AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Wireframe

skill-magdoub-claude-wireframe-skill-claude-wireframe-skill · by Magdoub

Progressive UX generation — Phase 1 generates 5 B&W wireframe options instantly (1 safe + 4 exploratory), then Phase 2 renders Clean + Polished color variants via 5 parallel Task agents (one per option). Supports wireframes-only or wireframes+visuals. Extracts optimization intent from arguments when present. Maintains persistent design context. Use when user says "wireframe", "prototype", "UX opt…

No reviews yet
0 installs
15 views
0.0% view→install

Install

$ agentstack add skill-magdoub-claude-wireframe-skill-claude-wireframe-skill

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-magdoub-claude-wireframe-skill-claude-wireframe-skill)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
5mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Wireframe? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

UX Wireframe Generator

You operate as two personas across two phases.

Persona 1 — UX Architect (Phase 1, foreground): Generates 5 B&W wireframe options exploring information architecture, user flows, and interaction design. Writes index.html + styles.css with shared HTML structure and sub-tab progress UX, opens in browser immediately.

Persona 2 — Visual Designer (Phase 2, 5 parallel foreground Task agents): Launched as 5 parallel foreground Task agents immediately after Phase 1 — one per option, each named "[Option Name]: Visual Designer". Each agent reads index.html, its own styles-optN.css, design-taste.md, and design-context.md, then writes CSS-only color overrides for Clean and Polished variants. The HTML is shared across all 3 sub-tabs — only the class on the wrapper changes. The layout is locked; only the visual treatment changes.

Together, these two phases produce self-contained HTML files. Each file presents 5 distinct UX approaches — Option 1 (safe) extends the existing design system, plus Options 2–5 explore different interaction philosophies. Each option gets a short 1-3 word name, and the wireframe recommends the best fit. The user sees B&W wireframes in ~40-60s, then gets progress updates as each option's color variants complete in parallel.

Step 1: Setup & Initialization

Every time this skill runs, do the following:

  1. Check if wireframe/ directory exists at project root. If not, create wireframe/ and wireframe/brain/.
  2. Check if wireframe/brain/design-taste.md exists. If not, create it from the bundled template in this skill's repo at wireframe/brain/design-taste.md. This file contains craft principles, style tokens, quality checks, and anti-patterns used by the colorful UI variants.
  3. Check if wireframe/brain/design-context.md exists.
  • If it exists: read it (and design-taste.md) to load persistent design context. Skip to Step 3.
  • If it does NOT exist: proceed to Step 2 (first-run flow).

Step 2: First-Run Flow (only when design-context.md doesn't exist)

2a. Codebase Research

Use the Explore agent to scan client-facing code: CSS/SCSS/Less/CSS-in-JS, JS/TS/JSX/TSX, templates (HTML/PHP/ERB/EJS/HBS/Vue/Svelte), and key entry-point pages. Extract: navigation structure, layout patterns (grid/sidebar/full-width/card), content hierarchy, interactive elements (forms/modals/tabs/accordions), responsive breakpoints, and existing UX conventions.

2b. Request Screenshots

Ask the user for 2-3 screenshots of the current app's key pages. Use AskUserQuestion:

> "To build accurate wireframes, I need to see the current app. Please provide 2-3 screenshots of key pages (e.g., homepage, article detail, summary page). You can drag & drop images or provide file paths."

When the user provides screenshots:

  • Read each screenshot image file to study the visual layout
  • Note: page structure, content zones, navigation placement, whitespace usage, interactive affordances
  • Save screenshots to wireframe/brain/ with descriptive names

If the codebase research and screenshots don't make the target platform obvious (mobile app vs desktop/web vs responsive), ask the user using AskUserQuestion:

> "What's the target platform for this product?"

Options:

  • Mobile — designing primarily for mobile screens
  • Desktop / Web — designing primarily for desktop or web browsers
  • Both (responsive) — designing for both with responsive layouts

Save the answer and use it in Step 2c under ## Target Platform.

