Thanks, Rifaat.  See my comments below prefixed by “Mike2>”.

From: Rifaat Shekh-Yusef <[email protected]>
Sent: Monday, March 2, 2026 6:47 AM
To: Michael Jones <[email protected]>
Cc: oauth <[email protected]>
Subject: Re: [OAUTH-WG] Shepherd Review - Updates to JWT Client Authentication 
and Assertion-Based Authorization Grants

See my comments below.


On Sun, Mar 1, 2026 at 10:40 PM Michael Jones 
<[email protected]<mailto:[email protected]>> wrote:
Thanks for your review, Rifaat.  Replies inline prefixed by “Mike>”.

From: Rifaat Shekh-Yusef 
<[email protected]<mailto:[email protected]>>
Sent: Monday, February 23, 2026 11:32 AM
To: oauth <[email protected]<mailto:[email protected]>>
Subject: [OAUTH-WG] Shepherd Review - Updates to JWT Client Authentication and 
Assertion-Based Authorization Grants

Hi,

As the shepherd for this document, I have reviewed version 05 of the draft
https://www.ietf.org/archive/id/draft-ietf-oauth-rfc7523bis-05.html

and I have the following comments/questions:

Section 3
It is RECOMMENDED that SAML Bearer Assertions not be used for client 
authentication.

Should the RECOMMENDED be a MUST? If not, can you add some text to explain when 
SAML Bearer Assertions could still be used?
Mike> How about we change it to say:
It is RECOMMENDED that SAML Bearer Assertions not be used for client 
authentication for any new applications.  (The authors are not actually aware 
of any applications using this feature of RFC 7522.)  Should any applications 
already be doing this in the manner described in RFC 7522, it is left to the 
discretion of their implementers and deployers whether to migrate away from 
this feature and/or potentially tighten the audience values used in a manner 
parallel to the changes being made in RFC 7523.

If you want new applications to not use SAML Bearer Assertions, then I think 
RECOMMENDED should be a MUST; otherwise, you need to explain when SAML Bearer 
Assertions can still be used.

Mike2> OK – https://github.com/oauth-wg/draft-ietf-oauth-rfc7523bis/pull/25 
updated to use MUST NOT.  Please review.
Section 5, Second paragraph,
“The paragraph describing the audience value in Section 2”

You might want to explicitly state which paragraph this is referring to.
Mike> How about we change it to say:
The last paragraph of Section 2 of [RFC9126] , which describes the audience 
value, is replaced by:

Yep


Section 5, Last paragraph,
“Client authentication JWTs SHOULD be explicitly…”

Can you elaborate on what this is a “SHOULD” to make it clear to the 
implementer?
Mike> The previous paragraph, beginning “The introduction of strong typing for 
JWTs” already gives ample reasons why this should be done but is not required.  
I don’t believe any additional text is needed in this case.
If I understood the previous paragraph, it implies that you can still not use 
strong typing for backwards compatibility reasons.
If this is the reason for the RECOMMENDED and SHOULD in this section, then it 
would be great if you can be explicit about it.

Mike2> That same paragraph already contains the rationale for why this is 
RECOMMENDED:
“Since strong typing alone does not prevent the attacks described in 
[private_key_jwt.Disclosure<https://drafts.oauth.net/draft-ietf-oauth-rfc7523bis/draft-ietf-oauth-rfc7523bis.html#private_key_jwt.Disclosure>]
 and 
[Audience.Injection<https://drafts.oauth.net/draft-ietf-oauth-rfc7523bis/draft-ietf-oauth-rfc7523bis.html#Audience.Injection>],
 the use of explicit typing is RECOMMENDED for clients, enabling them to signal 
their intention of sending a JWT conforming to the requirements herein.  This 
approach balances security signaling with practical deployment considerations, 
avoiding disruption to client deployments that already conform to the tightened 
audience requirements but have not yet adopted explicit typing.”

Mike2> Explicit typing can be relied on for new deployments, since new clients 
will follow the recommendation.  Let us know if there is a specific additional 
wording change that you recommend.
Section 8.2

It seems to me that the following references should be moved to the Normative 
References section:
IANA.MediaTypes, IANA.OAuthParameters, OpenID.Core, RFC2046, and RFC6838
I agree with you about OpenID.Core because implementers need to use normative 
definitions contained in it.  The other references are only used to inform the 
IANA registrations – not implementers, and so need not be normative.  It’s not 
customary to make such references normative.  (Of course, if the IESG 
disagrees, we can always do this later.)

Yeah, this makes sense to me.

Regards,
 Rifaat


Regards,
 Rifaat

I will create a PR applying these proposed changes later this evening.

                                                                Thanks again,
                                                                -- Mike

_______________________________________________
OAuth mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to