[Openid-specs-ab] OpenID Provider Commands - proposed WG specification

[Dick Hardt]

Hi Michael, I just saw this email, sorry for the delayed response ...

On Mon, Feb 17, 2025 at 9:12 PM Michael Schwartz via Openid-specs-ab <
openid-specs-ab at lists.openid.net> wrote:

> Dick,
>
> 1. I don't really see the relationship between SCIM and SAML. Gluu
> customers use SCIM to update all kinds of account information. Whether they
> use SAML or OpenID depends on the RP.  But I agree that more RPs support
> SAML in B2B websites... it just seems like a non-sequitur.
>

SCIM does provisioning & OP Commands does provisioning.
SCIM is a way to manage a resource. The users and groups are
independent, though often related, to SSO.

OP Commands allows an OP to manage accounts from the OP at an RP.


>
> 2. You could in fact make a SCIM extension to send this kind of
> information. For example, Gluu defined a FIDO SCIM extension because there
> was no way to get a list of passkeys for a user, or to delete a user's lost
> key.
>

 SCIM is based on the resource vs the OP data model


> 3. Are you intending to specify that the RP will expose a non-SCIM stable
> Internet facing backchannel web URL endpoint?
>

Yes


>
> 4. How would you protect this endpoint? SCIM left security out of scope.
> And thus SCIM client libraries might or might not work, depending on how
> they handle "OAuth" security.
>

This is one of the key advantages of OP Commands over SCIM. The commands
are signed the same way that an ID Token is signed. IE reuses the existing
JWKS endpoint. Does not need to use the same key as the ID Token, but
public key is found same way by following `iss` value


>
> 5. So your solution does not want to support mobile or browser based
> clients?
>

If there is a userpool, it is in a server.


>
> 6. Per my post
> <https://www.linkedin.com/posts/nynymike_token-status-list-activity-7295937395122655233-F7Rs?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAArUy4Bb5Ha4b5n1mmyBhevew7nxXkSV14>
> on Linkedin, I still think a pull based solution (i.e. based on OAuth
> Status List) would be more lightweight. If you "push" messages, you will
> need an RP endpoint or a long lived connection. Your spec could accomplish
> all its goals by publishing a new kind of "Account JWT", and enabling the
> RP to check the status of the JWT to see if anything changed. The advantage
> of this is that the RP can get an update for **all** account tokens issued
> by an OP at one time.  Plus the status token is very small. Once the RP is
> aware of the change, then it can interact with the OP, perhaps through an
> authorization request to get updated userinfo. Also, no need to deal with
> how to protect this very sensitive RP endpoint. One more thought... if
> Google has a billion users, as an RP do I really need to know about all the
> accounts that got compromised, even if they logged into my system only once
> in the past? I'd rather just check the account status when I see the person
> again.
>

I think that is significantly more complicated. :)


>
> So I'm agreeing with your idea, just not with your approach on how to
> solve it :-)
>
> - Mike
>
> --------------------------------------
> Michael Schwartz
> Gluu
> Founder/CEO
> mike at gluu.org
> https://www.linkedin.com/in/nynymike
>
>
> ------------------------------
> *CONFIDENTIALITY NOTICE*
> This message may contain confidential or legally privileged information.
> If you are not the intended recipient, please immediately advise the
> sender by reply e-mail that you received this message, and delete this
> e-mail from your system.
> Thank you for your cooperation
> _______________________________________________
> Openid-specs-ab mailing list
> Openid-specs-ab at lists.openid.net
> https://lists.openid.net/mailman/listinfo/openid-specs-ab
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-ab/attachments/20250302/83c77d30/attachment.htm>