Suggestion-card output
Agent output as a card
Suggestion-card output is useful when an agent finds one concrete next step.
A card should not be a long report. It should be an action that a user can review and complete.

Review source gap on comparison prompts
Competitors are named from review pages where your brand is missing. Inspect the prompt evidence and decide whether this needs a page update or source review.
Context to include
Give the agent the evidence source, the decision rule, and the action format.
For example, tell it to inspect prompt results, use source analytics, and create one card only when the action is clear.
Output to expect
Expect a short title, useful reasoning, supporting evidence, and a clear CTA.
The CTA should point to the place where the user can act, such as prompts, analytics, suggestions, calendar, sitemap, or page writer.
Review the card before acting
Check whether the card is based on real workspace evidence. Archive cards that are vague, duplicated, or not worth doing.
Cards are valuable when they keep the action queue focused.
Evidence
Name the prompt, source, metric, or file behind the card.
Issue
Explain the problem in simple words.
CTA
Point to one next dashboard view or action.
Review
Let the user accept, ask, complete, or archive.