision we can
agree on okay this is what we've been doing I don't think there's official
guidelines for this but this seems like a pretty reasonable take we don't
want to prevent people from leveraging AI tools but this needs to be done
in agreement and in good faith

Patrick St-Louis: Yes, Joe. Yes.

Joe Andrieu: I want to suggest one writer on that…

Joe Andrieu: which is and that the person is a legitimate contributor
either they're a member of the work group or they're an invited expert. I
think it's going to be too easy to pretend that you represent some person
that we don't actually know. I think in other words, from public people
that we don't know is hard to validate. so that feels like a threshold that
we should pay attention to.

Ted Thibodeau Jr: along similar lines. I don't think I've seen Rodrigo
before and there's nobody listed in the group as a participant with that
name.

Rodrigo: I actually applied a few days ago. I've introduced myself
sometimes. I work in the Spanish University in Madrid and we are currently
studying self identity but applied for data spaces. I have a question I
don't know when…

Patrick St-Louis: Man, want to reply.

Rodrigo: but I applied and I don't know when I'm getting an answer from it.
I don't know if there's a problem with it anything like that.

Manu Sporny: Are Rodrigo you're referring to your invited expert request?

Rodrigo: Okay. Yes.

Manu Sporny: I guess the question is which request? So, this is an official
working group of the worldwide web consortium. This is not a community
group. and to join and participate in this you need to be either a member
of the verifiable credential working group or an invited expert and there's
a way to get an invited expert form and get that filled out. Is that the
thing that you're referring to or are you referring to the credentials
community group thing?

Rodrigo: Erh, I'm kind of young and new into this, so if you could guide me
a little bit. Okay.

Manu Sporny: Yes, happy to. let's put you in touch with the chairs and the
staff contact to see if we can invite you in as an invited expert because
we don't have participation from people in your area of the world and I
think it would be good for that to happen. and then of course the standard
you can't contribute or participate to anything that we're doing until that
happens and it is up to the organizers of this group whether to allow you
to participate in the group until your invited expert form goes through.
So, I'll keep it there.

Rodrigo: All right.

Manu Sporny: And then hand it over to you, Patrick. I do have something
else to, talk about, from an AI perspective, though. but go ahead, Patrick.

Patrick St-Louis: Yeah. …

Patrick St-Louis: I just wanted to know is there a weekly credential
community group that Rodrigo could go and introduce himself and have this
as a first kind of introduction pathway to the verifiable credential
community group.

Manu Sporny: That would be a good thing for you to do Rodrigo to kind of
present your work to the credentials community group.

Rodrigo: Okay.

Manu Sporny: That group is open to anyone from the general public. you can
sign up yourself. You don't need to be approved, all that kind of stuff. So
I think that that would be one good thing for you to consider doing.

Manu Sporny: But separately and in parallel, I don't think Patrick that's
required for Rodrigo to apply for an invited expert position in this group.

Patrick St-Louis: definitely not was just proposing a way again like to
introduce himself.

Patrick St-Louis: As for this group I personally have no problem if Rodrigo
want to attend. if there's any objection we can discuss this but yeah I
think the credential community group is a very nice place to be especially
if you're new to the community you'll find a lot more sort of incubation
specifications and there's some very interesting discussions that happen on
that side as far as the verifiable credential working

Patrick St-Louis: group we are mostly focused on bringing different
specification to a recommendation status. So often time what will happen is
that some specification will be proposed and incubated at the CCG and then
once they reach some level of maturity they will apply to be promoted to
the verifiable credential working group which is discussed with more of a
production mindset and we take more time to go into the different details
security considerations and whatnot.

Patrick St-Louis: So yeah that could be a good place for you to join…

Patrick St-Louis: but I personally have no issue with you attending this
call again unless there's objection from any of the other organizers.

Rodrigo: Okay.

Rodrigo: All right. I'll join the community group.

Patrick St-Louis: Other than that, thank you for the introduction and
welcome Manuel

Manu Sporny: To follow up on one of Ted's comments about the use of
recorders and everything, agree with everything that's been said to date.
The one only caveat is there it is possible for us to go off the record by
going into another room using this stuff. those AI tools will follow us
into the other room and keep recording if we're not careful. Just pointing
that out as that's the only place where I can think of an AI tooling kind
of going wrong. in addition if they're meeting summarizers and things like
that, it's kind of like we already record this stuff and we already put
transcriptions out and we put it out there.

