Calling a dashboard a command center supplies a different description without explaining what changes. The schedule preset adds a practical reason for publishing on Tuesday. Those passages deserve different review notes even though both contain a contrast.
The row-count preset includes its own synthetic manifest. The model can identify that the sentence uses evidence to correct a number. It cannot authenticate the manifest or establish that the real export contains 480 rows.
Use the result as an editorial observation. The page does not rewrite the paragraph or automatically delete contrasting phrases; context and house style still determine the edit.
The API-collected roundup links Sniff Test. Its pinned rules define the text distinction adapted here. This original Choice demonstration uses synthetic inputs and adds explicit missing-context handling; the source linter uses its own yes-or-no judgments. It does not reproduce that application or validate the truth of a claim.