- From: Atul Tulshibagwale <atul@sgnl.ai>
- Date: Tue, 22 Jul 2025 13:52:05 -0700
- To: Rory Hewitt <rory.hewitt@gmail.com>
- Cc: Amos Jeffries <squid3@treenet.co.nz>, ietf-http-wg@w3.org
- Message-ID: <CANtBS9dj6nN00+X6-jcve58bqV0wV_OF+GYQP7Up_m9C4Zo5Pg@mail.gmail.com>
Hi Rory and Amos, I see these relevant headers in the HTTP field names registry: - Authentication-Control - Authentication-Info - Authorization Of these, the "Authentication-Info" field seems to be a HTTP Response field (as seen in this description <https://www.rfc-editor.org/rfc/rfc9110.html#section-11.6.3> of the field). The "Authentication-Control" field also appears to be a HTTP Response field (see description here <https://www.rfc-editor.org/rfc/rfc8053.html#section-4>). The Authorization header cannot be used because it needs to be kept available for service-to-service authorization such as SPIFFE. The TraTs spec clarifies this here <https://www.ietf.org/archive/id/draft-ietf-oauth-transaction-tokens-05.html#section-8> . This is why we thought we would need a new header field. Thanks for your feedback and review, Atul On Tue, Jul 22, 2025 at 9:16 AM Rory Hewitt <rory.hewitt@gmail.com> wrote: > Is there a benefit to creating a specific new header for JWTs? > > I would suggest either passing them in an Authentication header (commonly > currently used by some apps) or the application which needs them can define > its own header (also common). > > On Tue, Jul 22, 2025 at 8:24 AM Amos Jeffries <squid3@treenet.co.nz> > wrote: > >> On 22/07/25 06:28, Atul Tulshibagwale wrote: >> > Hello, >> > We are currently working on a draft for Transaction Tokens <https:// >> > datatracker.ietf.org/doc/draft-ietf-oauth-transaction-tokens/>, which >> > envisions a new HTTP Request Header called "Txn-Token" <https:// >> > www.ietf.org/archive/id/draft-ietf-oauth-transaction- >> > tokens-05.html#name-txn-token-http-header>. The header value is >> expected >> > to be a JWT. >> >> >> Taking a brief looks at the document ... >> >> > 2.1. What are Transaction Tokens? >> > >> > Txn-Tokens are short-lived, signed JWTs [RFC7519] that assert the >> > identity of a user or a workload and assert an authorization context. >> >> >> So, if I am reading that correctly these are a cross between login >> credentials and a session ID. >> >> >> I am wondering why these credentials are using a custom header instead >> of being sent as part of HTTP Authentication (request) and >> Authentication-Info (response) headers. >> >> There is a lot of HTTP security behaviour that can be leveraged just by >> using the Authn headers instead of re-inventing the wheel. >> >> >> Cheers >> Amos >> >> > > -- > Rory Hewitt > > https://www.linkedin.com/in/roryhewitt >
Received on Tuesday, 22 July 2025 20:52:26 UTC