design-with-claude

agent
Security Audit
Fail
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Fail
  • process.env — Environment variable access in scripts/test-e2e-roundtrip.mjs
  • network request — Outbound network request in scripts/test-e2e-roundtrip.mjs
  • Hardcoded secret — Potential hardcoded credential in scripts/test-e2e-roundtrip.mjs
  • process.env — Environment variable access in scripts/test-mcp-handshake.mjs
  • process.env — Environment variable access in scripts/test-phase2-flow.mjs
  • network request — Outbound network request in scripts/test-phase2-flow.mjs
  • spawnSync — Synchronous process spawning in scripts/test-setup-dry-run.mjs
  • network request — Outbound network request in src/api-client.ts
  • spawnSync — Synchronous process spawning in src/bin/setup.ts
  • network request — Outbound network request in src/bin/setup.ts
  • process.env — Environment variable access in src/config.ts
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This project is a collection of markdown-based agent files and Claude Code commands that provide structured, expert-level design guidance. It injects domain-specific knowledge (like WCAG compliance and token architecture) directly into your coding workflow.

Security Assessment
The overall risk is rated as Medium. While the core concept relies on dependency-free markdown files, the repository includes scripts and source code that introduce notable security concerns. The scanner flagged a hardcoded credential (FAIL) and several outbound network requests within the end-to-end test scripts. Additionally, the codebase uses synchronous process spawning (`spawnSync`) in the setup scripts, which can execute system commands. While environment variables are accessed, no inherently dangerous permissions are requested by the tool itself. Developers should be cautious of the setup scripts and avoid running them without reviewing the hardcoded secrets and network endpoints first.

Quality Assessment
The project is freshly maintained with a push made just today, and it benefits from a clear MIT license. However, community trust and visibility are currently very low. The repository only has 5 GitHub stars, meaning the codebase has not been broadly vetted or battle-tested by a large audience yet.

Verdict
Use with caution. The core markdown agents are safe, but you should review the setup scripts and source files for hardcoded secrets and unexpected network behavior before executing them locally.
SUMMARY

37 specialist agents + 1 master command — 38 design experts as Claude Code commands. Bring senior design expertise into every coding session. No runtime. No dependencies. No API keys. Just markdown files that give Claude deep design knowledge.

README.md

Design with Claude

37 design specialists + 7 technical guides + 1 master command — 45 experts as Claude Code commands. Design expertise and hands-on technical help for designers who build.

No runtime. No dependencies. No API keys. Just markdown files that give Claude deep design knowledge.

designwithclaude.com

What This Is

A collection of 44 agent files — 37 design specialists and 7 technical guides for designers. Install them as a Claude Code plugin, and you get expert-level design guidance plus hands-on help with setup, debugging, and deployment directly in your coding workflow.

The design agents contain structured domain knowledge: WCAG specifics, token architecture patterns, motion timing curves, healthcare UX compliance rules, checkout conversion best practices. The technical guides help designers get unstuck with common walls — environment setup, database connections, auth, deployment, and error debugging. This isn't generic prompting — it's deep, opinionated expertise.

Install

As a Plugin (recommended)

# Add the marketplace
/plugin marketplace add imsaif/design-with-claude

# Install the plugin
/plugin install design-with-claude@design-with-claude

Commands are namespaced: /design-with-claude:design-brief, /design-with-claude:accessibility-specialist, etc.

As Standalone Commands

If you prefer shorter command names (no namespace prefix):

# Clone the repo
git clone https://github.com/imsaif/design-with-claude.git ~/.design-with-claude

# Copy commands to your user-level commands directory
cp -r ~/.design-with-claude/commands/ ~/.claude/commands/

This gives you /design-brief, /accessibility-specialist, etc. directly.

Quick Start

/design-brief Build a SaaS analytics dashboard with dark mode and accessibility focus

The master command analyzes your brief, identifies the relevant design domains (out of 44), and returns structured guidance — token recommendations, component specs, layout decisions, and implementation notes.

When invoked inside a code project, commands are context-aware: they detect your stack, read your existing files, and generate output that matches your conventions.

Example Output

## Brief Analysis
"SaaS analytics dashboard with dark mode and accessibility focus"
Product type: B2B SaaS dashboard
Key requirements: data visualization, dark theme, WCAG AA compliance

## Relevant Domains (7 of 44)
1. Dashboard Designer — KPI cards, chart layout, data density
2. Dark Mode Specialist — surface hierarchy, elevation tokens, contrast
3. Accessibility Specialist — WCAG AA, focus management, screen readers
4. Data Visualization Specialist — chart selection, color-blind palettes
5. Design System Architect — dark/light token architecture
6. Color Specialist — semantic colors with dark mode variants
7. Typography Specialist — data-dense type scale, monospace for numbers

## Token Recommendations
--surface-primary: hsl(220 20% 10%)     /* main background */
--surface-secondary: hsl(220 18% 14%)   /* card surfaces */
--surface-elevated: hsl(220 16% 18%)    /* modals, dropdowns */
--text-primary: hsl(220 10% 93%)        /* high-emphasis text */
--accent-primary: hsl(210 100% 60%)     /* interactive elements */
--data-series-1: hsl(210 90% 60%)       /* chart color 1 */
...

