Agent skill

Master Design

by templetongroup in templetongroup/radiant

Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps.

MITAuto-check passedFrontend & Design

Install Master Design

skills CLI
$ npx skills add templetongroup/radiant --skill master-design -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install templetongroup/radiant master-design --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/templetongroup/radiant.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/master-design .claude/skills/master-design && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
master-design
GitHub stars
113
Token cost
~2.9k tokens
SKILL.md length
1,468 words
Files
9 (incl. scripts, references)
Skills in repo
9
Repo updated
First seen
Licence
MIT

At a glance

Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps.

  • Works in 7 steps: Establish the truth before designing → Form a coherent direction → Build the design system before the pages → …
  • Product framing
  • SKILL.md covers Non-negotiable hierarchy, Choose the operating mode, 1. Establish the truth before… and 2. Form a coherent direction, plus 6 more sections
  • Runs Shell scripts from its folder

What it does

Master Design is an agent skill from templetongroup/radiant. Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps. Use for product framing, visual direction, UX flows, design systems, implementation, accessibility, responsive and window behavior, AI interactions, screenshot or reference reconstruction, whole-product reviews, pre-launch polish, and installed or deployed interface QA. Triggers on master design, design this, redesign this app, make this production ready, review every screen, polish the UI…

Its SKILL.md is about 2.9k tokens, which your agent loads only when the skill is triggered. The skill folder holds 10 other files, including scripts and reference files (for example `SOURCE.md`, `references/evidence-and-verification.md` and `references/product-and-flow.md`).

It sits in Frontend & Design, covering Design systems and Accessibility. The repository describes itself as: A local coding harness for Mac. Chat with coding agents across cloud and local models, watch every tool call in a live activity feed, and drive a real terminal — in one window… The licence is MIT.

When your agent uses it

  • Product framing
  • Visual direction
  • Responsive and window behavior
  • AI interactions

Example prompts

  • “/master-design”

Requirements

  • A Bash shell

Workflow steps

7 steps, taken from the step headings in SKILL.md.

  1. Establish the truth before designing
  2. Form a coherent direction
  3. Build the design system before the pages
  4. Implement in task order
  5. Review as one product
  6. Verify the actual experience
  7. Run the final slop gate

What it can do on your machine

Read from SKILL.md and the folder at commit 94838ca. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Shell), which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Master Design loads about 2.9k tokens when it runs, and up to ~10k if it reads all its reference files. Until then it costs about 175 tokens; SKILL.md has 1,468 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~175
When it runs · the whole SKILL.md, loaded when a task matches
~2.9k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~10k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from templetongroup/radiant at commit 94838ca, republished under its MIT licence (© templetongroup). 1,468 words, ~2,939 tokens.

Download SKILL.mdSave it as .claude/skills/master-design/SKILL.md (or your agent's skills folder). This skill also uses 8 other files; get the full folder from GitHub.
name
master-design
description
Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps. Use for product framing, visual direction, UX flows, design systems, implementation, accessibility, responsive and window behavior, AI interactions, screenshot or reference reconstruction, whole-product reviews, pre-launch polish, and installed or deployed interface QA. Triggers on master design, design this, redesign this app, make this production ready, review every screen, polish the UI, improve the UX, match this reference, design system, visual audit, interface audit, accessibility review, responsive review, pre-launch design check, and ship-quality UI.

Master Design

Create interfaces that are coherent, specific, usable, accessible, technically sound, and proven in the product people actually use. Treat visual quality and behavioral quality as one discipline.

This skill synthesizes a broad interface-design library without disabling or replacing the source skills. Read source-map.md only when auditing this synthesis, updating the master skill, or explaining its provenance.

Non-negotiable hierarchy

Resolve conflicts in this order:

  1. Follow the user's explicit brief, constraints, and approved acceptance target.
  2. Protect accessibility, truthfulness, safety, privacy, and user data.
  3. Follow the platform's interaction conventions and native semantics.
  4. Preserve the project's established design system, stack, brand, and density when they remain fit for purpose.
  5. Make task hierarchy, real content, and product state legible.
  6. Apply stylistic taste and novelty.

Never let a style heuristic overrule a higher item. A brutalist, minimal, editorial, playful, luxury, or hardware direction is a route, not a universal rule.

Choose the operating mode

Infer one mode from the request and state it briefly when useful.