Manu Sporny: I think my in and in and there are times where those AI bots
have messed up the recording where I've had to kind of recover it. I
haven't seen that happen in the last year. So that's good. I think they're
behaving themselves mostly. But that's another kind of like, I'm a bit
skeptical that pulling an AI tool into one of these meetings is going to
actually provide useful is going to be better than worse. But I think
what's been said today is fine and we can deal with special cases as they
come up. That's it.
VC Test Suite Progress

Patrick St-Louis: Yeah, I think I agree with this talking about it but no
need to make it a big issue until it is one which it is not currently. I
think the way we deal with it is swift and very reasonable. any other
comments before we get started with today's agenda? Great. so I want to
start with the test suite progress. So I've been putting a lot of work into
this. I do have the first initialization PR to the VCOM test suite repo.

Patrick St-Louis: I do have a lot more work than what is in this PR
covering most of the sections and I want to take some time today to discuss
a little bit so the PR that's open there I'll open it in a second it's more
a scaffolding of the repo so put in the required reporter making sure all
configuration file are there and an initial readme that explains a little
bit the approach for the test suite there are a couple topics I want to
address mostly for how will an submit their implementation configuration
file and how should the test a little bit talking about how should this
happen in conjunction with the other test suites.

Patrick St-Louis: Because the way the test suite work is there's a
centralized implementations repository and then you just tag which test
suite that you want to run. up until this time we have been sort of
unofficially using vcom defined service endpoints to test specific feature.
but this being the VCOM test suite there's a few things that we need to
consider and probably a few decision point I have of course a proposition I
reflected on this came up with a proposition which I'd like to share with
the group get some feedback and then I can think about this some more so I
will open the PR I've tagged a few people in

Patrick St-Louis: Okay, so this is a brand new test suite. The way that
we've been usually building test suite is to have a look at the spec, break
down the normative requirements only including what is required. So these
include must not and required statements for may or should.

Patrick St-Louis: They are only included if the spec is written in a way
that if you choose to include this optional statement, there is a normative
statement following it. trying to think of an example for example in the
verifiable credential data model there's a should for people may use
verifiable credential for envelope proof if they choose to do so. there's a
strict data model that they need to adhere by. so in those situation we
want to make sure that if people do not support envelope verify credential
they're not penalized on the test suite reporting side. So they're not
getting showed up as they don't support it because it's not required.
However, if they do we want to ensure that they've implemented in the
correct way.

Patrick St-Louis: For how this test suite is scaffolded. so since here we
have a blank slate I wanted to create this pipeline of how we designed the
test suite what's the matic so in this initial PR I've included a file that
contains all the normative statements that are to be tested. so before we
had some Excel spreadsheet that contained these and we always had to kind
of cross reference. so I think for here I'd like to have these. So this is
a sort of here what your spec needs to be tested.

Patrick St-Louis: This should be generated with some kind of script parsing
that parses the spec and extract the must statement. So this parser needs
to be respect aware. respspec already tags these keywords in the spec
itself with some specific ID in the HTML itself. So it's very convenient to
parse. and then we want to kind of, just arrange them by section. So, this
allows us at any moment to see if what is covered in the test suite
reflects the current version of the spec. Yes, it means that if there's a
change to the spec, we'll need to come and change this file. but by
centralizing everything here, from experience, it makes this a little bit
better.

Patrick St-Louis: And for Ted's answer, you should be looking at the M test
suite. The UPSC ID is my fork. it's where the PR is coming from. so if we
go into the W3CVOM test suite, there is one pull request which is coming
from my fork into the W3C main. this is usually how I contribute to open
source project. if we prefer that we only create branches on the W3C repo,
I'm happy to change to that behavior. so if you see the diff up here, is
there? Yes, Manu.

Manu Sporny: Yeah, plus one to doing that. I mean, we should just give you
admin and maintenance access to this repo and you should be able to just
raise pull requests directly on it. yeah.

Patrick St-Louis: Yeah, I can do that.

Patrick St-Louis: I believe I have the rights already. This is just how
I've been kind ofuting to project.

Manu Sporny: No, I could totally understand why you did it the way you did,…

Patrick St-Louis: Yeah,…

Manu Sporny: but I think it's better to just keep it all in the BCOM test
suite.

