build-prototype

Build self-contained HTML prototypes and decision comparisons for product screens before implementation.

Updated Sep 17, 2026
One-click install
npx skills add https://github.com/hkw2028/toy-project --skill build-prototype-hkw2028
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: build-prototype
Source: https://github.com/hkw2028/toy-project/tree/main/.agents/skills/build-prototype
Command: npx skills add https://github.com/hkw2028/toy-project --skill build-prototype-hkw2028

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Teams often commit to product structure and visual decisions only after expensive implementation. This Skill lets you settle screen structure, interactions, and visual choices using a finished-looking, disposable HTML prototype before any production code is written. ## Core Features & Use Cases - Full-surface prototyping: Builds one self-contained HTML file containing every product screen with realistic dummy data, working interactions, state presets, and viewport simulation, grounded in the project's existing design tokens. - Decision comparisons: Renders up to three alternatives for a single contested visual or interaction decision in a temporary comparison file, then integrates the chosen option back into the canonical prototype. - Spec handoff: Once all screens are approved, records settled behavior, assumptions, deferrals, and risks in docs/specs/<slug>/spec.md and saves the approved prototype alongside it. - Use Case: Before building a mobile notification center for an existing dashboard app, generate a clickable prototype in the app's own design language, compare two layout options for the alert list, and hand off an approved spec to implementation. ## Quick Start Ask the agent to build a prototype of the screens you want to review, for example: build a prototype of the home, saved, and settings screens so I can review them before implementation.

Frequently Asked Questions about build-prototype

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

FAQPage Schema
How do I prototype product screens before implementation?

Describe the screens you want to review and the agent builds one self-contained HTML file containing every screen with realistic dummy data and working interactions. You review the surface in a browser, request changes, and approve it before any production code is written.

How do I compare two or three design alternatives for one decision?

The agent creates a temporary compare.html showing up to three alternatives for the specific contested decision, holding everything else fixed. After you choose, the winner is integrated into the main prototype and the comparison file is deleted.

Can the prototype use my real API or production data?

No. The prototype intentionally uses realistic dummy content and stays within a single-file, no-network constraint. Real data, APIs, latency, and production wiring are out of scope because they are not what the alignment review is about.

Does this work for CLI or terminal interface design?

No. Command-line, terminal, and voice interfaces are explicitly outside this Skill's scope, and it will not build an HTML mockup of a terminal. It covers screen-based web and mobile product surfaces only.

What happens after all prototype screens are approved?

The agent records settled behavior, acceptance criteria, assumptions, deferrals, and risks in docs/specs/<slug>/spec.md and saves the approved surface as a linked prototype.html. The prototype is a reference for implementation, not production code.

What is the difference between standard and high effort?

Both efforts pass the same base gate of finished screens, working interactions, and browser verification. High effort adds an independent visual convergence audit against an inspectable reference product, ideally delegated to a fresh reviewer subagent.