Intent to Ship: Web Authentication Immediate UI mode

125 views
Skip to first unread message

Ken Buchanan

unread,
Mar 5, 2026, 2:34:59 PM (11 days ago)  Mar 5
to blink-dev, Adem Derinel, Martin Kreichgauer
Contact emails
ke...@chromium.org ,  der...@google.com

Explainer
https://github.com/w3c/webauthn/wiki/Explainer:-WebAuthn-immediate-mediation

Specification
https://github.com/w3c/webauthn/pull/2291

Design docs

https://github.com/w3c/webauthn/wiki/Explainer:-WebAuthn-immediate-mediation

Summary
A new mode for navigator.credentials.get() that causes browser sign-in UI to be displayed to the user if there is a passkey or password for the site that is immediately known to the browser, or else rejects the promise with NotAllowedError if there is no such credential available. This allows the site to avoid showing a sign-in page if the browser can offer a choice of sign-in credentials that are likely to succeed, while still allowing a traditional sign-in page flow for cases where there are no such credentials.

Blink component
Blink>WebAuthentication

Web Feature ID
webauthn

Motivation
Most sign-in experiences on the web use sign-in pages that offer multiple options for accessing an account, such as username/password input fields, federated sign-in buttons, and sometimes explicit WebAuthn or passkey buttons. When the browser is aware of passkeys or passwords that the user has for the site, this API mode makes the sign-in page unnecessary by instead showing simple browser account selection UI when the user begins a sign-in attempt. Signing in with this flow reduces friction and avoids user confusion from having to remember which sign-in option they have used previously on a given site. The main difference between this and current modal WebAuthn sign-in UI is that for users without any such credentials, no browser UI will be shown, and their sign-in experience will be unchanged from what it is today (typically, a navigation to the site's sign-in page).

Initial public proposal
https://github.com/w3c/webauthn/issues/2228

TAG review
https://github.com/w3ctag/design-reviews/issues/1092

TAG has closed its review with unsatisfied on the basis that they do not believe a modal browser dialog is preferable to a form for user sign-in experiences. There was extensive discussion of this topic on both the TAG review issue and the WebAuthn WG issue.

This conflicts with the feedback we have received from developers of major relying parties who believe this enables them to build meaningfully better user experiences. They believe that a modal dialog that appears only when passkeys are available will be more successful for users attempting to sign in. Additionally, achieving the goal of signing in a user while keeping them in the current page context is very difficult with the current API.

Apple has stated that it supports the goals of this mode, but has objected on a different basis from TAG. It favors an alternative API form that it believes will have better privacy properties ( https://github.com/w3c/webauthn/issues/2228#issuecomment-3443764943 ).  Notably, Apple's proposal and Immediate mode would be invoked in different situations, and are not mutually exclusive.

Since Immediate mode does not guarantee that UI will be shown on invocation, we maintain flexibility to tweak this later in ways that limit its use.

TAG review status
Issues addressed

Origin Trial Name
Immediate Mediation for Passkeys and Passwords

Chromium Trial Name
WebAuthenticationImmediateGet

Origin Trial documentation link
https://github.com/w3c/webauthn/wiki/Explainer:-WebAuthn-immediate-mediation

WebFeature UseCounter name
kCredentialsGetImmediateMediationWithWebAuthnAndPasswords

Risks


Interoperability and Compatibility
Gecko : No signal ( https://github.com/mozilla/standards-positions/issues/1239 )

WebKit : Negative ( https://github.com/WebKit/standards-positions/issues/504 ) Feedback is on the WG issue:  https://github.com/w3c/webauthn/issues/2228#issuecomment-3443764943

Web developers : Positive ( https://github.com/w3c/webauthn/issues/2228#issuecomment-3999513181 )

Other signals :

WebView application risks

Does this intent deprecate or change behavior of existing APIs, such that it has potentially high risk for Android WebView-based applications?

