The Scope Creep Automated reviews of new AI-related working group submissions

Update review · November 20, 2026

Revocation Closure and Authority Epochs for Agent Delegation Update

draft-watts-oauth-agent-revocation-closure -04View on Datatracker →

What changed

Nothing technical. The -04 revision changes the date, the expiry and the revision number.

The only other change is a mistake. Four informative references now name this draft as their own draft string.

The entries for Transaction Tokens, the WIMSE Architecture, AIMS and draft-chen-oauth-agent-revocation all read "draft-watts-oauth-agent-revocation-closure-04". In -03 they correctly read draft-ietf-oauth-transaction-tokens-11, draft-ietf-wimse-arch-08, draft-ietf-wimse-aims-00 and draft-chen-oauth-agent-revocation-01. The titles, authors and URLs are still right. This looks like a find-and-replace on the revision string that caught too much.

The change log wasn't updated either. The appendix is still titled "Changes since -02" and describes the -03 edits.

Was the previous feedback addressed?

No. Every open item from the -03 review is unchanged.

  • Cite Global Token Revocation and Shared Signals: not done. Neither Global Token Revocation nor OpenID Shared Signals and CAEP is mentioned.
  • Nothing to interoperate on: not done. The Closure Receipt still has no encoding. There's still no way to request closure status or deliver it.
  • Epoch identifier: not done. It's still "opaque to this specification", with no claim name and no way to carry it through a token exchange.
  • Cross-domain signaling: not done. Domain A still can't claim CLOSED without knowing domain B's state, and there's still no way for A to learn it.
  • Untestable MUSTs: not done.
  • State diagram vs. text on PARTIAL: not done. The diagram still shows PARTIAL as an endpoint. The text still says it applies "while closure work can continue".
  • Leftover cut-set wording: not done. The receipt still lists a "barrier/cut-set evaluation" digest.

What's new

Nothing. No mechanisms were added.

What was removed

Nothing.

So is it closer to something the WG can use?

No. This is a keep-alive revision, and it's slightly worse than -03 because of the broken references.

The -03 draft wasn't close to expiring. It was good until May 2027. There was no reason to publish -04 without content changes.

Author

Deonté Watts is still the sole author, still listed with GoodShyt Group Inc. No authors were added.

The draft has now had four revisions since October 11. The last two made no technical changes.

Questions worth raising

The questions from the last review all stand:

  • Would the epoch-inheritance rule work better as a contribution to draft-chen-oauth-agent-revocation than as a separate document?
  • How does an epoch identifier get carried across a token exchange?
  • How does domain A learn domain B's closure state?
  • Can PARTIAL or UNKNOWN move to CLOSED later?

One new one for the author: fix the four reference entries in -05 and update the change log.