Patrick St-Louis: that makes a lot of sense. so I'm happy to, instead of
merging to main, I can just merge create a new branch on VCOM and recreate
this PR. So yeah so this is the tally we have now from experience it's been
the reason why we adopted this way of just focusing on the must statement
is because we need to think suite right the purpose of this test suite is
really to say is there enough implementers of this spec right so we want to
stay as much black and white as we can only covering what is normatively
required

Patrick St-Louis: And if we come up we see there's a statement which cannot
be tested. There's no way to test this statement. we can then rethink
should this really be a normative statement so far all the normative
statement I've identified are testable and usually with each of the
statement we need to include at least one positive test and one negative
test right so the goal for this is just so that we don't have an
implementation that just returns 200 at every single request and they can
cheat so it's finding this balance

Patrick St-Louis: of making sure it accurately represents the spec right
there's always going to be if someone really wants to go to extra miles and
really cheat the test suite go ahead be my guest we'll do a reasonable
effort to make sure that it is accurate so in the vcom it's a bit
interesting thing because we have a OAS specification along with it.
usually most of the other test suite are really either data model based or
algorithmic based right.

Patrick St-Louis: So whenever we talk about the data model obviously we
want to make sure that the response are whether their software accepts
reject incorrect bodies and returns correct bodies right with any kind of
crypto suite test we want to ensure that the solution can actually verify
the crypto suite and detect bad misformed proof and also that they can
issue credential with their selected crypto street that can be verified by
our test client. so this means that in things like the verifiable
credential data model test suite, we do not make sure that the solution
generates a valid proof. We make sure that the body of the credential is
correct. The proof verification is done with our crypto suite test suite.

Patrick St-Louis: So usually what people will do is they'll have an
endpoint tagged with verifiable credential 2.0 and whatever crypto suite
that issuer service is meant to issue and this will run both test suite.
The test suite will ensure that the data model The crypto suite will ensure
that the securing mechanism is well implemented and then when you mix both
together you end up with a comprehensive test suite. we made this because
otherwise we would just have every test suite repeat everything and it gets
a bit convoluted. So in this approach this VCOM test suite will not focus
on testing the data model of the credential and it will not focus on
ensuring that credential have a valid signature mechanism in them.

Patrick St-Louis: Instead it will focus on testing I endpoints and API
behavior. the only part that will have more extensive testing will be
ensuring that workflows can be conducted end to end. so we can foresee
someone tagging an endpoint with VCOM data model 2.0 and a specific crypto
suite, right? And this will run all three test suite on that endpoint. Each
test suite kind of testing their own respective things. for the layout of
the test suite, so we're going to have, our normative taste statement file,
which is what all the test description will be based on.

Patrick St-Louis: And then we're going to have directories kind of matching
the specification. It should be very easy when we look at the project and
the test to see what it corresponds to in the spec, We shouldn't have to
guess it's talking about issuing credential it must be the issuing service.
No right you're going to have the section 3.2 two issuing service and then
you're going to have the normative statements and series of tests. so that
when we want to answer the question do we have enough coverage it should be
fairly straightforward to get the answer right it should be able to do that
with a script or a program. for the configuration file. So this is where
things get a bit interesting.

Patrick St-Louis: So I think it's fair to say for things tag vcom we want
to give them service instance endpoints. this makes a big change because
previously our issuers we gave the credential issue endpoint. Now in the
vcom spec when we think about issuers there's actually three endpoints that
they need to support credentials issue being one of them I think credential
get is another one and credential delete for verifiers right verifiers are
meant to support credential verify and credentials presentation and then
holders workflows service right they define the different

Patrick St-Louis: end points they have and interaction that maybe we'll
keep maybe not. So this creates a interesting dilemma of do we still want
the issuers to point to a specific credential issuance endpoint or do we
want issuers to point to the issuing servicebased endpoint and have the
test suite infer that the issuer service must have this and must have the
other endpoints. we have the benefit of having a tag. So we can tell
existing test suite that if you have a tag vcom you need to interpret the
endpoint as an issuer service and not an issuance endpoint. if that makes
sense. So that's kind of the way I've approached it.

Patrick St-Louis: Yes, manu. So the biggest difference is so we add
verifiers and…

Manu Sporny: I mean, I think it's fine to go with that approach for now and
see if implementers complain about it, if we get an implementer's like, I
didn't set my thing up and I don't know how to get it to work. I think at
that point we can revisit it. but I think this should work for most of the
implementations that I know of anyway.

Patrick St-Louis: VP verifiers So by doing this approach we're condensing
both and the verifiers meaning if you have a verifier with VCOM you are
normatively expected to support both credentials and presentation
verification. that's one of the normative requirement for the verifier
service.

