Agent skill
audit-ux-laws
Quellcode ansehen: yigitkonur/skills-by-yigitkonur
Installation
npx skills add yigitkonur/skills-by-yigitkonur --skill audit-ux-laws5
Installationen
EU-hosted inference API
Power your AI agent skills with open-source models.
Drop-in OpenAI-compatible API. No data leaves Europe.
MiniMax
MiniMax M3
$0.40 / $1.40
per M tokens
Z.ai
GLM 5.3 Flash
$0.20 / $0.60
per M tokens
MoonshotAI
Kimi K3
$4.00 / $18.00
per M tokens
DeepSeek
DeepSeek V4.1 Flash
$0.40 / $1.40
per M tokens
Audit UX Laws
Self-contained reference for the 30 Laws of UX (based on lawsofux.com by Jon Yablonski), tuned for an agent to find the relevant law, apply it to real code, and flag violations by severity. No live fetch — all law content is bundled in references/.
This skill does two jobs: proactive design (pick the right law while building UI) and review (audit an interface and report violations as CRITICAL / MINOR with concrete fixes). It works on UI only — components, layouts, navigation, forms, interactions, perceived performance. It is not a visual/CSS-correctness pass and it does not redesign brand or content.
The agent owns one decision: which laws are relevant to the thing in front of me? Everything downstream — the takeaway, the code pattern, the severity threshold — lives in the routed reference file. Read the router, open the one category file you need, apply it.
When to Use
Trigger on phrases and contexts like:
- "is this good UX?", "review this interface", "what's wrong with this design?"
- building or editing components, pages, layouts, navigation, menus, dropdowns, forms
- "too many options", "this feels overwhelming", decision points, pricing tiers
- touch-target sizing, click areas, spacing, visual grouping, hierarchy, what stands out
- loading states, response time, perceived performance, animation timing
- any named law: Fitts's, Hick's, Miller's, Jakob's, Doherty Threshold, Gestalt, Von Restorff, choice overload, chunking, cognitive load, Tesler's, Postel's, Pareto, Occam's, Zeigarnik, peak-end, goal-gradient, serial position
Do NOT use when
- The work is backend / API / data-layer with no user-facing surface.
- The ask is pure visual polish (exact colors, shadows, brand) — that is a CSS/design-system pass.
- You need a full persona-driven usability audit with saved findings → use
audit-ux-and-save-files. - You need a visual UI bug audit across pages with screenshots → use
audit-ui-and-save-files.
Step 1 — Route the symptom to a law
Match what you see to a row; open only the reference file in the last column.
| What you're checking | Laws | Reference file |
|---|---|---|
| Too many options (menus, dropdowns, nav, plans) | Hick's Law, Choice Overload | references/heuristics.md |
| Touch targets too small / far apart / hard to hit | Fitts's Law | references/heuristics.md |
| Beautiful-but-is-it-usable / first-impression trust | Aesthetic-Usability Effect | references/heuristics.md |
| Too much info on screen / dense forms / long lists | Miller's Law, Cognitive Load, Chunking, Working Memory | references/cognitive.md |
| What stands out / CTA emphasis / scanning order | Von Restorff, Serial Position, Selective Attention | references/cognitive.md |
| Onboarding / first-time users skip instructions | Paradox of the Active User, Cognitive Bias, Flow | references/cognitive.md |
| Grouping / spacing / layout relationships | Proximity, Similarity, Common Region, Uniform Connectedness, Prägnanz | references/gestalt.md |
| Users confused by an unfamiliar pattern | Jakob's Law, Mental Model | references/gestalt.md (Mental Model), references/principles.md (Jakob's) |
| Response time / loading / perceived speed | Doherty Threshold | references/principles.md |
| Forms too long / complexity | Tesler's Law, Parkinson's Law, Occam's Razor, Pareto | references/principles.md |
| Input handling / error tolerance | Postel's Law | references/principles.md |
| Motivation / progress / completion / memorable ending | Goal-Gradient, Zeigarnik, Peak-End Rule | references/principles.md |
If more than one row matches, open each named file once and apply every relevant law.
Step 2 — Apply or review
Designing (proactive): read the law's Key Takeaways + Frontend Code Implications, then build to them.
Reviewing (audit): walk the law's Review Criteria, classify each violation, and report it with the fix. Severity scale:
- CRITICAL — breaks usability, accessibility, or trust. Fix before ship. (e.g. touch target <44×44px mobile; >7 ungrouped options; primary CTA indistinguishable from secondary; no loading feedback >400ms; required cross-screen memorization.)
- MINOR — polish / optimization. (e.g. >6 colors, inconsistent spacing, missing smart default, no progress indicator, suboptimal CTA placement.)
Always connect the finding to the why (the psychological principle) and give a concrete, actionable fix — code pattern or specific change, never "improve the UX."
Step 3 — Follow Related links when warranted
Each law entry ends with Related pointers (e.g. Choice Overload → Hick's Law → Doherty Threshold). Follow them only when the current law's fix creates a new question the related law answers.