No information provided


Debuggability
No information provided

Will this feature be supported on all six Blink platforms (Windows, Mac, Linux, ChromeOS, Android, and Android WebView)?
Yes

Is this feature fully tested by  web-platform-tests ?
Yes
https://wpt.fyi/results/webauthn/getcredential-ui-mode-immediate.https.html?label=master&label=experimental&aligned&q=getcredential-ui-mode-immediate.https.html

DevTrial instructions
https://docs.google.com/document/d/18iV5eUBM4NVoNx0gqPSxPyJAjPdrfIR75vcMDBewzZU/edit?tab=t.0#heading=h.uj0x12ysuohk

Flag name on about://flags
web-authentication-immediate-get

Finch feature name
WebAuthenticationImmediateGet

Rollout plan
Will ship enabled for all users

Requires code in //chrome?
True

Tracking bug
https://issues.chromium.org/issues/408002783

Launch bug
https://launch.corp.google.com/launch/4394539

Measurement
Use counters: CredentialsGetImmediateMediationWithWebAuthnAndPasswords CredentialsGetImmediateMediationWithWebAuthnOnly CredentialsGetImmediateMediationWithPasswordsOnly We are also tracking user interactions with the modal UI that will be shown when this is used.

Estimated milestones
Shipping on desktop 147
Origin trial desktop first 139
Origin trial desktop last 141
Origin trial extension 1 end milestone 144
DevTrial on desktop 136
Shipping on Android 147
DevTrial on Android 142
Shipping on WebView 147


Link to entry on the Chrome Platform Status
https://chromestatus.com/feature/5164322780872704?gate=5177075746734080

Links to previous Intent discussions
Intent to Prototype:  https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CALjHGKrQEs4TDzuzb%3D0B00S4OmkE4a1NbZGi19sCueTKvN_m9w%40mail.gmail.com
Ready for Trial:  https://groups.google.com/a/chromium.org/g/blink-dev/c/zC13ioLIZ_E/m/P-P6B6gNCQAJ
Intent to Experiment:  https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CALjHGKpJkA9G6De6D4%3DRNSbLMRdy8Yfa6B%3DgDNWeqTyHfv8sSg%40mail.gmail.com
Intent to Extend Experiment 1:  https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CALjHGKpLbqYVnSMfNgxh45TSbP9j6AU2JvLWow%3DH1ihr5v%2Bj0A%40mail.gmail.com


This intent message was generated by  Chrome Platform Status .

Alex Russell

unread,
Mar 9, 2026, 2:51:08 PM (7 days ago)  Mar 9
to blink-dev, Ken Buchanan, Adem Derinel, mart...@google.com, Jeffrey Yasskin, Jeffrey Yasskin
Hey Ken,

It's disappointing to hear that the TAG was trying to litigate specific UI choices, rather than helping to guide the API's design. As a general matter, we should not be specifying *any* explicit UI:


If we've made that mistake here, it's not too late to change course (although it's reasonable to leave non-normative "for instance" examples and Explainer illustrations). If we didn't, then I'm inclined to suggest that we have a conversation with our friends on the TAG about the opportunities for UI treatments that APIs provide vs. requirements. As I understand your API, an explicit form treatment is still possible in any UI that prefers that. Is that wrong?

Best,

Alex

On Thursday, March 5, 2026 at 11:34:59 AM UTC-8 Ken Buchanan wrote:

Yoav Weiss (@Shopify)

unread,
Mar 11, 2026, 11:21:30 AM (5 days ago)  Mar 11
to blink-dev, Alex Russell, Ken Buchanan, Adem Derinel, mart...@google.com, Jeffrey Yasskin, Jeffrey Yasskin
Naively reading through WebKit folks' feedback, they seem supportive of the use case, but interested in a different shape that won't expose the presence of the passkey to the site.
Is there a chance to converge here?

Ken Buchanan

unread,
Mar 11, 2026, 11:21:44 AM (5 days ago)  Mar 11
to Alex Russell, blink-dev, Adem Derinel, mart...@google.com, Jeffrey Yasskin, Jeffrey Yasskin
On Mon, Mar 9, 2026 at 2:51 PM Alex Russell < sligh...@chromium.org > wrote:
Hey Ken,

It's disappointing to hear that the TAG was trying to litigate specific UI choices, rather than helping to guide the API's design. As a general matter, we should not be specifying *any* explicit UI:

If we've made that mistake here, it's not too late to change course (although it's reasonable to leave non-normative "for instance" examples and Explainer illustrations). If we didn't, then I'm inclined to suggest that we have a conversation with our friends on the TAG about the opportunities for UI treatments that APIs provide vs. requirements.

Thanks for the response, Alex. We are not specifying the UI, other than that it must allow the user to select an existing passkey for sign-in. The objection stems from this API's functional behaviour, which is intended to enable sign-in via an in-context browser dialog rather than the site having to show a login form. TAG "disagree[s] that a browser dialog is a better user experience than an inline login form," but that is the goal of this API, not the actual thing being specified.