Patrick St-Louis: Where this causes problem is if someone already has an
issuance endpoint and they just want to add the VCOM tag, this will likely
not work. but I don't think that's something we want to encourage. I think
we want to encourage if you already have an issue endpoint with this suite,
leave this one as is and if you want to support VCOM, just add, your new
issuer VCOM endpoint. otherwise, we're going to get into all sort of
backward compatibility. weird artifact. So what you have now needs to keep
running. Basically that's my own focus when we define this test suite and
we had implementation we shouldn't require anyone to change what they have
in there to support it. We should just require if they want to support this
new tag to present this new issuer service.

Patrick St-Louis: Yes madam.

Manu Sporny: Plus one to that. I think that's exactly the right thing to
do. and I'm wondering yeah I totally forget that I think the issuance and
verification of the previous test suites that we had for this those were
preliminary first attempts at it, And I think that the VCOM test suite is
going to effectively deprecate those old things if I remember correctly. So
I think we should expect those things to go away in time and it's but I
agree with you let's not break the ecosystem while we're doing this upgrade.

Manu Sporny: Let's go ahead and upgrade. let's get a nice healthy set of
implementations for VCOM and then once we feel pretty comfortable that
there's enough broad implementation of VCOM we can kind of let those other
implement implementers say hey we're going to start deprecating this and
we're probably going to end up removing it in a year or…

Patrick St-Louis: Yeah. Yeah.

Manu Sporny: so right give them plenty of heads up to migrate that's

Patrick St-Louis: I was going to touch on that. So the two test suite
you're referring is the VC API issuer and API verifier test suite. And from
what I've heard, these were originally kind of put there as a C playground
integration step. Meaning if you wanted to integrate with VC playground,
just run this test suite as a first step. and yes, that's one of the first.
So there's the two next PR I'm going to open on there. The first one is
going to be some open API testing, which I'll get to in a minute.

Patrick St-Louis: And the second one will be essentially porting this VC
API issuer issuer and verifier service tests right which should cover the
equivalent of what these test suites do. one last thing I wanted to add the
challenge we have for this. So let's say I have an endpoint I want to tag
VCOM VC 2.0 and a specific crypto suite. one thing we could look at doing
if we want to support this in the future is that for the crypto suite test
suite or data model 2.0 test suite, they can look has the current
implementation has a VCOM tag on it. If I'm going to append credentials
issue to the base endpoint. Right?

Patrick St-Louis: So then people they don't need to have two separate
entries if they want to test VCOM and VCDM 2.0 they can just do the same.
that's going to be maybe later on because currently if we have VCOM and 2.0
the 2.0 is going to pull that domain expect that it has credential issue
send the credential to sign and the endpoint is going to be like it's not
the right endpoint it's just the base service endpoint so since we have the
vcom tag I think that's a backward compatibility path that can exist for
people that want to tag more than just the vcom service that's going to
have minimal footprint

Patrick St-Louis: and all the other test suite right I think we just need
to add a vcom handler that means if you have an issue with vcom and you
want to test 2.0 I'm just going to add credentials issue at the end simple
as that yes man Yes.

Manu Sporny: Yeah, I think that's okay. I'm wondering if we might be kind
of overdesigning it, meaning the alternative is they just duplicate the
entry, right? They just put another thing. And I think that's okay, I mean,
it's not super great. And I get what you're saying. but we're talking about
designing for a group of 10 or 20 people on the planet, right? I think they
can copy and…

Patrick St-Louis: Yeah. Yeah. Yeah. Yeah.

Manu Sporny: paste in the very worst case and in the best case, it's
output. Yeah, that's

Patrick St-Louis: I think that makes sense, right?

Patrick St-Louis: This idea was really at the end because some of the
implement they have already quite a few endpoints but yeah you can just
serve add another issuer entry same base you just add in the entry the
credentials issue and it's going to be sorted so definitely not priority
but yeah it's just kind of what I was thinking about the first goal we just
want to have something that ts vCom and then tests vcom and then we can
think

Patrick St-Louis: about how we can improve it. yeah, so the last thing I
wanted to mention so initially I had this idea of using a thing called
schema thesis for testing after some more research so there is a chai open
API testing it's a little bit less extensive but I think it's going to fit
a lot better more seamlessly into the current reporter. So we use mocha and
chai as kind of part of it.