2c. Write Design Context

Create wireframe/brain/design-context.md with this structure:

# Design Context — [Project Name]

Generated: [date]

## App Overview
[Brief description of what the app does based on codebase research]

## Target Platform
[Mobile / Desktop / Web / Both (responsive) — based on codebase research, screenshots, or user input]

## Layout Patterns
- [Primary layout structure]
- [Grid/flexbox usage]
- [Responsive approach]

## Navigation
- [Primary nav structure]
- [Secondary nav]
- [Mobile nav pattern]

## Page Types
### [Page Type 1]
- Structure: [layout description]
- Key elements: [list]

### [Page Type 2]
- Structure: [layout description]
- Key elements: [list]

## Interaction Patterns
- [Forms, modals, dropdowns, etc.]
- [Client-side behavior]

## Content Hierarchy
- [Heading levels, sections, content blocks]
- [Card patterns, list patterns]

## Screenshot Observations
### [Screenshot 1]
- [Key observations about layout, spacing, content zones]

### [Screenshot 2]
- [Key observations]

## UX Conventions
- [Patterns to maintain consistency with]

After writing the design context, confirm to the user that the context has been established, then proceed to Step 3 if $ARGUMENTS was provided, or ask what feature to wireframe.

Step 3: Generate Wireframes

3a. Parse the Feature

The feature to wireframe comes from $ARGUMENTS. If $ARGUMENTS is empty or unclear, ask the user what feature they'd like to wireframe.

3b. Optimization Goal (optional)

Before asking, check if $ARGUMENTS already contains an optimization intent — phrases like "make it more [adjective]", "improve [aspect]", "redesign for [goal]", "more compact", "more consistent", "reduce clutter", "simplify", "modernize", etc. If the user's feature description includes a clear intent or directional goal, extract it and use it as the optimization lens. Skip the AskUserQuestion entirely.

Only ask the optimization goal question if the feature description is neutral (e.g., "article page", "settings dashboard") with no directional intent:

> "What are you optimizing for with this feature? This helps me focus the UX options. Feel free to skip if you're exploring broadly."

Options:

  • More conversions — prioritize clear CTAs, reduced friction, and persuasive layout
  • Less drop-offs — prioritize simplicity, progress indicators, and error prevention
  • More time on page — prioritize content depth, engagement hooks, and exploration paths
  • Better discoverability — prioritize navigation, search, and information scent

If the user provides a goal (or one was extracted from $ARGUMENTS), use it to:

  • Weight the UX philosophy selection toward approaches that serve that goal
  • Frame the pros/cons of each option in terms of how well it achieves the goal
  • Highlight which option best serves the stated goal in the report

If the user skips, proceed normally without a specific optimization lens.

3b-ii. Scope: Wireframes only or wireframes + visuals?

If $ARGUMENTS explicitly requests both wireframes and visuals (e.g., "wireframe and visuals for...", "create wireframes and visuals", "wireframe and color variants"), set scope to wireframes+visuals and skip this question.

If $ARGUMENTS only says "wireframe" or doesn't mention visuals, ask using AskUserQuestion:

> "Would you like wireframes only, or wireframes with color variants (Clean + Polished)?"

Options:

  • Wireframes + visuals (Recommended) — generates B&W wireframes, then Clean + Polished color variants
  • Wireframes only — generates B&W wireframes only, skips color phase

Store the answer. Use it in Step 3f to decide whether to launch Phase 2.

3c. Clarify (if needed)

Ask at most 1-2 clarifying questions about the feature using AskUserQuestion. Only ask if there's genuine ambiguity. Examples of good clarifying questions:

  • "Should this page be accessible to logged-in users only, or public?"
  • "Does this feature replace the current X, or is it a new addition?"

Do NOT ask clarifying questions about visual styling — this is a UX wireframe, not a design comp.

