More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Create Mockup — UI From Scratch, Matched to Your Stack

    1

    Interviews you on site type, look and pages, then builds accessible Astro, React or plain-HTML mockups with Tailwind CSS v4 that follow your CLAUDE.md, DESIGN.md and BRIEF.md. No Figma file needed.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Create Mockup — UI From Scratch, Matched to Your Stack

    Create Mockup — UI From Scratch, Matched to Your Stack

    Example session with this skill installed

    Create a landing page for a small specialty coffee roaster.

    • Read your context and instructions
    • Compiled the create mockup
    • Generated the app files

    The skill first reported the stack it found: "Astro v7, Tailwind CSS v4 (@tailwindcss/vite), icons: @lucide/astro, linter: Biome, CSS: src/styles/global.css, images: src/assets/images/", no DESIGN.md, no BRIEF.md, and all six companion skills detected. Then it asked:

    StepQuestionAnswer
    SetupNew project or add to existing?Start fresh
    SetupHow much creative direction?Avoid generic AI look
    SetupJapanese-language site?No
    SetupHow far to break from default layouts?Balanced
    SetupRun a requirements interview first (brief-me)?Yes, which wrote BRIEF.md: B2C, 20s to 30s and price-sensitive, freshness and value, goal = purchase or subscribe, avoid flashy effects
    SetupHow busy should decoration be? (this question is now asked only for Japanese sites)Restrained
    DesignColors, mood, imagesWhite base with one bold accent, modern and minimal, free stock photos from Unsplash
    SectionsCore and optional sectionsHero with CTA, featured beans and prices, how it works, subscription plans, roastery story, visit the roastery
    FinishWhat next?Finish here

    What it decided from those answers

    DecisionResult
    LayoutA zigzag carried by typography and numbers instead of alternating image rows. The hero is an asymmetric split, not 50/50, with a framed photo offset lower.
    ConversionThe header button is not the only CTA: an inline link sits at the last "how it works" step. In the plans, one button is filled and the other two are outlined, so the anomaly is deliberate.
    ColorAchromatic base plus one accent (cherry red), defined as @theme tokens (--color-surface, --color-surface-alt, --color-ink, --color-muted, --color-border, --color-accent). White background, because the brief gave no reason to tint it.
    RestrainedExtra whitespace between sections, the accent used only for buttons, one eyebrow label and link hovers, and no motion beyond a color change on buttons.

    What it built: Layout.astro (lang set to en), Header.astro, Footer.astro, index.astro with six sections, the @theme block, and four optimized photos.

    Checks it ran: duplicate classes, default-palette classes, space-* / divide-*, typed uppercase and aria-labelledby references all returned 0 hits. Biome, astro check and the build passed, and the dev server was left running.

    Three problems found while checking the page in the browser, and fixed:

    1. The hero photo dropped below the text: grid auto-placement pushes an overlapping item to the next row. Fixed by placing both items on the same row explicitly.
    2. The heading ended with one word alone on the last line, and the first attempt at an overlap covered its final period. Headings now use text-wrap: balance and paragraphs text-wrap: pretty, and the overlap was removed.
    3. After that, a scan for one-word last lines found none at 390px or 1440px, with no horizontal overflow.

    Not covered: the navigation list is hidden below the md breakpoint (no mobile menu yet), the copy, prices and address are placeholders, and the photos are Unsplash stock images.

    global.css

    ZIP · project files

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A skill that builds UI mockups from scratch — no Figma file needed — after a short interview, in the style your project already uses. Ask an agent for "a landing page" and you tend to get the same centered hero, the same purple gradient, and code that ignores how your project is set up. This skill puts a few batched questions in front of the build (site type, look, pages, sections, images), detects your stack first, follows your project files, and finishes by checking the result and starting the dev server.

