Agent skill

UI Craft

by MemTensor in MemTensor/memmy-agent

Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA.

MITAuto-check passedFrontend & Design

Install UI Craft

skills CLI
$ npx skills add MemTensor/memmy-agent --skill ui-craft -a claude-code

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

GitHub CLI
$ gh skill install MemTensor/memmy-agent ui-craft --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/MemTensor/memmy-agent.git skills-src && mkdir -p .claude/skills && cp -r skills-src/App/memmy-agent/src/skills/ui-craft .claude/skills/ui-craft && 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
ui-craft
GitHub stars
2.1k
Token cost
~2.9k tokens
SKILL.md length
1,551 words
Files
1
Skills in repo
10
Repo updated
First seen
Licence
MIT

At a glance

Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA.

  • Works in 12 steps: Classify the target by runtime needs,… → Use browser_navigate to open the real… → Use browser_snapshot to inspect page… → …
  • Svelte pages and components
  • SKILL.md covers Establish the design brief, Read the existing visual…, Choose the composition before… and Build a coherent visual system, plus 5 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

UI Craft is an agent skill from MemTensor/memmy-agent. Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA. Use for HTML, CSS, JavaScript, React, Vue, or Svelte pages and components, landing pages, dashboards, admin tools, settings, forms, onboarding, responsive layouts, prototypes, and any frontend change that alters visible browser output.

Its SKILL.md is about 2.9k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Frontend & Design, covering Responsive design, Frontend development and Landing pages. It works with React, Svelte, Vue.js and JavaScript. The repository describes itself as: 🍙 A personal AI agent & local memory hub for all AI agents, gives every AI one shared, fully controlled memory and persistent context — all AI remember the same you. Now… The licence is MIT.

When your agent uses it

  • Svelte pages and components
  • Responsive layouts
  • Any frontend change that alters visible browser output

Example prompts

  • “/ui-craft”

Workflow steps

12 steps, taken from the first numbered list in SKILL.md.

  1. Classify the target by runtime needs, not by file count
  2. Use browser_navigate to open the real rendered page. Do not substitute source inspection for opening it.
  3. Use browser_snapshot to inspect page structure, visible content, semantics, and important controls.
  4. Exercise the primary interaction and important states with browser_find, browser_click, browser_type, browser_select_option…
  5. Use browser_console_messages to inspect errors and meaningful warnings.
  6. Use browser_network_requests to inspect failed, blocked, duplicated, or unexpected requests.
  7. Use browser_take_screenshot and visually inspect the primary viewport.
  8. Use browser_resize for at least one materially different relevant viewport, then repeat structural and screenshot inspection there.
  9. Judge the screenshots, not merely the presence of DOM nodes. Check composition, hierarchy, alignment, spacing rhythm, typography…
  10. Run the anti-template review against the screenshots.
  11. Fix every material issue and repeat the complete relevant sequence until the implementation and the rendered result are sound.
  12. If this task started a development or preview service, terminate its Exec session after validation.

What it can do on your machine

Read from SKILL.md and the folder at commit ee0ed02. 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

    No scripts in the folder and no shell commands in SKILL.md.

    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

UI Craft loads about 2.9k tokens when it runs. Until then it costs about 110 tokens; SKILL.md has 1,551 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~110
When it runs · the whole SKILL.md, loaded when a task matches
~2.9k

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); files beside SKILL.md are not scanned.

SKILL.md

The full file from MemTensor/memmy-agent at commit ee0ed02, republished under its MIT licence (© MemTensor). 1,551 words, ~2,922 tokens.

Download SKILL.mdSave it as .claude/skills/ui-craft/SKILL.md (or your agent's skills folder).
name
ui-craft
description
Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA. Use for HTML, CSS, JavaScript, React, Vue, or Svelte pages and components, landing pages, dashboards, admin tools, settings, forms, onboarding, responsive layouts, prototypes, and any frontend change that alters visible browser output.

UI Craft

Create an intentional interface, not merely valid frontend code. Make the page fit its product, help the user act, and hold up under real browser inspection.

Establish the design brief

Before editing, identify:

  • the user and the situation in which they reach this interface;
  • the primary job, decision, or action the interface must support;
  • the information that must be noticed first, second, and later;
  • the dominant behavior: scan, operate, enter, compare, learn, or explore;
  • the important default, loading, empty, error, success, disabled, and permission states;
  • the target viewports, input methods, accessibility needs, and engineering constraints.

Ask a small number of focused questions only when a missing answer would materially change the result. Otherwise proceed with conservative assumptions and preserve the user's scope.

Read the existing visual language

Inspect the rendered page and the source that actually defines it. Read relevant:

  • routes and page composition;
  • layout shells and navigation;
  • theme, token, and global style files;
  • nearby components and their variants;
  • typography, icons, imagery, and other assets;
  • copy conventions and product terminology;
  • responsive rules and existing breakpoints.

