design-shotgun

Generate multiple visual design variants and assemble an HTML comparison board.

Updated Mar 29, 2026
One-click install
npx skills add https://github.com/ibalasite/gstack --skill design-shotgun-ibalasite
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: design-shotgun
Source: https://github.com/ibalasite/gstack/tree/main/design-shotgun
Command: npx skills add https://github.com/ibalasite/gstack --skill design-shotgun-ibalasite

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Rapidly explore multiple visual directions without manual mockup toil by generating several AI-produced design variants, presenting them side-by-side, collecting structured feedback, and iterating until a direction is approved.

Core Features & Use Cases

  • Parallel variant generation: Launch multiple generation agents in parallel to produce distinct visual directions and retry on rate limits.
  • Comparison board & feedback loop: Auto-build a served HTML comparison board, collect feedback files, and poll for choices or remix requests.
  • Taste memory & evolution: Read prior approved designs to bias new variants, and evolve from screenshots when improving an existing UI.
  • Persisted artifacts: Save all mockups and approval metadata to the user's global design store so designs persist across projects and branches.
  • Use cases: visual brainstorming for a new landing page, iterating on a checkout UI from a live screenshot, or generating style directions for a component library.

Quick Start

Ask the skill to generate three distinct design directions by saying: /design-shotgun generate 3 variants for the checkout screen including persona, constraints, and target platforms.

Frequently Asked Questions about design-shotgun

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate multiple UI design variants for visual comparison?

You generate UI design variants by launching parallel agents that produce distinct mockups and assemble them into a side-by-side comparison board. This enables rapid visual brainstorming without manual mockup toil.

What is the best way to collect feedback on AI-generated mockups?

The best way to collect feedback on AI-generated mockups is to auto-build a served HTML comparison board. The system polls for feedback files, collects structured choices, and handles remix requests to iterate until a direction is approved.

Can I iterate on an existing UI screenshot to evolve the design?

Yes, you can evolve an existing UI from a live screenshot. The system uses taste memory to read prior approved designs, biasing new variants to iteratively improve the visual direction based on that screenshot.

How do I persist AI design mockups across different projects?

You persist AI design mockups by saving all artifacts and approval metadata to a global design store. This ensures your generated visual designs and feedback history remain accessible across different projects and branches.

Does parallel variant generation handle API rate limits?

Yes, parallel variant generation handles API rate limits automatically. The execution logic includes built-in retry mechanisms for generation agents, ensuring all distinct visual directions are produced despite rate limiting.

What do I need to generate distinct visual directions for a checkout screen?

To generate distinct visual directions for a checkout screen, you need to provide a persona, design constraints, and target platforms. The system then launches parallel agents to produce the requested mockup variants for comparison.