Quick Reference — all 30 laws
| Law | Category | One-liner | Related |
|---|---|---|---|
| Aesthetic-Usability Effect | Heuristic | Beautiful design is perceived as more usable | Von Restorff, Jakob's |
| Choice Overload | Heuristic | Too many options cause decision paralysis | Hick's, Doherty |
| Fitts's Law | Heuristic | Larger, closer targets are faster to hit | Doherty, Choice Overload |
| Hick's Law | Heuristic | Decision time grows with number of choices | Choice Overload, Miller's |
| Chunking | Cognitive | Group info into meaningful units | Miller's, Cognitive Load |
| Cognitive Bias | Cognitive | Systematic thinking errors shape decisions | Selective Attention, Peak-End |
| Cognitive Load | Cognitive | Minimize mental effort to use the interface | Miller's, Tesler's |
| Miller's Law | Cognitive | Working memory holds ~7±2 items | Chunking, Working Memory |
| Working Memory | Cognitive | Don't force recall across steps/screens | Miller's, Chunking |
| Selective Attention | Cognitive | Users filter to goal-relevant stimuli | Von Restorff, Cognitive Bias |
| Serial Position Effect | Cognitive | First and last items remembered best | Von Restorff, Peak-End |
| Von Restorff Effect | Cognitive | The distinct item stands out — use for CTAs | Serial Position, Selective Attention |
| Flow | Cognitive | Balance challenge & skill; remove friction | Doherty, Goal-Gradient |
| Paradox of the Active User | Cognitive | Users skip manuals and learn by doing | Jakob's, Mental Model |
| Law of Proximity | Gestalt | Near elements are perceived as grouped | Common Region, Similarity |
| Law of Similarity | Gestalt | Similar-looking elements are seen as related | Proximity, Uniform Connectedness |
| Law of Common Region | Gestalt | A shared boundary groups elements | Proximity, Uniform Connectedness |
| Law of Uniform Connectedness | Gestalt | Visually connected elements are most related | Common Region, Similarity |
| Law of Prägnanz | Gestalt | Users read complex shapes in their simplest form | Occam's, Similarity |
| Mental Model | Gestalt | Users carry prebuilt expectations from other products | Jakob's, Paradox of Active User |
| Jakob's Law | Principle | Users expect your site to work like the others they know | Mental Model, Prägnanz |
| Doherty Threshold | Principle | Keep response <400ms to hold attention & flow | Flow, Fitts's |
| Goal-Gradient Effect | Principle | Motivation rises as the goal nears — show progress | Zeigarnik, Flow |
| Zeigarnik Effect | Principle | Incomplete tasks are remembered — use progress cues | Goal-Gradient, Peak-End |
| Peak-End Rule | Principle | Experiences are judged by their peak and ending | Serial Position, Zeigarnik |
| Postel's Law | Principle | Be liberal in what you accept, strict in what you output | Jakob's, Tesler's |
| Tesler's Law | Principle | Irreducible complexity must live in design, not the user | Cognitive Load, Occam's |
| Occam's Razor | Principle | Remove elements until you can't without breaking function | Tesler's, Prägnanz, Pareto |
| Pareto Principle | Principle | ~80% of effects come from ~20% of causes | Occam's, Tesler's |
| Parkinson's Law | Principle | Work expands to fill available time/space — constrain it | Doherty, Goal-Gradient |
Reference routing
Load only the file for the category you routed to in Step 1.
| File | Read when |
|---|---|
references/heuristics.md |
Decision time, target acquisition, choice volume, aesthetic trust. |
references/cognitive.md |
Memory limits, attention, scanning, learning, biases. |
references/gestalt.md |
Visual grouping, layout relationships, shape simplification, expectations. |
references/principles.md |
Trade-offs, simplicity, response time, input handling, motivation, conventions. |
Guardrails
- Self-contained: never fetch lawsofux.com at runtime; bundled references are the source of truth.
- Don't load all four category files for every request — route, then apply only what's relevant.
- Every review finding needs a severity (CRITICAL/MINOR), a why (the law), and a concrete fix.
- Don't invent thresholds; use the numbers in the reference files.
- Severity numbers are usability heuristics, not a substitute for WCAG accessibility testing.
Installationen
Sicherheitsprüfung
Quellcode ansehen
yigitkonur/skills-by-yigitkonur
Mehr aus dieser Quelle
Give agents clean web context
Search and extract the public web as Markdown or structured JSON through one API or hosted MCP server.
Verwandte Skills
Mehr aus dieser Quelle: yigitkonur/skills-by-yigitkonur
So verwenden Sie diesen Skill
Install audit-ux-laws by running npx skills add yigitkonur/skills-by-yigitkonur --skill audit-ux-laws in your project directory. Führen Sie den obigen Installationsbefehl in Ihrem Projektverzeichnis aus. Die Skill-Datei wird von GitHub heruntergeladen und in Ihrem Projekt platziert.
Keine Konfiguration erforderlich. Ihr KI-Agent (Claude Code, Cursor, Windsurf usw.) erkennt installierte Skills automatisch und nutzt sie als Kontext bei der Code-Generierung.
Der Skill verbessert das Verständnis Ihres Agenten für audit-ux-laws, und hilft ihm, etablierte Muster zu befolgen, häufige Fehler zu vermeiden und produktionsreifen Code zu erzeugen.
Was Sie erhalten
Skills sind Klartext-Anweisungsdateien — kein ausführbarer Code. Sie kodieren Expertenwissen über Frameworks, Sprachen oder Tools, das Ihr KI-Agent liest, um seine Ausgabe zu verbessern. Das bedeutet null Laufzeit-Overhead, keine Abhängigkeitskonflikte und volle Transparenz: Sie können jede Anweisung vor der Installation lesen und prüfen.
Kompatibilität
Dieser Skill funktioniert mit jedem KI-Coding-Agenten, der das skills.sh-Format unterstützt, einschließlich Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider und anderen Tools, die projektbezogene Kontextdateien lesen. Skills sind auf Transportebene framework-agnostisch — der Inhalt bestimmt, für welche Sprache oder welches Framework er gilt.