Treat the repository as the source of truth. Reuse its components and tokens before adding new ones. Do not introduce a parallel palette, spacing scale, radius family, shadow language, or component pattern for one screen.

Preserve the established product language unless the request explicitly calls for a redesign. If the current interface is inconsistent, improve the requested surface without casually rewriting unrelated global styles.

Choose the composition before decoration

Let the dominant user behavior determine the page structure:

  • For scanning or monitoring, favor density, stable alignment, glanceable status, and clear exceptions.
  • For operating or managing, favor selection state, nearby actions, feedback, and efficient repeated work.
  • For entering or configuring, favor grouping, progressive disclosure, validation, and a clear completion path.
  • For comparing, align equivalent information and make differences easy to inspect.
  • For learning or deciding, build a paced narrative with one clear idea or action at a time.
  • For exploring, make search, filters, results, preview, and wayfinding the primary composition.

Do not give every product surface a marketing hero. Do not default to a centered heading followed by three equal cards. Choose what should dominate, what should support it, and what should recede.

Form a one-sentence visual thesis before implementation. State the intended hierarchy, density, and tone, for example: “A compact operational surface where exceptions lead, controls stay close to data, and color is reserved for state.” Use the thesis to reject choices that do not belong.

Build a coherent visual system

Use a small number of deliberate decisions consistently.

When no established system exists, define the smallest useful set of local tokens for color, type, spacing, shape, and depth before styling individual components. Keep the system proportional to the deliverable.

Hierarchy and layout
  • Establish hierarchy with scale, position, spacing, alignment, weight, and contrast before adding containers.
  • Use whitespace to group and separate information, not simply to make the page airy.
  • Align related content to a shared grid.
  • Vary emphasis according to importance; avoid making every section, card, and action equally loud.
  • Keep the primary action easy to find without turning every control into a primary button.
  • Prefer content-driven dimensions over fragile fixed heights and viewport assumptions.
Typography
  • Reuse the product type system when one exists.
  • Choose a compact type scale with clear roles for display, heading, body, label, metadata, and numeric content.
  • Use weight and size intentionally; do not rely on low-contrast gray text for the entire hierarchy.
  • Keep line length, line height, wrapping, and truncation readable at each viewport.
  • Use monospace only for content that benefits from it.
Color, shape, and depth
  • Reuse semantic color tokens and verify important text and controls have sufficient contrast.
  • Reserve saturated color for meaning, focus, or a small number of high-value accents.
  • Use borders, surfaces, shadows, and elevation as one system; do not mix unrelated treatments.
  • Keep radii consistent with the product and appropriate to component size.
  • Avoid wrapping every group in a rounded card when spacing or a divider would communicate structure better.
Imagery, icons, and motion
  • Use supplied assets and the existing icon system when available.
  • Add icons only when they improve recognition or scanning; do not use decorative icon badges as filler.
  • Use motion to explain continuity, state change, or feedback.
  • Keep motion brief and interruptible, and respect prefers-reduced-motion.
  • Do not invent elaborate illustrations, fake screenshots, or stock imagery to fill an unresolved composition.

Keep content honest

Use real provided content or realistic structure that does not invent product claims.

  • Do not add fake metrics, testimonials, customers, integrations, or strategic claims.
  • Do not add generic feature sections merely to fill space.
  • Use specific labels and actions instead of vague words such as “Optimize,” “Insights,” or “Transform.”
  • Preserve domain terminology from the product.
  • Mark unavoidable placeholder copy clearly when it could be mistaken for final content.

Implement the real interface

  • Work in the repository's actual framework, component architecture, package manager, and styling approach.
  • Implement the requested deliverable in runnable form. In an existing product, modify the actual route or component; create a standalone artifact only when the user requests one.
  • Keep semantics meaningful and controls keyboard-accessible.
  • Provide visible focus, hover, active, selected, disabled, loading, empty, error, and success behavior where applicable.
  • Associate labels, descriptions, validation, and errors with the controls they explain.
  • Keep touch targets usable and avoid interactions that depend only on hover.
  • Make layout behavior intentional across relevant widths; do not treat mobile as a scaled-down desktop.
  • Prevent clipping, overlap, accidental horizontal scroll, unreadable wrapping, and controls moving off-screen.
  • Preserve working behavior while improving presentation. Do not trade correctness for visual novelty.
  • Keep changes scoped and avoid unrelated refactors.

For exploratory work, create multiple directions only when the user requests alternatives or the visual direction is genuinely unresolved. Make each direction differ in composition, hierarchy, density, or interaction model—not only color. Keep experiments separate from production code until a direction is selected.