Patrick St-Louis: schematasis was fun. it's very comprehensive. It does a
lot of fuzzy kind of testing but it's very I want to say parasitic to this
it's kind of its own thing. It doesn't fit well in the reporter as the chai
open API I think is going to be a lot more seamless. we can just kind of
give it the OAS file and ask it to test whatever is relevant to the
endpoint we're testing. I think that's probably the best solution here.
it's also less involvement, less configuration and everything. Schemais you
had to do your own configuration file. So that's yet another thing people
would have to do. yes, Mario.

Manu Sporny: Yeah, a plus one to that. That sounds good as well. are you
this fits in…

Manu Sporny: what you're proposing fits into the way that we're doing
reporting for all the other test suites. Is that correct? Did I hear you
correctly?

Patrick St-Louis: Yeah. Yeah,…

Patrick St-Louis: that's…

Manu Sporny: Okay, that's good.

Patrick St-Louis: what I meant. the digital bazaar reporter that feeds into
So I got as far as generating a test suite page,…

Manu Sporny: I think you mean the Can I VC site?

Patrick St-Louis: the report page that test it renders nicely. I've
obviously not tested all the way to VC playground, but if we get a nice
conformant respspec test result page, I'm assuming it's going to translate
nicely to the VC playground. because it Sorry. the KIVC,…

Manu Sporny: Yeah. Yeah, that's right.

Patrick St-Louis: Because it's the same JSON.

Manu Sporny: Yep. Yep.

Patrick St-Louis: I have a few suggestions to make some change there.

Patrick St-Louis: Might as well. It's going to be the last thing I show
there. Then we can kind of go on somewhere else. So, I kind of just wanted
to propose maybe on the reporter, we can kind of have a test title, Instead
of showing the statement here, it would still be tagged in the spec. the
conformant could appear when you hover it would still be tied to it but
maybe just give a nice test title that I can read because when I read this
it's like okay what is this doing because this could be securing mechanism
right okay yeah this is a test for securing mechanism I think this can be
done without issue so that'd be kind of one thing I'd propose yes manu

Manu Sporny: Yeah, plus one to that. none of that reporting infrastructure
is precious. we want to, greatly improve it's kind of ugly. It's all
smashed together. we definitely want to make those improvements you
mentioned to make readability better. plus one to that. the most important
thing of course is that it fits in some way to the KIVVC site so that
people that are implementers can show whoever they're trying to convince to
use their software hey this is how I'm doing and as you can see it it does
very well across these tests that the customer care about.

Manu Sporny: So I definitely want to make sure that we don't lose that and
you're saying absolutely we're not going to lose that all that continues to
work.

Patrick St-Louis: Yeah. Yeah.

Manu Sporny: So all that's great. but yeah, I mean the test reports are
ugly right now. It would be really, really nice to give them a refresher.

Patrick St-Louis: And I think kind of does that, right? Because when you
see here, it's nice. so here, the test name, we could have a shorter test
name with a little when you hover over it, it shows the normative test
statement that corresponds to it. It will maintain this link. here, it
already looks a lot better.

Patrick St-Louis: But things like this it's a lot to read just to know what
is this doing right we could summarize this in a short name that explain
what it is and then if you want more detail exactly how this test was
normatively requested in the spec you can kind of read this I think a small
name with an information bubble is pretty standard HTML thing so I can talk
with Ben and we should be able to just put this in small addition. It's not
changing anything. It's just adding this little test name and like it says
test name. That's not a name, right? That's a statement. yeah. So, that's
kind of where I'm at. So, I'd encourage people to read this PR. I'll open a
new one from the W3C.

Patrick St-Louis: mostly read me if there's anything that seems problematic
here or any suggestion to comment on it and then once this is in I'll
submit two new PR one for open API testing for the services and the other
one for covering the VC issuer verifier test suite and then after that I'll
do other probably divide

Patrick St-Louis: service and we'll probably leave workflows for last. I
think that's going to be the main thing we'll want to test here. All the
other service are pretty lowhanging fruit, right? the status update, issuer
endpoint. this should be easy to test and then we'll finish by working on
workflows and think about interop testing. If you want to have two services
kind of do workflow things. yeah, that's as much as I have. Is there any
comments?

Patrick St-Louis: Yes, man.

Manu Sporny: Yeah, this is all great. Patrick, thank you so much for
working on it. I mean people should I did, read all the sections as you
were going through them and they look good to me. my preference would be to
merge this sooner than later. maybe give it a week for folks take a look at
it. But this is very much aligned with, what I think we were at least I was
hoping for and it looks like it slots in really well and can get us in
really good shape to go into candidate w.

