Home / All Skills / Design & Motion / redesign-existing-projects
English (EN)简体中文 (ZH)日本語 (JA)
Design & Motion · leonxlnx/taste-skill

redesign-existing-projects

Upgrades existing websites and apps to premium quality. Audits current design, identifies generic AI patterns, and applies high-end design standards without breaking functionality. Works with any CSS framework or vanilla CSS.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit
Author leonxlnxGitHub leonxlnx/taste-skill ↗Stars 85698License MIT（仓库根 LICENSE 文件；GitHub API spdx MIT；Copyright (c) 2026 Leonxlnx）Commit ccbc15639c
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Three-beat workflow Scan→Diagnose→Fix: First reads the codebase to identify tech stacks and current status, runs audits to list issues, and finally applies targeted fixes solely on the existing stack, forbidding rewrites.

skills/redesign-skill/SKILL.md
**Fix** — Apply targeted upgrades working with the existing stack. Do not rewrite from scratch. Improve what's there.
"Do not rewrite from scratch" is the constitutional constraint across the skill, separating "legacy modernization" from greenfield draft-generation skills.

The core is an "Anti-pattern Audit Library": Lists common generic AI flaws across eight categories (Typography / Color and Surfaces / Layout / Interactivity and States / Content / Component Patterns / Iconography / Code Quality), pairing each entry with "Problem → Fix".

skills/redesign-skill/SKILL.md
**Purple/blue "AI gradient" aesthetic.** This is the most common AI design fingerprint. Replace with neutral bases and a single, considered accent.
Audit items are not vague suggestions but "fingerprint → remediation" pairs: first identifying common AI design fingerprints, then providing specific replacement directions—binding diagnosis directly to prescription.

Fixes are ordered by "Impact/Risk ratio": Fix Priority defines a 7-step sequence from font replacement to polishing, prioritizing actions with maximum visual gains and minimum risk.

skills/redesign-skill/SKILL.md
1. **Font swap** — biggest instant improvement, lowest risk
The "Fix Priority" directive begins with "Apply changes in this order for maximum visual impact with minimum risk:", followed by 7 steps: font replacement → palette cleanup → interactive states → layout spacing → component replacement → missing states → typography polish. Gives the agent a deterministic execution order to prevent wasting cycles on low-impact areas or recklessly altering high-risk parts.

"Do not break functionality" is enforced by four hard rules: no framework/style library migrations, test every change, check dependency files before importing libraries, use vanilla CSS if no framework exists, and keep changes small and focused.

skills/redesign-skill/SKILL.md
Do not break existing functionality. Test after every change.
Testing actions are executed by the host agent on the user's project (host-authorized tool surface); the skill itself contains no scripts.

The Upgrade Techniques toolkit (Typography/Layout/Motion/Surface Upgrades) provides advanced directions beyond merely "removing cliches", ensuring transformations go further than bug fixing.

skills/redesign-skill/SKILL.md
**True glassmorphism.** Go beyond `backdrop-filter: blur`. Add a 1px inner border and a subtle inner shadow to simulate edge refraction.
Complements the audit checklist: audits handle subtraction and correction, while Upgrade Techniques handle addition.

The "Strategic Omissions" checklist explicitly supplements what AI typically overlooks: legal links, back navigation, 404 pages, form validations, skip-links, cookie consent, etc.

skills/redesign-skill/SKILL.md
**No legal links.** Add privacy policy and terms of service links in the footer.
This section is titled "### Strategic Omissions (What AI Typically Forgets)", followed by several bullet points. It integrates "completeness" into the upgrade scope—many UIs look beautiful but have broken workflows; this section seals functional and UX gaps.

2Core Capabilities

01Existing codebase scanning and identification (frameworks, styling systems, existing design patterns)
02Eight-category anti-pattern auditing (typography/color/layout/states/content/components/icons/code-quality) with actionable fixes per item
03Strict compatibility rules: no tech stack migrations, no rewrites, test every change
04Prioritized refactoring execution plan (7-step impact/risk ratio)
05Interactive state completion: hover/active/loading/empty/error/focus/navigation states
06Content-layer de-AI and realism enhancement (realistic data, authentic names, banned buzzwords)
07Strategic omissions coverage: a11y, legal notices, 404, form validation, accessibility
08Advanced upgrade techniques (authentic glassmorphism, parallax, spotlight borders, variable font animations, etc.)