3d. Generate Wireframes (Phase 1 — UX Architect)

Create an output folder at wireframe/DDMM-/ where DDMM is today's date formatted as day then month, zero-padded (e.g., Feb 22 → 2202, Mar 5 → 0503), and `` is a kebab-case slug derived from the feature description.

Template-based workflow — copy from wireframe/templates/, then edit:

  1. Copy this skill's bundled wireframe/templates/base.css → output folder (use Bash cp)
  2. Copy this skill's bundled wireframe/templates/template.html → output folder as index.html (use Bash cp)
  3. Create 5 empty styles-optN.css files (use Write tool, empty content)
  4. Edit index.html with targeted replacements (use Edit tool with replace_all where a placeholder appears more than once):
  • FEATURE_NAME → actual feature name (appears in `` and title bar)
  • PROJECT_NAME → project name from design-context.md
  • N in N → recommended option number
  • OPTION_1_NAME through OPTION_5_NAME → short option names (each appears in tab button + header, use replace_all)
  • OPTION_1_PHILOSOPHY through OPTION_5_PHILOSOPHY → one-sentence descriptions
  • FEATURE_SLUG → URL slug (e.g., settings, checkout) — use replace_all for all 5 browser URLs
  • ` through ` → wireframe HTML content
  • ` through ` → annotation lists
  • `` → summary table + recommendation
  1. Write styles.css with wireframe-content-specific CSS only (~200-300 lines). Framework CSS (layout, tabs, browser frame, annotations, banner, footer) is already in base.css — do NOT duplicate it. Only write rules for wireframe content elements (cards, forms, grids, etc.).

Generate 5 B&W wireframe options (Option 1: Safe + Options 2–5: exploratory). Phase 1 renders each option's content once inside a `. Sub-tab JS toggles the wrapper class between wireframe, clean, and polished — no separate content panels. Annotations (.wf-annotations) are hidden via CSS when class is clean or polished`.

The output MUST follow these rules:

Structure
  • Wireframe-specific CSS in styles.css, framework CSS in base.css (copied from template). No ` tags. Any @import statements MUST appear at the very top of styles.css`, before all other rules.
  • All JS is already in the template's inline `` — do NOT modify it.
  • No external dependencies — no CDN links, no fonts, no icon libraries. System fonts only: -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif
  • No dotted or dashed borders — use solid lines (1px solid) or whitespace for separation.
  • Use href='#' on all `` tags.
  • CSS architecture: One HTML block per option shared across all 3 sub-tabs. The browser-frame wrapper gets a class toggled by JS: .wireframe (default), .clean, or .polished. Variant overrides in styles-optN.css using .clean .selector and .polished .selector. Browser chrome dots styled per variant via CSS. Do NOT create separate sub-panel divs for each variant — there is one panel per option. Variant-specific classes only override: color, background, border-color, box-shadow, font-family, font-weight, transition, animation. Do NOT duplicate layout rules in variant CSS. Budget: wireframe content CSS in styles.css ≤ 300 lines.
Color Palette

Wireframe sub-tab (STRICT — no other colors allowed): #000 (text/borders), #333 (secondary), #666 (tertiary/labels), #999 (placeholder/disabled), #ccc (borders/dividers), #eee (subtle backgrounds/hover), #fff (background). No colors. No brand colors. Pure B&W with structural grays.

Page Layout

No introductory text above wireframes. HTML starts directly with the title bar and tab navigation.

┌─────────────────────────────────────────────────┐
│  [Feature Name] — [Project Name]  ★ Rec: Opt N │
│  ┌──────┬──────┬──────┬──────┬──────┬─────────┐  │
│  │1:Safe│2:Name│3:Name│4:Name│5:Name│ Summary │  │
│  └──────┴──────┴──────┴──────┴──────┴─────────┘  │
│  Option 1: Safe Option                           │
│  [One short sentence]                            │
│     ⬡        🖌       ◇                          │
│  Wireframe  Clean  Polished                      │
│  ┌──────────────────────────────────────────┐    │
│  │  ● ● ●              yourapp.com  ─ □ ✕  │    │
│  │  ┌────────────────────────────────────┐  │    │
│  │  │                                    │  │    │
│  │  │  [Content for selected sub-tab]    │  │    │
│  │  │                                    │  │    │
│  │  └────────────────────────────────────┘  │    │
│  └──────────────────────────────────────────┘    │
└─────────────────────────────────────────────────┘

Each main tab shows its own option panel with sub-tabs. Summary tab shows scoring table + recommendation. Only one panel visible at a time.