## Component Recommendations
- KPI cards: icon + metric + trend + sparkline, 4-column grid
- Charts: use accessible palette with pattern fills for color-blind users
- Tables: sticky headers, alternating row contrast ≥ 3:1
- Navigation: sidebar with collapsible sections, active state at ≥ 4.5:1

## Implementation Notes
- Dark mode as default, light mode as toggle (not vice versa)
- All interactive elements need visible focus rings (2px solid, offset 2px)
- Chart tooltips must be keyboard-accessible (not hover-only)
- Minimum touch target: 44x44px for any clickable element

Commands

Master Command

Command What it does
design-brief Takes a natural language brief → identifies relevant agents → outputs comprehensive design guidance

Core Design

Command Use when...
visual-hierarchy-specialist Layout, hierarchy, spacing, focal points
interaction-designer User flows, states, gestures, feedback
design-system-architect Tokens, component APIs, theming
accessibility-specialist WCAG compliance, ARIA, keyboard nav

Visual Design

Command Use when...
typography-specialist Type scales, font pairing, vertical rhythm
color-specialist Color systems, palettes, semantic colors
spacing-layout-specialist Grid systems, spacing scales, density
icon-illustration-specialist Icon grids, sizing, illustration style, SVG accessibility

Interaction Design

Command Use when...
motion-designer Transitions, timing curves, micro-interactions
form-designer Input layout, validation, multi-step forms
navigation-specialist Nav patterns, wayfinding, menus
notification-designer Push notifications, toasts, badges, notification center
drag-drop-specialist Drag affordances, drop zones, reordering, canvas

Product Design

Command Use when...
dashboard-designer Data display, charts, KPI cards
mobile-specialist iOS/Android patterns, touch, thumb zones
responsive-design-specialist Breakpoints, fluid layouts, adaptive patterns
landing-page-specialist Hero sections, CTAs, conversion layout
settings-designer Settings pages, preferences, toggle patterns
auth-security-ux-specialist Login flows, 2FA/passkey, session management

Content & IA

Command Use when...
content-strategist Microcopy, tone of voice, content hierarchy
information-architect Navigation, taxonomy, content structure
conversational-ui-designer Chat interfaces, voice UI

Industry

Command Use when...
healthcare-ux-specialist Clinical workflows, HIPAA, medical UI
b2b-saas-specialist Enterprise patterns, multi-tenant, admin UIs
ecommerce-specialist Product pages, cart, checkout flows
checkout-specialist Cart UX, payment forms, trust signals

Specialized

Command Use when...
dark-mode-specialist Dark theme tokens, surface hierarchy
error-handling-specialist Error states, recovery flows
onboarding-specialist First-run experience, progressive onboarding
performance-specialist Skeleton screens, loading states, perceived speed
data-visualization-specialist Chart types, axis design, data-ink ratio
table-designer Data tables, sorting, filtering, pagination
search-specialist Search UX, filters, faceted navigation
brand-designer Visual identity, brand systems
empty-loading-states-specialist Skeleton screens, empty states, loading patterns
i18n-designer RTL layouts, locale-aware UI, string expansion
print-export-designer PDF generation, print stylesheets, export UX

Technical Setup

Command Use when...
setup-guide Installing Node, Claude Code, creating first project
code-explainer Understanding any file or error in plain language
database-setup Setting up Supabase — tables, queries, frontend connection
environment-setup .env files, API keys, what never to commit
auth-implementation Adding login/signup with Clerk or Supabase Auth
deploy-to-vercel Getting your project live, fixing build errors
debug-helper Pasting any error, getting the exact fix

Examples

# Get a full design brief for a new product
/design-brief Healthcare patient portal with HIPAA compliance and mobile support

# Deep-dive into a specific domain
/accessibility-specialist Audit this login form for WCAG AA compliance
/design-system-architect Create a token architecture for our React component library
/motion-designer Define transition specs for our modal and dropdown components
/dashboard-designer Design a KPI overview page for a logistics platform

# Combine agents for thorough coverage
/form-designer Multi-step onboarding form with file uploads
/color-specialist Define a semantic color system for light and dark themes
/checkout-specialist Guest checkout flow for a subscription product

# New design skills
/notification-designer Design a notification system for a team collaboration app
/settings-designer Redesign our settings page — it's a mess
/auth-security-ux-specialist Passkey login flow with 2FA fallback
/drag-drop-specialist Kanban board with drag between columns and keyboard support

# Technical guides — get unstuck
/setup-guide I've never used a terminal before, help me set up Claude Code
/debug-helper Error: Cannot read properties of undefined (reading 'map')
/database-setup I need to store user submissions in a database
/deploy-to-vercel My build is failing with "Module not found"
/auth-implementation Add Google login to my Next.js app
/environment-setup What's a .env file and why do I need one?
/code-explainer What does this middleware.ts file do?

How It Works

Each command file is a markdown document containing a specialized design agent's complete knowledge. When you invoke a command, Claude loads that agent as context — giving it deep, structured expertise in that domain.

There's no code running, no API calls, no build step. The commands/ directory IS the product. Claude's intelligence does the rest.

Alternative: Project-Local Commands

Copy commands into a single project:

cp -r ~/.design-with-claude/commands/ your-project/.claude/commands/

Contributing

Adding a new skill

  1. Create commands/your-skill-name.md with YAML frontmatter (description field) and a role statement using $ARGUMENTS
  2. Follow the structure of existing commands: Expertise, Design Principles, Guidelines, Checklist, Anti-patterns, How to respond, What to ask if unclear
  3. Add to the command table in this README and to commands/design-brief.md

License

MIT — designwithclaude.com

Reviews (0)

No results found