This feature allows developers to implement the following flow:
if (there is a passkey known by the UA to be available) then
  the UA shows sign-in UI offering that passkey
else
  reject the promise so the website can take some other action

The question is about whether this actually enables better user experiences, because in the current state the first part of that is available via autofill, but the 'else' part is not possible.

The discussion with TAG reached a point where they required a higher standard of evidence for user benefit than we can provide. We are satisfied with feedback from developers telling us how they intend to use this, and that this enables a lower-friction sign-in flow that makes it worth shipping.
As I understand your API, an explicit form treatment is still possible in any UI that prefers that. Is that wrong?

You are correct. This deliberately does not displace any existing WebAuthn usage. Invoking `navigator.credentials.get()` in this mode is not guaranteed to show any UI, so developers need to keep their existing sign-in options available for the cases in which the promise is rejected.

Best,

Alex

On Thursday, March 5, 2026 at 11:34:59 AM UTC-8 Ken Buchanan wrote:

Ken Buchanan

unread,
Mar 11, 2026, 11:36:14 AM (5 days ago)  Mar 11
to Yoav Weiss (@Shopify), blink-dev, Alex Russell, Adem Derinel, mart...@google.com, Jeffrey Yasskin, Jeffrey Yasskin
Hi Yoav,

That's correct. Apple agrees with the use case but dislikes the information leakage. The website can easily infer that browser UI is being shown, which lets them know a passkey exists for this user even if the user does not choose to sign in with one.

We spent some time exploring their alternative proposal. It ends up being something quite different, though, and we decided to continue with this based on partner feedback. Since in their case the promise would not be rejected if no passkeys are available, the website must offer an alternative on the page before the method is called. Much of the current design's value is that it allows the page to perform some action (such as a navigation) only if this API does not succeed.

The two proposals can co-exist in the specification, and we haven't ruled out pursuing Apple's alternative also. They would be invoked in different situations.

Alex Russell

unread,
2:54 PM (2 hours ago)  2:54 PM
to blink-dev, Ken Buchanan, blink-dev, Adem Derinel, mart...@google.com, Jeffrey Yasskin, Jeffrey Yasskin, Alex Russell
Thanks so much for the clarifications, Ken.

LGTM1

On Wednesday, March 11, 2026 at 8:21:44 AM UTC-7 Ken Buchanan wrote:
On Mon, Mar 9, 2026 at 2:51 PM Alex Russell < sligh...@chromium.org > wrote:
Hey Ken,

It's disappointing to hear that the TAG was trying to litigate specific UI choices, rather than helping to guide the API's design. As a general matter, we should not be specifying *any* explicit UI:

If we've made that mistake here, it's not too late to change course (although it's reasonable to leave non-normative "for instance" examples and Explainer illustrations). If we didn't, then I'm inclined to suggest that we have a conversation with our friends on the TAG about the opportunities for UI treatments that APIs provide vs. requirements.

Thanks for the response, Alex. We are not specifying the UI, other than that it must allow the user to select an existing passkey for sign-in. The objection stems from this API's functional behaviour, which is intended to enable sign-in via an in-context browser dialog rather than the site having to show a login form. TAG "disagree[s] that a browser dialog is a better user experience than an inline login form," but that is the goal of this API, not the actual thing being specified.

This feature allows developers to implement the following flow:
if (there is a passkey known by the UA to be available) then
  the UA shows sign-in UI offering that passkey
else
  reject the promise so the website can take some other action

The question is about whether this actually enables better user experiences, because in the current state the first part of that is available via autofill, but the 'else' part is not possible.

The discussion with TAG reached a point where they required a higher standard of evidence for user benefit than we can provide. We are satisfied with feedback from developers telling us how they intend to use this, and that this enables a lower-friction sign-in flow that makes it worth shipping.
As I understand your API, an explicit form treatment is still possible in any UI that prefers that. Is that wrong?

You are correct. This deliberately does not displace any existing WebAuthn usage. Invoking `navigator.credentials.get()` in this mode is not guaranteed to show any UI, so developers need to keep their existing sign-in options available for the cases in which the promise is rejected.

Best,

Alex

On Thursday, March 5, 2026 at 11:34:59 AM UTC-8 Ken Buchanan wrote:
Reply all
Reply to author
Forward
0 new messages