prompt-interaction-design

To design interactive prompts in VS Code, use the ask_questions tool to build guided, wizard-like components. This approach gathers structured user input, preferences, or multi-step decisions directly within the editor environment.

Reusable patterns for structuring multi-step agent prompts include Pre-Implementation Gathering for scoping, Feature Selection for capability choices, Named Input for custom values, and Free Text for open-ended user responses.

Yes, you can use VS Code widgets to batch questions before implementation. The Pre-Implementation Gathering pattern collects multiple user inputs at once to define project scope and preferences before any code is written.

Interactive prompt design supports both free text and multi-select inputs. You can capture open-ended user input with the Free Text pattern or use the Feature Selection pattern for single-select and multi-select capability choices.

Document interaction rules for structured agent prompts by specifying frontmatter metadata in SKILL.md. Include the name and description, and optionally reference supporting scripts, references, or assets alongside example components.

The best way to capture custom user preferences in a VS Code wizard is applying the Named Input pattern. It collects user-provided names or custom values, offering optional freeform input when preset selections are insufficient.