ModeUse whenDefault action
frameThe problem, audience, or feature is unclearClarify the job, risk, evidence, and success condition before drawing UI
directA new visual direction or system is neededProduce 2–3 coherent directions, recommend one, and wait if the choice materially changes the result
buildThe user asks to create or change the productImplement the smallest complete system in the existing stack and verify it
reviewThe user asks to inspect, critique, or auditStay read-only; return evidence-backed findings and a verdict
polishA working interface needs launch-quality refinementFix shared primitives and high-leverage inconsistencies, then recheck all consumers
shipThe user asks for production readiness or publicationRun the full product and release evidence gates; publish only when explicitly authorized

If the request combines modes, use this order: frame → direct → build → polish → ship. For an existing approved direction, skip direct. For a review-only request, do not mutate source or external state.

1. Establish the truth before designing

  1. Read repository and platform guidance completely: AGENTS.md, CLAUDE.md, CONTRIBUTING.md, design-system docs, ADRs, token files, Storybook, and release instructions.
  2. Inventory the actual stack, components, tokens, fonts, icons, assets, supported sizes, localization, themes, and preview/test commands.
  3. Inventory the product surfaces and states: entry points, windows/routes, subnavigation, overlays, menus, loading, empty, error, disabled, selected, success, destructive, narrow, large, zoomed, and restored states.
  4. Determine whether each supplied image is an acceptance target, brand source, content source, or inspiration only. If ambiguity would materially change the work, ask once. Do not blend conflicting references silently.
  5. Use real product content and state. Find hard-coded personal names, demo data, stale version text, fake readiness, placeholders, and copy that claims more than the implementation proves.

For existing products, preserve the architecture and interaction model unless the user explicitly authorizes a rewrite. Redesign the system, not just the most visible screenshot.

Read product-and-flow.md for product framing, cognitive load, AI behavior, trust, copy, and failure recovery.

2. Form a coherent direction

Define one sentence for each:

  • Audience: who uses this and what they already understand.
  • Job: the single primary outcome of the surface or flow.
  • Tone: a specific character such as restrained technical, warm editorial, precise utilitarian, refined hardware, or expressive cultural. Never use “clean and modern” as the whole direction.
  • Hierarchy: what earns first, second, and third attention.
  • Signature: one or two memorable moves that belong to this product.

When direction is genuinely open, show 2–3 distinct systems, not color variations. For each, specify structure, typography, palette/material, interaction character, strengths, risks, and fit. Recommend one based on the audience and job.

When an acceptance target exists, measure it. Reconstruct its structure, proportions, alignment, typography roles, density, and component geometry before adding decoration. Preserve its design DNA without copying unrelated content or accidental defects.

Read style-routing.md for selecting and reconciling aesthetics, brand, anti-slop rules, and reference-based work.

3. Build the design system before the pages

Define or normalize these shared primitives before leaf-screen polish:

  • semantic color roles and contrast pairs;
  • typography roles, line heights, measures, and numeric styles;
  • spacing scale and layout margins;
  • radii, borders, elevation, material, and lighting direction;
  • icon family, stroke/fill weight, optical size, and decorative semantics;
  • control heights, horizontal padding, label baselines, focus treatment, and state behavior;
  • motion durations, easing, reduced-motion behavior, and interruption rules;
  • window/container size policies and responsive breakpoints based on content.

Peer actions that are intended to read as a set must share height, radius, vertical padding, baseline, and state geometry. Use equal width when the actions have equal rank; use content width deliberately when rank or available space differs. Check optical centering, not only numeric constraints.

Never solve a repeated defect in one screen when its cause is a token or shared component. Audit every consumer after changing a primitive.

Read system-and-craft.md for component geometry, layout, typography, color, surfaces, icons, motion, and accessibility details.

Show full SKILL.md (661 more words)Show less

4. Implement in task order

  1. Build semantic structure and state ownership.
  2. Establish layout, reflow, window sizing, and scroll behavior.
  3. Apply typography and content hierarchy.
  4. Apply color, material, imagery, and iconography.
  5. Implement every interaction state and keyboard path.
  6. Add restrained motion only when it explains change, confirms action, or preserves spatial continuity.
  7. Add empty, loading, error, success, offline, permission, and recovery states where the product can reach them.

Use native controls and semantics before custom replicas. Match the existing framework and styling method. Do not introduce a second UI stack for a cosmetic change.

When an installed platform skill covers the actual stack—such as Apple HIG, macOS/SwiftUI/AppKit, iOS, web, or React Native—load it for implementation mechanics. This master skill owns direction, quality gates, evidence, and final coherence; the platform skill owns framework-specific APIs and conventions.

For AI features, make scope, inputs, sources, progress, confidence limits, cost, reversibility, and the boundary between suggestion and action visible. Require preview or confirmation proportional to the consequence. Never present a static or inferred status as a live verified fact.

