← Search

Design Taste Frontend
Design 🧪 Install ran in our test sandbox · Nov 11, 2026 📌 Pinned version: commit b9ca2fd
Install test only, on the date shown. This does not guarantee the code is safe.
⚙ Needs: none — no account, API key, or paid plan required.

Design Taste Frontend

Anti-slop frontend skill for landing pages and portfolios — brief inference, three design dials, per-case presets, strict pre-flight check

At a glance
What
Anti-slop frontend skill for landing pages and portfolios — brief inference, three design dials, per-case presets, strict pre-flight check
Cost
Free
Needs
none — no account, API key, or paid plan required. (Using the skill later wants Node 22+ and Chrome, Edge, or Chromium for its own scripts; installing it needs neither.)
Install
Copy the installer prompt below into your Muse — your agent does the rest.

Version:

Quick answer

Design Taste Frontend is a free Muse skill offering an anti-slop frontend skill for landing pages and portfolios, with brief inference, three design dials, per-case presets and a strict pre-flight check. As a substantial rewrite of the original, nothing fires automatically: the agent first reads the brief, covering page kind, vibe, references and audience, then declares a one-line design direction.

⌁

Install

Copy the install package below, then paste it into Muse🧪 Install ran in our test sandbox · Nov 11, 2026📌 Pinned version: commit b9ca2fd

