Update review · November 10, 2026
Authorization Schemas -01, Three New Authors and No Technical Changes
The -01 is an author list change and an IANA cleanup. Sections 1 through 11 are word-for-word identical to -00.
Everything in our review of -00 still applies. None of that feedback was addressed.
What changed
- Three new co-authors. Yaron Sheffer, Yaron Zehavi and Suresh Krishnan join Dapeng Liu. All three were listed as contributors in -00.
schema_refis registered in the right place. -00 pointed at a "JWT Access Token JWT Claims" registry attributed to RFC 9068, which doesn't exist. -01 uses the JSON Web Token Claims registry from RFC 7519.- A new Token Introspection Response registration for
schema_ref. The draft already said the claim could appear in introspection responses. Now it's registered. - RFC 9068 is gone from the normative references. RFC 7519 replaces it.
- A GitHub repo and issue tracker. The new "About This Document" note points to
maxpassion/oauth-authorization-schemas. - Editorial. The appendices are now lettered A and B.
The page count went from 41 to 43. That's the front matter and the longer author block, not new content.
No mechanisms were added. No mechanisms were removed.
Was the previous feedback addressed?
No. Here's where each point stands in -01.
- No normative schema.
transaction_authorityis still a non-normative example and the registry still starts empty. - Cumulative limits with more than one RS. Section 7.2 still has each RS track usage from its own state, and the AS still "does not track runtime usage." A 1000 USD limit across two checkout APIs is still 2000 USD.
- Guardrails have no wire format. Nothing tells the AS that an
authorization_detailsobject is a proposed guardrail. - Subset semantics. Section 3.2 still says SHOULD. Section 9 still says MUST. Appendix B.3 still calls subset verification "unconditional."
- Error codes. Table 1 still maps nearly everything to
invalid_request.invalid_authorization_detailsfrom RFC 9396 still isn't mentioned. - Status codes. Figure 3 still returns
401withinsufficient_authorization. Section 7.3 still says403. - Unknown fields. The AS rejects them. The RS still "MUST ignore or reject" them.
- Too many identifiers.
work_id,work_ref,schema_ref,evidence_refandtransaction_idare all still there, with the same definitions. confirmation_required. The RS still tells the client that the AS will need user consent.- Mechanisms we suggested removing. Administrative guardrails, the four ending reasons,
evidence_ref, the new IANA registry andschema_refare all still in.
On schema_ref, the draft went the other way. We suggested dropping it, and -01 adds a second IANA registration for it.
So what does the new author list mean?
The interesting name is Yaron Zehavi. He's the author of RAR Metadata and Error Remediation, the adopted draft this one overlaps with most.
Our main structural concern was that schema_profile_uri and the schema profile document are a second remediation mechanism, parallel to authorization_remediation and the JSON Schema metadata in Zehavi's draft. Section 5.3 still calls the two "complementary," and its text hasn't changed.
The person who can resolve that overlap is now on the author list. The right outcome is still for reusable profiles to be an extension of the remediation draft, with one description format for agents to parse. That's the question to ask the authors.
Two small things to fix in the next revision:
- Yaron Sheffer appears twice, as an author with "Independent" and in the Contributors section with "Intuit."
- The informative reference to the authorization evidence draft points at -01. That draft is now at -02.
Background of the Authors
Datatracker couldn't be reached for this post. Liu, Zehavi and Krishnan are described from our earlier 2026 reviews. Sheffer hasn't appeared in a previous review, so his entry is from general knowledge and has no verified counts.
- Dapeng Liu (Alibaba): 3 RFCs from 2014 and 2015 in distributed mobility management, around 10 active drafts and 80+ expired ones. An experienced author who is writing a lot right now.
- Yaron Sheffer (Independent, new in -01): a long-time IETF security contributor with many published RFCs, including the TLS recommendations BCP and the JWT Best Current Practices. Not a new author.
- Yaron Zehavi (Raiffeisen Bank International, new in -01): 0 RFCs and around four active drafts, one of them adopted by the OAuth WG. One of the most active individual contributors in the WG right now.
- Suresh Krishnan (Cisco, new in -01): dozens of RFCs, former Internet Area AD and IAB member. He has recently been added to several agent authorization drafts, including Liu's authorization evidence draft and the actor chain documents.
The author list is much stronger than it was on -00. The document hasn't caught up to it yet.
What to do with this version
Don't re-read it. If you read -00, you've read -01.
The -00 feedback is all still open, and there's now an issue tracker to file it in. The three to start with are the multi-RS cumulative limit problem, the missing guardrail wire format, and merging the profile mechanism into RAR Metadata and Error Remediation.