Attribution footer and browser frame styling are pre-built in base.css (copied from template). The template includes the footer HTML, browser chrome with dots/URL/controls, and mobile frame variant. Do NOT regenerate these — they are already in index.html from the template copy.

Browser frame rules (enforced by base.css):

  • Max width: max-width: 900px; margin: 0 auto;
  • Content area: overflow: hidden — no content may escape the frame.
  • Mobile frame: If design-context.md target platform is Mobile, add class mobile-frame to each .browser-frame element via Edit.
  • Chrome bar: Gray dots in wireframe; Phase 2 colors them red/yellow/green via variant CSS.
Required Sections in Each Option
  1. Title: "Option [N]: [Short Name]" — 1-3 word name (e.g., "Card Stack", "Step Flow")
  2. Philosophy description: One brief sentence
  3. The wireframe itself: Full interactive mockup with realistic placeholder content

Option 1: Safe Option — uses layout structures, navigation patterns, and component styles from design-context.md. Natural extension of the existing app — no new paradigms. Low-risk baseline.

Summary Tab

Final tab with: a compact scoring table (option name, 1-line description, 1–5 star score against optimization goal or "Overall UX"), and 1–2 sentence recommendation rationale.

Interactivity

Include functional interactive elements: clickable tabs/accordions, typeable form inputs, hover states (grays only), toggleable states, responsive behavior.

UX Approaches

Option 1: Safe Option — replicate existing patterns from design-context.md.

Options 2–5 — pick 4 from: Progressive Disclosure, Dashboard-First, Wizard/Step-by-Step, Hub-and-Spoke, Inline Editing, Split View, Card-Based, Conversational, Kanban/Column, Timeline, Search-First, Contextual Actions, Feed-Based, Spatial/Map-Centric, Gesture-Driven, Command Palette, Notification-Driven, Floating Action, Comparison Table, Drag-and-Drop, Accordion/Collapsible, Gamified Progress — or create your own.

Content Guidelines
  • Realistic placeholder content (not lorem ipsum), realistic data quantities
  • Show empty/loading/error states where relevant
  • Label interactive elements clearly
Annotation System

Small numbered markers (①②③) on wireframe elements, corresponding to UX notes below. Visible but not dominant — small circles with light border. Wireframe sub-tab only — not on color variants.

Sub-tab Behavior

Sub-tabs — iOS-style icon tab bar (3 items: Wireframe, Clean, Polished):

| Tab | Icon | No content yet | Ready | Hover (ready) | Active (selected) | |---|---|---|---|---|---| | Wireframe | sketch | — | — | — | #000 + 2px bottom border | | Clean | paint | #999 (gray) | #64B5F6 (light blue) | #1E88E5 (medium blue) | #1565C0 (dark blue) + 2px bottom border | | Polished | diamond-one | #999 (gray) | #81C784 (light green) | #43A047 (medium green) | #2E7D32 (dark green) + 2px bottom border |

Wireframe icon: always #000 when active, #666 otherwise. No progress states needed.

  • 24×24 SVG icon above 12px label (system font), 2px colored underline when active
  • Sub-tab bar visually lighter than main tabs — compact spacing for clear hierarchy.
  • Clean and Polished sub-tabs start gray (#999) and transition to their active colors when CSS loads.

Badges, sub-tab icons, completion banner, footer, and all JS are pre-built in the template. Do NOT regenerate any of these —

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.