The clue model

A ViewCue clue has three parts:

meaning + target + moment

Meaning

A short message in human language. It should explain what matters now, not repeat a tool schema or expose raw data.

Target

A stable semantic target such as payments.method or checkout.total. The host application declares the target and supplies its label and description.

Moment

The application decides when the clue is useful:

• a tool is explaining a term;
• a result has changed the visible page;
• a validation message needs attention;
• the agent is asking the person to review a choice;
• an optional activity state should be shown.

The last case can use a lifecycle trace, but a lifecycle is only one possible moment for a clue.

A good clue is modest

ViewCue should not cover the page, trap focus, auto-submit a form, or claim that a payment happened when the host application has only prepared a simulation. The person should still understand the underlying application and remain able to interact with it.