amvcp-choice-tables

You can capture structured user selections by rendering an interactive single-choice or multi-choice form table in the browser. This replaces error-prone chat typing by injecting controls that return a deterministic, typed selection payload back to the agent.

To set up agent selection, create a root table marked with data-ve-type="table-form" and data-ve-mode set to "single" or "multi". Define rows with data-ve-row-id and data-ve-row-label, then the runtime injects radio or checkbox controls automatically.

Yes, you can include a free-text "Other" option by adding a row marked with data-ve-row-text. This captures typed text alongside the structured selections, allowing users to submit custom responses within the same table-form payload contract.

The best way to get a deterministic payload is using a structured table form with typed row data. By defining data-ve-row-id attributes, the runtime ensures the agent receives exactly the selected row data without parsing ambiguous text inputs.

No external dependencies are required to render the interactive selection controls. The Skill operates independently using scripts and references to inject radio buttons or checkboxes into the table markup and manage the submission workflow.

You should use a table form instead of chat input when you need exact, structured selections for scenarios like strategy comparison or voting. It prevents typing errors and ensures the agent receives a clean, typed payload of the selected rows.