[Openid-specs-native-apps] Please review specs

Fri Feb 14 15:29:24 UTC 2014

The connect spec was modified to allow for it but it is not explicitly called out anyplace as how you do 3rd party id_tokens.

In Core Sec 2
We modified "aud" from a single value to an array:
aud
REQUIRED. Audience(s) that this ID Token is intended for. It MUST contain the OAuth 2.0 client_id of the Relying Party as an audience value. It MAY also contain identifiers for other audiences. In the general case, the aud value is an array of case sensitive strings. In the common special case when there is one audience, the aud value MAY be a single case sensitive string.

We added "azp" to identify the party to whom the token was issued:

azp
OPTIONAL. Authorized party - the party to which the ID Token was issued. If present, it MUST contain the OAuth 2.0 Client ID of this party. This Claim is only needed when the ID Token has a single audience value and that audience is different than the authorized party. It MAY be included even when the authorized party is the same as the sole audience. The azp value is a case sensitive string containing a StringOrURI value.

That way an id_token can identify who issued it "iss", to what client was it issued "azp" and who are the valid audiences.

In Core Sec 3.1.3.7  for id_token validation by a client the rules state:

The Client MUST validate that the aud (audience) Claim contains its client_id value registered at the Issuer identified by the iss (issuer) Claim as an audience. The aud (audience) Claim MAY contain an array with more than one element. The ID Token MUST be rejected if the ID Token does not list the Client as a valid audience, or if it contains additional audiences not trusted by the Client.
If the ID Token contains multiple audiences, the Client SHOULD verify that an azp Claim is present.
If an azp (authorized party) Claim is present, the Client SHOULD verify that its client_id is the Claim Value.

We also took out a prohibition on returning a id_token from the token endpoint in response to a "refresh_token" grant_type.

Google presented the use case for third party id_tokens relatively late in the process,  so we allowed for them to be fully defined in an extension like napps.  
The important thing we did was include the processing rule about multiple audiences and "azp" in the core spec so that clients who don't know about third party id_tokens won't blindly accept them if they have someone else as the "azp" even if they are one of the audiences. 

That's why the Connect spec is silent on how to request them and all of the possible additional semantics.

I think it is up to us to define the missing parts.

John B.

On Feb 14, 2014, at 11:26 AM, Paul Madsen <paul.madsen at gmail.com> wrote:

> John, can you point me to the relevant piece of Connect that enables these 3rd party id_tokens?
> 
> thanks
> 
> paul
> On 2/12/14, 3:39 PM, John Bradley wrote:
>> Hi Chuck,
>> 
>> I will get to this over the next couple of days. 
>> 
>> We do have the 3rd party id_tokens that can be used as JWT assertions that were added to connect for Google.  In principal those should be exchanged in the assertion flow for access tokens when crossing security domains.
>> 
>> So I suppose the type of token would depend on if the app directly accepted access tokens from the AS of the napps agent.
>> 
>> Apps using Google Play services directly use the id_token as a access token in general but that places a potential burden on the RS to accept tokens of different types.   I prefer to use the token endpoint to exchange the assertion so the RS only needs to worry about access tokens from it's AS whatever those happen to be.
>> 
>> John B.
>> 
>> On Feb 5, 2014, at 11:48 PM, Chuck Mortimore <cmortimore at salesforce.com> wrote:
>> 
>>> One other thought  - Perhaps instead of opaque access tokens for the apps, we should issue id_tokens that are audience restricted 
>>> 
>>> 
>>> On Wed, Feb 5, 2014 at 3:58 PM, Chuck Mortimore <cmortimore at salesforce.com> wrote:
>>> Comments on Agent Core 1.0
>>> 
>>> 5.0 - Do we need to make client credentials mandatory?   Can we make this a MAY?
>>> 
>>> 7.1 - in general seems redundant to oauth/openid connect, with the exception of the AZA scope.  Do we need to respecify all of this?
>>> 
>>> 7.1.1 - Why is response_type=code MUST?  Is this oauth carry over?  (same as my question on 5.0 I think)
>>> 
>>> 7.4.1/2 - By issuing on the token endpoint, we are basically saying that only administrative authorization models will work.  If end-user authorized oauth is being used, the user doesn't have a chance to approve access to and new app.    Shouldn't we be performing a new Authorization request, rather than a straight refresh token exchange?
>>> 
>>> 
>>> Comments on Agent API bindings 1.0
>>> 
>>> 2.0 - "Rather than the user individually authenticating and authorizing each native application, they do so only for the authorization agent"  - same as my last comment; from an authorization model perspective, this basically kills off end-user approval models with this profile.   There's no way for the user to make effective authorization decisions for future unknown applications.   
>>> 
>>> 4.0 - this seems to really be the meat of what we should specify, but the entire section is basically silent on detail.   For this spec to be successful, shouldn't we take a stand and actually specify interaction patterns?
>>> 
>>> 4.1 - "The TA MUST NOT deliver a secondary access token to an application for which it was not issued." seems at odds with the rest of this section.   For example, the custom scheme approach would potentially violate this on iOS.  I'm not certain there is a reliable way not to violate this when supporting an TA intiated flow.
>>> 
>>> 4.2 - We should really spec out a Native App intiated flow.  It may be the only way we can reliably handle the security contraint in section 4.1.    One option could be to issue a public key with the authorization request and then encrypt the use JWE to responds, so if the Native app's custom scheme url were hijacked, the returned token wouldn't bleed to the wrong app.
>>> 
>>> 
>>> 
>>> 
>>> 
>>> On Wed, Feb 5, 2014 at 1:18 PM, Paul Madsen <paul.madsen at gmail.com> wrote:
>>> Both core & bindings are available at
>>> 
>>> http://hg.openid.net/napps/wiki/Home
>>> 
>>> John has some editorial fixes to make but is hoping to combine with those with any more normative changes
>>> 
>>> Our next call is Wed feb 19 @ 6 pm EST
>>> 
>>> Paul
>>> 
>>> _______________________________________________
>>> Openid-specs-native-apps mailing list
>>> Openid-specs-native-apps at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-specs-native-apps
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Openid-specs-native-apps mailing list
>>> Openid-specs-native-apps at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-specs-native-apps
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-native-apps/attachments/20140214/a4cc40a4/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4507 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-specs-native-apps/attachments/20140214/a4cc40a4/attachment-0001.p7s>