5. Review as one product

Review foundational domains before surface polish:

  1. Accessibility and input methods
  2. Product flow, clarity, trust, and recovery
  3. Layout, resizing, scrolling, and localization growth
  4. Writing and information hierarchy
  5. Typography
  6. Color and contrast
  7. Components, imagery, surfaces, icons, and motion
  8. Cross-screen consistency and release truth

Use one finding per root cause. Rank shared primitives above isolated symptoms. Every finding needs evidence from the correct layer:

  • source evidence for implementation facts;
  • rendered evidence for visual claims;
  • interaction evidence for behavior claims;
  • installed or deployed evidence for release claims.

Do not infer keyboard access, screen-reader output, clipping, animation behavior, or deployed state from a static screenshot. Do not infer visual quality from source alone.

Read review-contract.md for scope, severity, output, caps, and verdict rules.

6. Verify the actual experience

Create a surface-by-state matrix and exercise it. At minimum:

  • normal, minimum, and large window/viewport sizes;
  • 200% zoom or the platform's relevant text-size setting;
  • light/dark/high-contrast modes the product claims to support;
  • keyboard traversal, focus visibility, shortcuts, Escape, and focus restoration;
  • accessible name, role, value, and dynamic announcements;
  • real content extremes, localization growth, empty, loading, error, and permission-denied states;
  • destructive confirmation and recovery;
  • repeated actions, cancellation, interruption, and reopen/restoration;
  • actual installed app or deployed URL when evaluating a release.

Inspect rather than assume. Capture before/after evidence at matching dimensions when implementing visual changes. A successful build, signature, notarization, test suite, or preview does not prove the installed or deployed interface works.

Read evidence-and-verification.md for the complete proof ladder, platform matrices, and release gate.

7. Run the final slop gate

Before calling work complete, check all of these:

  • no clipped, overlapped, off-screen, or unreachable content;
  • no accidental horizontal scrolling;
  • no critical action below the fold without a visible continuation cue;
  • no mismatched peer-button geometry or label baselines;
  • no one-off spacing, radius, font, color, or icon that should use a shared token;
  • no tiny or low-contrast secondary text masquerading as polish;
  • no active state shown only by color;
  • no missing hover, focus, active, selected, disabled, loading, empty, error, or success state;
  • no unlabeled icon control, pointer-only path, broken focus order, or ignored reduced-motion preference;
  • no generic filler copy, unexplained jargon, fake data, hard-coded personal greeting, or false readiness claim;
  • no decorative effect competing with the task hierarchy;
  • no CSS or rendering trick that fights the platform;
  • no claim of completion beyond the evidence actually gathered.

Use scripts/scan-interface-risks.sh <project-root> as a fast candidate scan when the project contains source code. Treat its output as leads, never findings.

Completion contract

Lead with the outcome. State what changed or what was reviewed, what was physically verified, what remains unverified, and the verdict or release status. Link to real files with exact lines when possible.

Do not publish, deploy, buy, send, delete, or grant permissions unless the user explicitly authorized that action. “Production ready” authorizes preparation and verification, not publication by itself.

© templetongroup, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 8 other files (scripts, references) in .claude/skills/master-design of templetongroup/radiant.

  • SKILL.md
  • SOURCE.md
  • references/evidence-and-verification.md
  • references/product-and-flow.md
  • references/review-contract.md
  • references/source-map.md
  • references/style-routing.md
  • references/system-and-craft.md
  • scripts/scan-interface-risks.sh

Open the folder on GitHubat commit 94838ca

Compare with similar skills

Master Design next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Master Design compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Master Design this skilltempletongroup/radiant113—~2.9kAutomated safety check: PassMIT
UI/UX Design System AdvisorGalaxy-Dawn/claude-scholar5.7k1 repos~1.1kAutomated safety check: PassMIT
Extract DesignManavarya09/design-extract4.2k—~786Automated safety check: NotesMIT
Color Auditrome-os/rome743—~2.7kAutomated safety check: PassMIT
Material Design 3 UI/UX Guideskydashnet/material-design-3-ui-skill138—~3kAutomated safety check: PassMIT
UI Component Spec Designerplugin87/ux-ui-agent-skills1.6k—~2.8kAutomated safety check: PassMIT

Similar skills

  • UI/UX Design System Advisor

    Galaxy-Dawn/claude-scholar

    Turns a vague UI request into a concrete design system with style, palette, typography and layout guidance from a search script, plus stack-specific implementation advice.

    5.7k GitHub starsUsed in 1 repo~1.1k tokens
    Frontend & DesignAuto-check passed
  • Extract Design

    Manavarya09/design-extract

    Extract the full design language from any website URL. An agent skill from Manavarya09/design-extract.