Manu Sporny: Plus one emerge from me. my suggestion is we give it another
week for people to provide any comments they want to and then we just go
forward with what you've done. Patrick.

Patrick St-Louis: Perfect. So,…

Patrick St-Louis: what I'll do, I'll close this squash the comments, reopen
one from a branch at W3C. I'll send a message on Slack, and then next week
we can merge it, if there's no objection. Perfect. So, this took way longer
than I expected, but I think it was, significant, to discuss this.

Patrick St-Louis: So the next thing I don't know if we want to take time to
discuss threat model because I saw there's a PR open so maybe we want to
fold this discussion in the PR review. is there any kind of updates
discussion that's not covered by the PR that someone would like to discuss.
Threat Model PR Review

Patrick St-Louis: looking at you, Eric and Joe. Don't want to put you on
the spot, but yes, Eric.

Eric Schuh: Yeah, I guess I can just talk a little bit about maybe the path
forward and…

Eric Schuh: see what the group thinks. I missed last week's call, but I see
that especially Manu and Ted have started commenting on the PR. I'm not
sure how long we want to give that. I guess I was tentatively planning on,
maybe giving it to the end of this week and before next week's call, I'll
go through and, basically accept any of the proposed changes that seem
uncontroversial, so there's no discussions happening around them. a lot of
TED's editorial changes would fall into that category. as well as a few of
Manu's, suggestions around, different responses. And then I guess where I'm
not as sure how the group would like to handle it is for some of the I
guess controversial changes in that I know there were a number of responses
that Manu disagreed with.

Eric Schuh: And I think my general sense was the cleanest way to do this
might be to simply remove a number of those since there's clearly
disagreement about what the content I provided. which is all fine and and
then open up individual issues after we kind of merge the massive PR in to
respond to any of the threats that are missing responses in that manner. So
basically accept anything that seems uncontroversial. strip out the
responses that were causing controversy and then open up individual issues
so that we can deal with things more peace meal rather than in the large
file.

Eric Schuh: But happy to take suggestions. So I think that's kind of where
my head's at. So I'll leave it there.

Patrick St-Louis: Thank you.

Patrick St-Louis: Manuel

Manu Sporny: Yeah, I mean I appreciate the care that you're taking in
processing but I don't want to slow it down. I mean, I don't know if we
need to remove things. I'm fine with it being merged with those kind of
things that are I don't think not quite accurate, being in there. I think
it's less at stake, Because threat model is a living document. We're going
to be modifying it as things go on and I've made it very clear, where I
think the misalignments are.

Manu Sporny: I don't think it's the end of the world if we just merge it
and iterate on it quickly. So, I would much rather we iterate on it rapidly
instead of kind of waiting for, something more perfect to happen. plus one
to leave it till the end of the week and then just basically merge in a way
that you feel is appropriate. that's certainly what my preference would be.
because I would much rather us get to the point…

Patrick St-Louis: Joe.

Manu Sporny: where we're firing off the horizontal review and actively
moving towards

Manu Sporny: then trying to polish the threat model more than we need to at
this point. that's it.

Joe Andrieu: Yeah,…

Joe Andrieu: plus one for managed comments. I think we can move this a
little faster than maybe you were thinking, ic, I have already merged in a
bunch of things that I deem to be editorial. so a lot of tall Ted's
comments I pulled in already. there are some things where also I said hey
that's interesting manu but do you have text and manu did in fact come back
and…

Patrick St-Louis: Eric. My

Joe Andrieu: provide some suggestions so I think we are getting some good
healthy back and forth in the comment thread itself. So please engage there
and I think to the extent that there are things that don't resolve with
that back and forth in comments then definitely next week we should have
some time on agenda to see…

Joe Andrieu: what they are and try and clear them out.

Eric Schuh: Yeah,…

Eric Schuh: that sounds good. And then I think what my plan and this is
partly for Joe as well in case you're doing more work through this week
will be to come through this on Monday or Tuesday morning before the call
here and try to clean up anything that is uncontroversial. I'll leave
anything open that the group might want to look at before we merge and then
we can hopefully deal with a few outstanding things and merge this next
week.

Patrick St-Louis: Yeah. Heat.

Manu Sporny: And again, I appreciate the care, but'm not as long as if it's
just my comments, merge it on Monday. and then we can talk about it, on the
follow-ups on Tuesday.