    What it does

    • Detects the framework, router, icon library, style utilities and linter from package.json (Astro, React, or plain HTML with Vite), then reads only the reference for that mode
    • Asks about site type, design style, page type and sections in one or two batched questions, not a long form
    • Defines your colors as Tailwind CSS v4 @theme tokens, and avoids token names that collide with built-in utilities
    • Treats DESIGN.md as the source of truth when you have one, and uses BRIEF.md as requirements context when it exists
    • Builds accessible markup: nav lists, labelled inputs, uppercase applied as a style instead of typed in the HTML, no dangling aria references
    • Checks for duplicate and conflicting classes, runs your linter and build, and starts the dev server

    How it works

    1. Reads your project: package.json, CLAUDE.md, and DESIGN.md / BRIEF.md if present, then reports the stack it found.
    2. Asks the setup questions: new project or an addition to an existing one, and how much creative direction you want.
    3. Asks about the design: site type, colors and mood, images, pages and sections.
    4. Builds: cleans up starter demo content, then implements the pages.
    5. Checks and runs: class checks, lint and build, then starts the dev server and gives you the URL.

    Works with (all optional)

    If one of these is installed, the skill offers the matching question. If it is not, the question is skipped.

    • Brief Me (brief-me): a requirements interview that writes BRIEF.md first
    • Escape AI Slop (cognitive-ui-design): layouts that break the centered-hero default
    • Japanese Web Guideline and Japanese LP Seasoning: typography, carousels, scroll reveals and vertical text for Japanese sites
    • Code to Figma: export the finished page to Figma
    • Anthropic's frontend-design plugin: a bolder design direction

    Requirements

    Tailwind CSS v4 is required. The skill stops and points you to the install guide if the project uses Tailwind v3 or no Tailwind at all. You also need a project with a package.json and a CLAUDE.md in the project root (run /init first if it is missing), and Node.js. Stock photos are downloaded from Unsplash to your image folder, which needs internet access; you can choose solid-color placeholders instead. It does not set up a new project for you.

    Live demo

    A page built with this skill from one short request, a landing page for a fictional coffee roaster: https://create-mockup-demo.gaspanik.workers.dev/

    The questions it asked and the decisions it made are in the example session above.

    It needs a project that already has package.json and a CLAUDE.md; it does not scaffold one. Tailwind CSS v4 is assumed (colors go in @theme, not tailwind.config.js). Stock photos come from Unsplash and need internet access. It builds from your answers, not from a Figma file. The optional companion skills change the questions it asks, not what it can build. Visual polish beyond the first pass (mobile refinements, copy tuning) is left to follow-up requests.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Requires

    Unsplash
    Node.js

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Write Files

    Allowed Hosts

    images.unsplash.com

    File Scopes

    src/**
    public/**

    This skill reads and writes the project's own files (pages, components, the CSS entry) and runs the project's lint, build and dev-server commands. Its only outbound request is downloading stock photos from images.unsplash.com with curl, and only if you choose stock photos over placeholders. Pages it writes may load Google Fonts in the browser. To detect optional companion skills it checks whether their folders exist under the project's and your user-level .claude/skills directories; it does not read their contents.

    Listed1 month ago
    Updated1 month ago

    What's inside

    Frequently Asked Questions

    No. It builds from a short interview. If you already have a Figma URL, an implement-from-Figma workflow is a better fit; this skill is for when you don't.

    No. You need a project with a package.json and a CLAUDE.md first (run /init for the CLAUDE.md). The skill points you to the Vite setup guide if there is no package.json.

    Yes. Choose "add to the existing project" and it reads your @theme tokens and conventions, then builds the new page in the same style.

    Brief Me, Escape AI Slop (cognitive-ui-design), Japanese Web Guideline, Japanese LP Seasoning, Code to Figma and Anthropic's frontend-design plugin. When one is installed, the skill offers the matching question; when not, it skips it. It works fine without any of them.

    Only to download stock photos from Unsplash, and only if you choose stock photos. Choose solid-color placeholders and nothing is downloaded.

    It is written for Tailwind CSS v4: colors are defined in @theme, and there is no tailwind.config.js step.

    New in Frontend & Web Apps