ux-design

name	ux-design
description	Applies psychology-informed UX principles to design and review interfaces, flows, forms, navigation, onboarding, dashboards, and product journeys. Use for UX critiques, interaction decisions, heuristic reviews, accessibility-aware recommendations, and diagnosing confusion, overload, or misleading behavior.
license	See LICENSES.md
compatibility	Designed for Pi and Agent Skills-compatible coding agents. Core use requires only file access; browser access is optional for reviewing a live product or checking sources.
metadata	{"category":"design","source":"https://lawsofux.com/","source-author":"Jon Yablonski","source-license":"CC-BY-NC-ND-4.0"}

UX Design

Use UX laws as diagnostic lenses, not proof or commandments. Observed user behavior, accessibility requirements, domain constraints, and product evidence outrank a named principle.

Workflow

1. Inspect the context. Establish the intended users, task, device and input methods, environment, user familiarity, business constraints, accessibility needs, and available research or analytics. Inspect the real implementation and states when available.
2. Frame the problem behaviorally. Describe what users may fail to notice, understand, choose, remember, reach, recover from, or complete. Do not start by choosing a law.
3. Route directly. Use the reference routing below and load only the law groups relevant to the observed user problem.
4. Choose a small set. Usually 2–5 principles are enough. Name a law only when it materially explains the issue.
5. Translate each law into a design move. Tie it to a specific element, user action, expected benefit, tradeoff, and falsifiable validation method.
6. Check conflicts. Simplifying a novice flow may slow experts; adding emphasis may create noise; hiding choices may harm comparison; accepting flexible input may create ambiguity or security risk.
7. Separate evidence levels. Label findings as observed evidence, accessibility/platform requirement, research-informed principle, heuristic judgment, or hypothesis.
8. Prioritize. Rank by user harm, frequency, task criticality, confidence, and implementation cost—not by how many laws support the finding.

Output format

For each material finding, report:

Finding: [specific interface behavior]
User impact: [who is affected and how]
Principle: [one primary law; optional supporting law]
Rationale: [context-specific causal argument]
Recommendation: [smallest useful design change]
Tradeoff: [cost, conflict, or failure mode]
Validation: [method and observable success/failure signal]
Evidence level: [observed | requirement | research-informed | heuristic | hypothesis]
Source: [specific source URL when naming a law]

For implementation work, also include acceptance criteria and verify responsive, loading, empty, error, success, keyboard, touch, zoom, reduced-motion, and assistive-technology-relevant states.

Non-negotiable rules

• Prefer measured behavior over appeals to authority. Never say a law “proves” a design decision.
• Do not invent users, research findings, conversion lifts, accessibility results, or certainty.
• Do not use “Miller’s 7±2” as a menu-item limit or treat the Doherty Threshold as a universal performance standard.
• Preserve necessary complexity; remove extraneous effort rather than merely hiding information.
• Prefer recognition over recall and put memory burden on the interface where practical.
• Keep familiar conventions unless evidence justifies retraining costs.
• Do not rely on color, motion, hover, or pointer precision alone.
• Never recommend deceptive defaults, manufactured urgency, obstruction, hidden costs, forced continuity, or other dark patterns.
• Treat accessibility as an independent requirement, not merely another UX preference.
• Make validation capable of disproving the recommendation.

Reference routing

• Visual grouping, hierarchy, salience, or missed content: perception-and-visual-hierarchy.md
• Memory, attention, density, scanning, or recall: memory-attention-and-load.md
• Menus, options, defaults, prioritization, or decisions: choice-and-decision-making.md
• Targets, latency, forms, tolerance, or task time: interaction-and-response.md
• Progress, motivation, completion, interruption, or remembered journeys: motivation-and-journeys.md
• Conventions, onboarding, learnability, or irreducible complexity: models-learning-and-complexity.md
• End-to-end critique: review-checklist.md
• Attribution and provenance: sources.md and LICENSES.md

Final quality gate

Before delivery, confirm that every named principle maps to a concrete issue; recommendations account for users and constraints; accessibility was checked directly; hypotheses are labeled; conflicts are visible; citations are traceable; and the proposed test could reveal that the advice was wrong.