Show full SKILL.md (571 more words)Show less

Run the anti-template review

Inspect the rendered composition before polishing details. Look for:

  • a page archetype copied from an unrelated use case;
  • a centered intro or oversized headline where users need to act or scan;
  • repeated equal-weight cards with no visual priority;
  • excessive rounded containers, pills, icon tiles, or badges;
  • arbitrary blue-purple gradients, glow, blur, or glass effects;
  • decoration compensating for weak hierarchy;
  • invented data or copy used as visual filler;
  • every section using the same rhythm and proportions;
  • default typography with no deliberate hierarchy;
  • low-contrast text, tiny labels, or color carrying meaning alone;
  • responsive layouts that merely stack everything without reconsidering priority.

Diagnose the cause before changing details:

  • Recompose when the page structure or dominant behavior is wrong.
  • Rebalance hierarchy when everything has equal emphasis.
  • Remove decoration when it does not communicate meaning.
  • Refine type, color, and spacing only when the underlying composition is sound.

Do not declare the interface polished while a structural problem remains.

Browser visual QA

When the browser_* tools are available, complete the following workflow before declaring visible frontend work finished:

  1. Classify the target by runtime needs, not by file count:
    • For independent static HTML whose files run as written without a build, backend, or application route, pass the permitted local .html/.htm path directly to browser_navigate.
    • For framework or service-backed work, start the project's existing development or preview command in the foreground with yield_time_ms, wait until it is ready, and use its HTTP or HTTPS route.
  2. Use browser_navigate to open the real rendered page. Do not substitute source inspection for opening it.
  3. Use browser_snapshot to inspect page structure, visible content, semantics, and important controls.
  4. Exercise the primary interaction and important states with browser_find, browser_click, browser_type, browser_select_option, browser_press_key, or browser_wait_for as relevant.
  5. Use browser_console_messages to inspect errors and meaningful warnings.
  6. Use browser_network_requests to inspect failed, blocked, duplicated, or unexpected requests.
  7. Use browser_take_screenshot and visually inspect the primary viewport.
  8. Use browser_resize for at least one materially different relevant viewport, then repeat structural and screenshot inspection there.
  9. Judge the screenshots, not merely the presence of DOM nodes. Check composition, hierarchy, alignment, spacing rhythm, typography, contrast, content density, clipping, overflow, control prominence, and visual consistency.
  10. Run the anti-template review against the screenshots.
  11. Fix every material issue and repeat the complete relevant sequence until the implementation and the rendered result are sound.
  12. If this task started a development or preview service, terminate its Exec session after validation.

Treat console, network, snapshot, and screenshot checks as different evidence. A clean console does not prove the layout is good, and an attractive screenshot does not prove the interactions work.

If browser tools are unavailable, run the strongest static and project checks available and explicitly report that browser validation was not performed. Do not claim observations that were not made.

Completion gate

Finish only when:

  • the interface supports the primary user job and has a clear hierarchy;
  • the visual choices form one coherent system and fit the surrounding product;
  • relevant interaction and content states are implemented;
  • accessibility and responsive behavior have been checked;
  • project validation passes;
  • browser structure, console, network, interactions, and screenshots have been inspected when tools are available;
  • material visual problems found during review have been repaired and rechecked.

Report concisely:

  • the visible outcome;
  • the important design decisions;
  • the files or routes changed;
  • the viewport and interaction coverage;
  • the browser and project validation performed;
  • any remaining limitation that affects confidence.

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

Files

Just SKILL.md in App/memmy-agent/src/skills/ui-craft of MemTensor/memmy-agent.

Open the folder on GitHubat commit ee0ed02

Compare with similar skills

UI Craft 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.

UI Craft compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
UI Craft this skillMemTensor/memmy-agent2.1k—~2.9kAutomated safety check: PassMIT
Premium UI Revampbuildfastwithai/gen-ai-experiments785—~2kAutomated safety check: PassMIT
UI/UX Design System AdvisorGalaxy-Dawn/claude-scholar5.7k1 repos~1.1kAutomated safety check: PassMIT
Design StyleCastor6/tactus3761 repos~2.1kAutomated safety check: PassApache-2.0
Frontend Designjjyaoao/HelloAgents3.2k1 repos~7kAutomated safety check: PassMIT
Figma To StaticLeoYeAI/openclaw-master-skills2.2k—~7.1kAutomated safety check: PassMIT

Similar skills

  • Premium UI Revamp

    buildfastwithai/gen-ai-experiments

    Audit and revamp existing web interfaces into polished, professional, product-specific experiences while preserving behavior and working within the current frontend stack.

    785 GitHub stars~2k tokensUpdated 15 days ago
    Frontend & DesignAuto-check passed
  • 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
  • Design Style

    Castor6/tactus

    A skill your agent uses whenever the user asks to build, create, design, develop, improve, or style any frontend interface or visual element.