Manu Sporny: But clearly, if other people are objecting have an issue with
it, then that's fine. we can discuss. But just underscoring my desire to
see this thing merged and in there and for us to have a base to work that's

Patrick St-Louis: That's okay.

Ted Thibodeau Jr: Yeah, I'll just suggest that anything that seems to be
controversial that remains including manage comments. If they just get
turned into fresh issues and then this gets merged,…

Ted Thibodeau Jr: it means that we don't lose any of the comments and all
the material that's there is useful. That's it.

Patrick St-Louis: Thank you.

Patrick St-Louis: Yeah, I think this all makes sense. what I would propose?
So to ensure that by the time we start the meeting next week, everything
here is addressed, issues are created, it's in a good place to be merged
and we focus on getting it merged at the meeting. that leaves you the time
to have other things prepared and we can start with this and ideally we
should be able to merge it right away and then branch off in the other
issues.

Patrick St-Louis: does that work with what we discuss? it's not clear to me
if we wanted to add this merge before the meeting after I would propose
getting it in the place that we can basically click the button on the
meeting and present the issues that have come out of it.

Patrick St-Louis: Is that reasonable with everyone? yes, Okay.

Eric Schuh: Yeah,…

Eric Schuh: I was going to just say I'll plan on trying to getting through
kind of my review having all the issues created by the end of the day
Monday and then I'll ping you Patrick in Slack and if you want to choose to
merge it before the call you can do that and if you want to leave it to the
call for a quick review we can do that as well.

Patrick St-Louis: Does that work with what you said, is that fast enough or
you'd really like this to be merged before the next call?

Manu Sporny: Nope, that works just fine.

Patrick St-Louis: We'll make sure to put a pin on it.

Patrick St-Louis: I'll make sure it's the first topic we discuss. Any other
closing comments on threat I put a little topic. We can skip this for
today. Maybe talk about this the next time. So I think Maned some of it
about our goals to get horizontal review. So I just wanted to have a quick
chat about kind of where we are at what's next. it sounds like we are on a
good track. So maybe we can just being mindful of the time there is a PR
I'd like to get us to kind of leave this for another time. Yes. Okay.

Manu Sporny: Never mind. I didn't know there was a PR to get to. So, let's
get to that.

Patrick St-Louis: Yeah. This was just sort of making sure we're all on the
same page. Different people working on different things.
OpenAPI Specification Simplification

Patrick St-Louis: but it was not a necessary thing. so there's a big one
I'd like to take time to decide this one that's going to have impact on the
test suite. some people have already reviewed it. I have addressed some of
the comments. So this is about not the response bodies themselves but
whenever we want to refer as a credential in the spec whether that's an
input to an endpoint to be signed a response a verifiable credential a
workflow item.

Patrick St-Louis: So currently the way it's made is that we define what the
credential is in a verifiable credential very extensively in the OAS file.
and this includes a lot of normative statement about what the body of the
credential is. some of the things we discussed is we wanted to simplify
this in the same aspect that the data model spec already defines all of
these in its own spec and it's not the primary candidate we want to
highlight in the VCOM spec right we want to highlight the VC API the
interactions between parties and the security around this by kind of
copying over what's in the VC data model spec just adds

Patrick St-Louis: a lot of normative statement and a lot of things and then
we say yeah it's normative but it's only in description so we don't
actually enforce it so this is an attempt to address this so what it does
it only leaves the context and the type define and the spec so the spec
longer mentions fields like issuer credential subject groove move it only
defines the context and the type and it clearly explains the relationship
between this and the VC data model 2.0. this makes it that we primarily
focus here on verifiable credential and envelope verifiable credential and
presentations.

Patrick St-Louis: and we'll simplify a lot the OAS spec simplify testing
and yeah so I still kept the terms credential and verifiable credential
even though if they are exactly the same as a logic signaling that when you
have an issue endpoint you give a credential and when you have the response
you give a verifiable credential. They are essentially the same thing. it's
in the spec we can infer that well as a result of the issuance endpoint
you've secured the credential with a securing mechanism but we don't
require it to be either a data integrity or a envelope verifiable
credential I've also updated some things as coyote pointed out update to 1.0

Patrick St-Louis: So this was open it's been reviewed a little bit so if
people could have a look and that's like something else we could merge in
early on next week. yeah so this removes a lot of these fields because we
had issue is it VCDM 1.0 2.0 No. instead the spec just says your context
and your type will define what should be and the rest of the body. But the
VCOM spec itself will not test this it will not test to make sure that you
have a valid credential subject and so on. It will make sure you have a
context and a type. We may if we want to test it must be either a ential or
envelope verifiable credential.

