Review · November 10, 2026
OAuth User Authorization Management URI
The Problem
Most authorization servers have a page where the user can see which apps have access to their account and revoke them. A client has no standard way to find that page.
When the AS belongs to one well-known service, this doesn't matter. Users know where Google's account settings are.
It breaks down when clients and authorization servers are many to many. The draft calls out the decentralized social web. Think of a Mastodon or Bluesky client talking to thousands of independently run servers, each with its own settings URL. The client can't show a "manage this app's access" link because it doesn't know where to point it.
How It Fits In
This is a single new AS metadata parameter, user_authorization_management_uri. It extends RFC 8414 and depends on nothing else beyond RFC 6749. It fits the working group charter.
RFC 8414 already carries human-facing URLs such as service_documentation, op_policy_uri and op_tos_uri. This is the same kind of parameter.
It creates no new roles, endpoints, or protocol messages. The management page already exists at most authorization servers. The draft only makes it discoverable.
So how does it relate to revocation? Token Revocation lets a client revoke its own tokens. This draft sends the user to a page where they manage every client's access themselves.
Summary
The AS publishes a URL in its metadata. The client reads it and shows the user a link. The user lands on a page at the AS, signs in if needed, and manages their authorized clients there.
The draft leaves the page itself out of scope. It recommends that the page shows all authorized clients, and it mentions revoking clients or sessions and reviewing security history as examples.
There is one normative requirement. The client MUST NOT open the URL in a webview or embedded browser. It has to hand off to the user's browser so the client never sees the user's credentials.
Interoperability
The idea is simple enough that two implementations will mostly agree. The problem is that the draft never formally defines the parameter.
Gaps worth raising
- No parameter definition in the body. The parameter is only described in the IANA section. There is no statement that it's OPTIONAL, no data type, no
httpsrequirement, and no example metadata document. RFC 8414 parameters all get this treatment. - What the page contains is undefined. It could be an authorized apps list, a general security settings page, or an account home page. A client can't write accurate link text like "Revoke this app's access" without knowing what the user will find there.
- No way to identify the client or the user. The client can't deep-link to its own grant, and it can't hint which account is involved. A user signed in to a different account in the browser lands on the wrong account's page, and nothing tells them. An optional
client_idorlogin_hintquery parameter would fix this, or the draft should say it's a deliberate omission. - The browser requirement is ambiguous. "Delegate to the user's default browser" could exclude in-app browser tabs like ASWebAuthenticationSession and Android Custom Tabs. OAuth for Native Apps (RFC 8252) explicitly allows those. The draft should cite RFC 8252 and use its terms.
- No guidance on what the client does afterward. If the user revokes the client on that page, the client finds out when its next request fails. A sentence telling clients to expect this would help.
Mechanisms that could be removed
Nothing. There is only one mechanism.
The recommendation about what the UI should show is not needed for interoperability as written. It should either become a clear minimum ("the page MUST let the user view and revoke authorized clients") or be dropped.
Writing Quality
The draft is very short. The body is about one page, and it's too thin for a Standards Track document. Section 3 is titled "Authorization Management UI" and describes the page in general terms, then mentions the parameter in a single sentence.
The language is mostly informal. Only one requirement uses a BCP 14 keyword. The rest are lowercase "should" and "may", including the statement that the AS should authenticate the user.
A few sentences need editing. "When you don't have a primary service for the Authorization Server" is unclear and uses second person. "However, recommends that" is missing a subject. "An URI" should be "a URI".
There are no informative references. RFC 8252 and RFC 7009 are the obvious ones to add. There is no privacy section, though there's little to say beyond the account mix-up case above.
This is a rename of an earlier draft, draft-emelia-oauth-authorization-management-uri. The change added "user" to the parameter name and nothing else.
Author
Emelia Smith (no affiliation listed). No published RFCs that I'm aware of. She co-authors the adopted Client ID Metadata Document draft, and has one other individual draft in this working group, AS Metadata Parameter for Requiring DPoP. A newer author who is active in the working group and comes from the Mastodon and fediverse community, which is where this use case comes from.
These details come from the local document archive. The Datatracker profile could not be fetched for this review, so the counts are not confirmed.
Questions for the Working Group
So what should the working group decide?
- Is a human-facing settings link worth an RFC, or is a direct IANA registration from another document enough? The registry only requires Specification Required.
- Should the draft set a minimum for the page, such as viewing and revoking authorized clients?
- Should the client be able to pass
client_idor a user hint so the user lands on the right grant in the right account? - Does the browser requirement match RFC 8252, including in-app browser tabs?