What problem does it solve? Exploring visual directions for a UI screen usually means committing to one idea before seeing alternatives. This Skill generates several distinct AI design mockups in parallel, opens them side-by-side in a comparison board, and captures structured feedback so you can pick a direction before writing any frontend code. ## Core Features & Use Cases - Parallel Variant Generation: Launch multiple design directions at once via subagents, each with its own brief, quality check, and retry logic. - Comparison Board with Feedback Loop: Serve an HTML board over HTTP where you review variants side-by-side and submit structured feedback to iterate. - Taste Memory: Reads previously approved designs to bias new generations toward your demonstrated aesthetic preferences. - Evolve Mode: Screenshot a running local page and generate improvement variants when you don't like how something currently looks. - Use Case: You're building a settings page and unsure of the layout. Run the skill, get 3 distinct visual directions in about a minute, compare them in your browser, and iterate on the winner with feedback like "more whitespace, darker sidebar". ## Quick Start Ask the AI to run /design-shotgun to explore design options for the screen you are working on.