    4.2k GitHub stars~786 tokensUpdated 2 days ago
    Frontend & DesignAuto-check: notes
  • Color Audit

    rome-os/rome

    Audit a design system's color palette against measurable color-science disciplines — WCAG/APCA contrast of declared token pairs, perceptual (OKLCH) ramp uniformity, color-blindness safety of…

    743 GitHub stars~2.7k tokensUpdated today
    Frontend & DesignAuto-check passed
  • Material Design 3 UI/UX Guide

    skydashnet/material-design-3-ui-skill

    Guides designing, reviewing or implementing interfaces that follow Google's Material Design 3 system: semantic tokens, component states, adaptive layout and accessibility.

    138 GitHub stars~3k tokensUpdated 10 days ago
    Frontend & DesignAuto-check passed
  • UI Component Spec Designer

    plugin87/ux-ui-agent-skills

    Writes a complete UI component spec with anatomy, variants, sizes, eight interaction states, design-token mapping and accessibility notes before or alongside code.

    1.6k GitHub stars~2.8k tokensUpdated 3 days ago
    Frontend & DesignAuto-check passed
  • Typeui Fundamentals

    bergside/typeui

    Universal UI/UX design principles covering visual hierarchy, interaction laws, typography foundations, and WCAG accessibility requirements.

    2k GitHub stars~861 tokensUpdated 3 mo ago
    Frontend & DesignAuto-check passed

More from templetongroup/radiant

All 9 skills in this repo
  • Design System

    templetongroup/radiant

    A skill your agent uses to generate or audit design systems, check visual consistency, and review PRs that touch styling.

    113 GitHub stars~775 tokensUpdated 4 days ago
    Auto-check passed
  • Tinystruct Patterns

    templetongroup/radiant

    Expert guidance for developing with the tinystruct Java framework.

    113 GitHub stars~3k tokensUpdated 4 days ago
    Auto-check passed
  • Agent Payment X402

    templetongroup/radiant

    Add x402 payment execution to AI agents with per-task budgets, spending controls, and non-custodial wallets.

    113 GitHub stars~2.7k tokensUpdated 4 days ago
    Auto-check passed
  • Motion UI

    templetongroup/radiant

    Production-ready UI motion system for React/Next.js. An agent skill from templetongroup/radiant.

    113 GitHub stars~3.5k tokensUpdated 4 days ago
    Auto-check passed
  • Plan Orchestrate

    templetongroup/radiant

    Read a plan document, decompose it into steps, design a per-step agent chain from the ECC catalogue, and emit ready-to-paste /orchestrate custom prompts.

    113 GitHub stars~4.5k tokensUpdated 4 days ago
    Auto-check passed
  • Star System

    templetongroup/radiant

    Ask the user to rate the code/output just produced on a 1-5 star scale, then ask rating-appropriate follow-up questions, log the rating, and iterate until the work reaches 4+ stars.

    113 GitHub stars~2.2k tokensUpdated 4 days ago
    Auto-check passed

Questions about Master Design

What does Master Design do?

Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps. Master Design is an agent skill from templetongroup/radiant. Design, redesign, build, audit, polish, and production-verify exceptional interfaces across web, mobile, and native apps.

When should I use Master Design?

Master Design fits situations like: product framing; visual direction; responsive and window behavior; AI interactions.

How do I install Master Design in Claude Code?

Run `npx skills add templetongroup/radiant --skill master-design -a claude-code`. Or copy the skill folder (.claude/skills/master-design in templetongroup/radiant) into .claude/skills/master-design in your project. Claude Code loads it when a task matches its description.

How do I install Master Design in Codex?

Run `npx skills add templetongroup/radiant --skill master-design -a codex`. Or copy the skill folder (.claude/skills/master-design in templetongroup/radiant) into .agents/skills/master-design in your project. Codex loads it when a task matches its description.

Can I use Master Design in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add templetongroup/radiant --skill master-design -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/master-design, .gemini/skills/master-design, .github/skills/master-design and .opencode/skills/master-design in your project.

What does Master Design need to run?

Going by SKILL.md and its folder, Master Design needs a shell for the scripts in its folder. Our summary lists: A Bash shell.

Does Master Design access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Master Design safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Master Design use?

Master Design is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Master Design use?

About 2.9k tokens (SKILL.md is roughly 12k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 7.5k tokens, read only when the agent opens those files.

What are the alternatives to Master Design?

Skills that share tags, products or a category with Master Design: UI/UX Design System Advisor (Galaxy-Dawn/claude-scholar, 5.7k stars), Extract Design (Manavarya09/design-extract, 4.2k stars), Color Audit (rome-os/rome, 743 stars) and Material Design 3 UI/UX Guide (skydashnet/material-design-3-ui-skill, 138 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Master Design?

templetongroup (a GitHub user) maintains it in templetongroup/radiant, which has 113 GitHub stars. The repository holds 9 skills in this directory. The repository was last updated on October 6, 2026.

Source: templetongroup/radiant on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.