Exactly, and to Brian's point:

> Looking at it from a different perspective, this could be represented as a 
> limited CWT claims set (to only allow specific claims) alongside the COSE key 
> which would separate the two concerns.

One could imagine a subset of CWT Claims, called COSE_Key_Usage, and a 
COSE_Key_With_Usage that would be a pair of (COSE_Key, COSE_Key_Usage), and 
then you could have a Set of them much like you can have a COSE_KeySet today, 
and OIDC-style workflows are exactly one of the places where this would come in 
handy.

Amaury

________________________________
From: [email protected] <[email protected]> on behalf of Ilari 
Liusvaara <[email protected]>
Sent: 23 February 2026 17:26
To: cose <[email protected]>
Subject: [COSE] Re: [EXTERNAL] Re: Key restriction metadata

On Mon, Feb 23, 2026 at 03:56:14PM +0000, Amaury Chamayou wrote:
> For what it's worth, we have a similar requirement in our
> implementation of SCITT, where the keys exposed by the service for
> the purpose of receipt verification are currently exposed as a
> COSE_KeySet.
>
> We currently use application-specific, custom tstr values in the
> COSE_Key, as allowed by RFC_9052, to describe the sequence number
> range that a given key is applicable to.
> It would be useful to have an interoperable way to describe
> constraints on key usage though, such as time range, sequence number
> range etc, whether it is embedded in the key, or travels next to it
> in a COSE_KeySetForPurpose.

That sounds like application-specific grab bag of attributes.

However, there are cases where that kind of thing is appropriate (e.g.,
attested workflow attributes for OIDC, as those are wildly dependent on
the runner).




-Ilari

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

Reply via email to