    376 GitHub starsUsed in 1 repo~2.1k tokens
    Frontend & DesignAuto-check passed
  • Frontend Design

    jjyaoao/HelloAgents

    Transform UI style requirements into production-ready frontend code with systematic design tokens, accessibility compliance, and creative execution.

    3.2k GitHub starsUsed in 1 repo~7k tokens
    Frontend & DesignAuto-check passed
  • Figma To Static

    LeoYeAI/openclaw-master-skills

    Convert Figma design files to pixel-level mobile-first static HTML/CSS pages.

    2.2k GitHub stars~7.1k tokensUpdated 2 mo ago
    Frontend & DesignAuto-check passed
  • GSAP Core Animation

    greensock/gsap-skills

    Covers the GSAP core API for tweens, easing, staggers, defaults and matchMedia, and when to choose GSAP over CSS animations or other JavaScript animation libraries.

    16k GitHub starsUsed in 4 repos~3.7k tokens
    Frontend & DesignAuto-check passed

More from MemTensor/memmy-agent

All 10 skills in this repo
  • DOCX

    MemTensor/memmy-agent

    Create, read, edit, review, redline, comment on, merge, audit, render, and verify .docx Word documents, including layout-sensitive and OOXML-level work.

    2.1k GitHub stars~1.9k tokensUpdated 7 days ago
    Auto-check passed
  • Skill Creator

    MemTensor/memmy-agent

    Create, edit, improve, tidy, review, audit, or restructure memmy-agent skills and SKILL.md files.

    2.1k GitHub stars~944 tokensUpdated 7 days ago
    Auto-check passed
  • Update Setup

    MemTensor/memmy-agent

    One-time setup wizard for the memmy upgrade skill. An agent skill from MemTensor/memmy-agent.

    2.1k GitHub stars~1.3k tokensUpdated 7 days ago
    Auto-check passed
  • Agent Memory Onboarding

    MemTensor/memmy-agent

    On-demand provisioning guide for a Memmy GUI task that fully connects an explicitly named local Agent: discover its active history store, install or remove its rendered Memmy Skill, bootstrap a scan…

    2.1k GitHub stars~4.6k tokensUpdated 7 days ago
    Auto-check passed
  • PPTX

    MemTensor/memmy-agent

    Create, inspect, edit, validate, render, and deliver PowerPoint presentations and templates while preserving OOXML structure and user metadata.

    2.1k GitHub stars~861 tokensUpdated 7 days ago
    Auto-check passed
  • Computer History

    MemTensor/memmy-agent

    Answer questions about what the user recently did on their computer — who contacted them, what they were working on, where they left off — by reading the local Computer History summaries and raw…

    2.1k GitHub stars~947 tokensUpdated 7 days ago
    Auto-check passed

Questions about UI Craft

What does UI Craft do?

Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA. UI Craft is an agent skill from MemTensor/memmy-agent. Design, build, redesign, or modify browser-visible web interfaces with product-quality composition, visual systems, interaction states, responsive behavior, accessibility, and browser visual QA.

When should I use UI Craft?

UI Craft fits situations like: svelte pages and components; responsive layouts; any frontend change that alters visible browser output.

How do I install UI Craft in Claude Code?

Run `npx skills add MemTensor/memmy-agent --skill ui-craft -a claude-code`. Or copy the skill folder (App/memmy-agent/src/skills/ui-craft in MemTensor/memmy-agent) into .claude/skills/ui-craft in your project. Claude Code loads it when a task matches its description.

How do I install UI Craft in Codex?

Run `npx skills add MemTensor/memmy-agent --skill ui-craft -a codex`. Or copy the skill folder (App/memmy-agent/src/skills/ui-craft in MemTensor/memmy-agent) into .agents/skills/ui-craft in your project. Codex loads it when a task matches its description.

Can I use UI Craft 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 MemTensor/memmy-agent --skill ui-craft -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ui-craft, .gemini/skills/ui-craft, .github/skills/ui-craft and .opencode/skills/ui-craft in your project.

What does UI Craft need to run?

SKILL.md names no scripts, command-line tools or credentials: UI Craft is instructions for the agent only.

Does UI Craft 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 UI Craft 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. Review the folder before installing.

What licence does UI Craft use?

UI Craft 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 UI Craft 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.

What are the alternatives to UI Craft?

Skills that share tags, products or a category with UI Craft: Premium UI Revamp (buildfastwithai/gen-ai-experiments, 785 stars), UI/UX Design System Advisor (Galaxy-Dawn/claude-scholar, 5.7k stars), Design Style (Castor6/tactus, 376 stars) and Frontend Design (jjyaoao/HelloAgents, 3.2k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains UI Craft?

MemTensor (a GitHub organization) maintains it in MemTensor/memmy-agent, which has 2,064 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on September 30, 2026.

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