Intent to Prototype: Agentic Login Toolkit

[Chromestatus]

Contact emails

go...@google.com

Explainer

https://github.com/samuelgoto/agentic-login-toolkit

Specification

No information provided

Summary

A series of annotations that a website developer can introduce to their pages to help agentic browsers log users in.

Blink component

Blink>Identity>FedCM

Web Feature ID

fedcm

Motivation

In LLM-powered Agentic browsers, many user journeys involve logging in to websites. As with most of the LLM-powered actuation, the LLM has a baseline understanding using statistical models that allows it to click on links and fill forms to assist the user through the process. However, as much as user agents are and should develop as many heuristics as possible to retrofit into the existing content on the Web, heuristics are, by design, unreliable and are expected to have lower precision and quality than structured content that is opted-into by website owners. Not every website developer will have the incentives (expertise or demand) to annotate their content to be accessed by assistive browsers, but for those that do, what’s the best way that they can annotate their page to make user agents (e.g. browsers and search engines) better aware of their login flows?

Initial public proposal

https://github.com/samuelgoto/agentic-login-toolkit

Requires code in //chrome?

False

Tracking bug

https://b.corp.google.com/issues/473826292

Estimated milestones

No milestones specified

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5093444592730112?gate=5156788951449600

This intent message was generated by Chrome Platform Status.