3External Dependencies

KindDependency / Endpoint
networkpicsum.photos（占位图服务）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Can be auto-loaded
For Administrators
Register source and version
Evidence & Triggers (File Paths & Excerpts)
  • SKILL.md: Pure prompt instructions and rule text without script execution or external impact
Audit Notes: Pure prompt-based single-file skill: no script execution, no external network requests, no credential reading, and no direct file writing by the skill itself (reading code, modifying files, and running tests are executed by the host agent within the user project per instructions via host authorization). The only network domain, picsum.photos, appears in placeholder image URLs within delivered code, not as runtime network activity from the skill.

5Second-Pass Verification

  • [ok] 目录资产面（含隐藏/未跟踪文件） — git ls-files + ls -A 双查：该 skill 目录仅 1 个文件 SKILL.md，无 scripts/references/模板/数据/隐藏文件，无可执行代码。
  • [ok] 执行类 token 扫描 — 逐文件扫描 bash/exec/child_process/curl/wget/fetch/axios/process.env/API_KEY/.env/fs./writeFile/.github 等零命中（brutalist 一处 'sans-serifs.' 单词结尾 fs. 为误报，人工复核排除）。
  • [ok] 引文逐条 grep -F 字节复核 — 本 JSON 所用全部 quote 均以 grep -F 在 pin commit 工作树中逐条验证存在，无 paraphrase 冒充。
  • [ok] 元数据与 pin 核对 — 本地 git HEAD == 任务 pin ccbc15639c97057cbfcf32ecebc38ef716e4bb37（README 显示该提交为 docs/sponsor 修订 #102，2026-08-24）；GitHub API 实查 stars=85698、MIT、pushed_at=2026-08-24T15:23:56Z。frontmatter name 与任务给定名一致。
  • [ok] 功能声明 vs 实际能力夸大检查 — description 声称均由 SKILL.md 正文纯 prompt 机制兑现，无脚本可调用、无越界承诺（详见本文件 how_it_works 引文）。
  • [ok] 『without breaking functionality』是否有保障机制 — Rules 段四条硬规则（不迁移栈/每改必测/引库查依赖/保持小改动）+ Fix Priority 的风险排序共同兜底；虽为文本约束，但机制完整、无虚假承诺。
  • [ok] audit 分类是否有漏项/夸大（声称 8 类其实不足） — 重读核对标题：Typography/Color and Surfaces/Layout/Interactivity and States/Content/Component Patterns/Iconography/Code Quality 恰好 8 类，另加 Strategic Omissions 独立一节，与正文结构一致。
  • [ok] picsum.photos 引用的性质复核 — 原文明确『when real assets are not available』的占位图用途，且仅此一处 URL；无隐藏网络调用。

6Verdict & Scenarios

  • Disciplined restraint for legacy modernization: no rewrites, no stack migration, test every change, minimizing breaking risks.
  • Extensive and actionable anti-pattern library: 8 categories × dozens of "fingerprint → fix" pairs for model execution.
  • Impact/risk prioritized execution plan, avoiding wasted effort on low-value edits.
  • High completeness: covers legal links, skip-links, 404 pages, cookies, and import hallucinations, ensuring production-ready deliverables.
  • Framework-agnostic and auditable: explicitly preserves the existing stack with small-step commits, ideal for human-in-the-loop review.
    • Best for: Ideal for scenarios with existing website/app codebases needing visual and completeness upgrades to a premium level without switching tech stacks: legacy marketing sites, SaaS dashboards, and modernizing legacy projects; particularly suited for collaborative workflows requiring "audit first, review by human", compatible with any CSS framework or vanilla CSS.
    Not for: Not suitable for greenfield scratch generation (no codebase to audit; use taste-skill / high-end-visual-design instead); not suitable for full rewrites or framework migration projects; not suitable for pure design ideation without code implementation; not suitable for high-risk core systems lacking tests where zero breakage is demanded—prompt rules cannot replace human verification.
    Install Agent-Ready
    ① Verified Mirror 2026-09-09
    Option A · Direct DownloadDownload redesign-existing-projects.tar.gz
    sha256: de9382abc8e410ed…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit 3cca18b368.