Patrick St-Louis: But I would probably not even go there. it gets rid of
things like unsecured it add this thing called credential document which
both credential and verifiable credential with outline from and this can be
either a verifiable credential or verifiable red Same thing for
presentation. yeah so this is just a attempt at making this a little bit s
simpler to test u there is a section so this is probably the section that
will need the most review to make sure that the informative note that we
added here captures this well and is easy to understand.
OAuth 2.0 Authorization Flow ZCAPS Support in Specification

Patrick St-Louis: it's really to highlight that this open API specification
focuses on the endpoint request body and response mostly at the first layer
but not the semantic of the credential and presentation themselves and it
tries to explain how it relates to the VCDM 2.0 know spec. yeah. So, that's
pretty much what I wanted to present. so at least we had this discussion so
we can be in a good place to merge next week. I'm not going to merge until
next week. yes, Chaotic.

Kayode Ezike: Yeah, I just wanted to propose if we could quickly look at
656 and 657 just because they were also open for a week now. all so this
one is for basically adding more detail to the scope requirements for oath
20 authorization flow.

Patrick St-Louis: we can have a really quick view. there four minutes left,
but yes, so these are your PR. So I'll let you go over both try to make it
fairly quick.

Kayode Ezike: So what we had at the time you needed a little bit more
explanation as to what the format needed to be. and so did that and also
improve the examples that were provided. so it's gotten some review from
David and Ted and as an approval on it now. I'm not sure if we need anyone
else to review it, but it's just …

Patrick St-Louis: Is that it for this one? And this one. So the same thing
I assume similar kind of vein.

Kayode Ezike: yeah, difference is that there was no section at all for
authorization capabilities. So, I have to add a section for that because I
mean we do refer to it in this back and I think now that we're spinning up
the capability storage task force, we're more comfortable using these
things. And so added a new section for ZCAPS and basically defined what the
format of them should be and examples of what they would look like for
different endpoints. Kayode Ezike:

Patrick St-Louis: Okay, good. So,

Ted Thibodeau Jr: By this one, what do we mean? 656 or 662 the right one
we're talking about.

Kayode Ezike: No 656 for the first one. This is 657. Yeah. Yeah,…

Ted Thibodeau Jr: Okay. Thank you.

Manu Sporny: The only challenge with the Zcap one coyote is that we cannot
normatively reference it. and you took that into account. I take it. the 70
line 870 looks like a normative.

Kayode Ezike: that's right.

Patrick St-Louis: Yeah. …

Manu Sporny: It's going to create a normative reference is the only reason
I'm bringing it up.

Kayode Ezike: Even if I'm not using any of the verbs or…

Patrick St-Louis: I think required is a norative verb.

Kayode Ezike: verbs. Okay.

Patrick St-Louis: Yeah,…

Manu Sporny: And…

Manu Sporny: and if you just square bracketed, you have to use a question
mark to make it non-normative ref or you need to make mark the section as
informative. either one of those should do it. Just a heads up to you, what
to do. I think just look at the rendered version and see if it puts zcap in
the normative references section. And if it does then you have to prepend
it with It's a square bracket question mark and…

Kayode Ezike: Yeah. I think we also agreed last week that we would also
leave the OOTH one as non-normative…

Manu Sporny: then the ref correct.

Kayode Ezike: because of the confusion. So, I guess maybe I'll have to take
one more look at these two PRs then just to make sure. That's all.

Patrick St-Louis: I think this looks good. So, we'll make sure to lube back
this in next week along with thre model and the OS a simplification.
awesome.

Patrick St-Louis: Any closing thoughts anyone would like to share before we
adjourn the meeting? In that case, thank you very much for attending and
I'll see you all in a week. Meeting ended after 00:59:55 👋 This editable
transcript was computer generated and might contain errors. People can also
change the text after it was created.

This transcription was generated by a large language model (LLM) and might
contain errors. When in doubt, check the audio recording. This page was
formatted by scribe.perl <https://w3c.github.io/scribe2/scribedoc.html>
version 248 (Mon Oct 27 20:04:16 2025 UTC).

Received on Wednesday, 8 July 2026 01:17:51 UTC