Before you start: tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead. Prerequisites: none — no account, API key, or paid plan required. (Using the skill later wants Node 22+ and Chrome, Edge, or Chromium for its own scripts; installing it needs neither.) Install "Design Taste Frontend" for me — upstream name "tasteskill v2", an anti-slop frontend skill: art direction for landing pages, marketing sites, portfolios, product pages, and any public-facing web page, in any stack (HTML/CSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — and now also for app screens: dashboards, admin and analytics views, settings, billing, inboxes and queues, wallets, editors and builders, and the detail page of one object. Its page method is distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day) and runs in ten steps: (1) read the brief and inventory the subject — eight to twelve things that belong to this client and no competitor, above all what imagery truly exists; (2) write the argument: the page as claims with their proof, one line per section, which decides which sections exist; (3) draw the direction: a bundled script deals three of twenty researched directions from the skill's library, you choose the one that can carry the subject's idea, and the script deals its type system, palette, hero, and section blueprints; (4) find the idea: one sentence built from the subject itself, plus one live thing the visitor can do; (5) write the words: the full copy deck before any markup, every headline fitting in two lines; (6) build it, still: tokens, the hero exactly as dealt, each section from its blueprint, and real imagery — the project's own, generated, or real pictures of the subject's world searched in open collections (Openverse, Wikimedia Commons, the Art Institute of Chicago, the Met; the skill's assets script does this over the network, no API keys, and records each picture's credit) — with a floor of at least four real pictures; (7) gate: a linter plus a real-browser inspector at 1440, 768, and 390 pixels measure the page (headline line counts, type sizes, colours, contrast, overflow) and every FAIL must be fixed; (8) look: critique the inspector's screenshots against a juror's rubric, fix the top three problems, look again, three rounds at most; (9) move: the motion layer — a signature move, staggered entrances, feedback on every control, one moving picture; (10) report: the direction dealt, the idea, the final scores, every found picture with its credit, and any sample content to swap. Over all of it, a strict house style the inspector measures: headlines of two lines or fewer, no small labels above headings, no gradients, no shadows on things at rest, no icon-topped feature cards, no monospace outside real code, no index numbering, at most two typefaces. App screens do not follow that page run: they follow their own run in the bundled references/app-screens.md, distilled from a study of 23 real product screens — inventory the screen's data and jobs and write a small dataset whose totals are computed from its rows; name the screen's one question and its one finding; choose a style: the draw script's app mode (draw.mjs --app) deals up to three of seven styles from the library, you pick one, choose the face it takes from the brief and one signature element only this product has, and the script deals its tokens, shell, and screen patterns; build the shell first (the bundled assets/app.css is a neutral start), then render every block and chart from the dataset (the bundled assets/charts.js, never a chart library's defaults, never placeholder data or invented vanity numbers), then build every empty, zero, loading, and error state; check with the linter and inspector in app mode (lint.mjs --app, inspect.mjs --app) at the same three widths and fix every FAIL, then answer the written critique questions, fix the worst three, three rounds at most; and report the style dealt, the face and signature element, each screen's one question, and the sample data to replace. Its app house rules differ from the page rules: quiet borders, small labels, and repeated panels are normal there; colour goes on the data, not the chrome; KPIs are type, not tiles; surfaces are flat; charts carry almost no chrome; and gradients, glows, icons in tinted squares, rainbow charts, and invented vanity numbers are banned. A product page that shows its app stays a page, but each interface fragment on it is built as a real piece of that app — the app style dealt once with the page's palette and type kept, the bundled app stylesheet and chart helpers, real data that adds up. Repository: https://raw.githubusercontent.com/leonxlnx/taste-skill/b9ca2fd8047f4c72d7da90ebc1c1203e57c72cd2/skills/taste-skill/SKILL.md Pinned version: install the code at commit b9ca2fd8047f4c72d7da90ebc1c1203e57c72cd2 exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. [FINGERPRINTS] SKILL.md at the pinned commit: SHA-256 d955dae4ba8ab2d994f4bb06b4bf412a1d9875433a8597939c391f4bae91dd7a, 38,458 bytes (38,420 characters). If the fetched file does not match, it changed since the review: stop, tell me what you found and let me decide. Updated October 19, 2026: Skill Harbor regenerated this install prompt from the upstream file of the day. Upstream extended tasteskill v2 (SKILL.md grew by about a third): the skill now covers app screens as well as pages. It adds a whole second route — the bundled references/app-screens.md (read it whole before starting an app screen; the page run does not apply to them), the draw and inspector scripts' app modes (draw.mjs --app, inspect.mjs --app, lint.mjs --app), app assets (assets/app.css, assets/charts.js), and app styles, shells, and screen patterns in the library. Correction to the previous prompt: it said "Not for dashboards or app screens" — that exclusion is gone; dashboards and the other app screens listed above are now in scope, under their own run and rules. What changed for you: the pinned commit above is the new version; the page run, its ten steps, and its house style are unchanged, and the install still brings the whole skill directory (instructions, references, scripts, assets, and the direction library), now including the app-screens reference and assets. 1. Download the repository at the pinned commit into a temporary folder (the commit archive, or a clone checked out at that commit). The skill is the skills/taste-skill/ directory — install that entire directory, not just its SKILL.md. 2. Compare its SKILL.md with the fingerprint from Skill Harbor's review of 2026-10-19 (see [FINGERPRINTS] above): SHA-256 d955dae4ba8ab2d994f4bb06b4bf412a1d9875433a8597939c391f4bae91dd7a, 38,458 bytes. If it does not match, the file changed since the review: do not install it yet — tell me what you found and let me decide. 3. Read SKILL.md and tell me in plain words: what the skill does when used, whether it connects to a service, whether it reads or sends any personal information or keys, and anything that seems unusual. 4. When I say yes, copy the entire skills/taste-skill directory into my skills folder, in a folder named "leonxlnx-design-taste-frontend", keeping its support folders (references/, scripts/, assets/, library/) intact. 5. Check the copy (valid frontmatter, files in place), start a fresh session so the host re-indexes the skill, and tell me what was installed, where, and what is left for me to do (bring a page to build or redesign, or an app screen to build; for the skill's own gate and picture search when I use it: Node 22+, a Chrome-family browser, and network access). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Rules: only touch the temporary folder and the install target. Nothing is sent anywhere during installation; the picture search described above happens only later, when the skill is used, never during installation. If something seems off, stop and ask me.

🔒 Security, provenance and what this touches
🧭 What this touches
Based on the install prompt we scanned, this skill does not run commands, make third-party network calls, install packages, write files, or ask for API keys.
Install prompt analysis · 2026-10-13
🔒 Security and provenance
📌 Pinned version: commit b9ca2fd · 2026-10-19
🧪 Install ran in our test sandbox · Nov 11, 2026
👁 Observed during the install test · Nov 11, 2026 : no network contact
🔎 Last checked · Oct 19, 2026 : no change detected in the files linked by this install prompt
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
Wanna be extra careful?

The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?

How to check a skill before installing →

✓

Verification certificate

Dated facts about what was checked for this listing, and how. This is not a safety certification.

Pinned version
b9ca2fd · 2026-10-19
Sandbox install test
Install ran in our test sandbox · Nov 11, 2026
Creator identity
Not verified
Fingerprint watch
Last checked · Oct 19, 2026 : no change detected in the files linked by this install prompt
Install prompt analysis
2026-10-13 : No commands, network calls, packages, file writes or API keys detected.

Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify

?

Questions

How do I install a skill?

Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.

Where does my money go?

Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.

What does the ✓ next to a creator’s name mean?

It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.