agentsclimarketplace

Ux psychology skill

Skill vivaldi007/ux-psychology-skill

Use when designing, building, or reviewing any user-facing interface — forms, onboarding, signup flows, pricing pages, dashboards, CTAs, loading/empty/error states, checkout — or when diagnosing conversion drop-off, form abandonment, or user-trust problems.From its SKILL.md

Install
npx -y skills add vivaldi007/ux-psychology-skill

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

7.2 KB, ~1.6k tokens by cl100k_base, as published. Nobody here has run it

UX Psychology Principles

Overview

21 evidence-based psychology principles for building interfaces, plus Nielsen's 10 heuristics as an audit layer. Each principle in reference.md has a rule, its psychological mechanism, concrete implementation moves, and guardrails separating the honest version from the dark-pattern version.

Priority order when principles conflict: usability > honesty > persuasion. A persuasion pattern never justifies breaking a usability rule or a trust rule. Usability laws are load-bearing; persuasion is decoration on top.

How to Use

  1. Before designing: scan the Quick Reference below, then read the matching numbered sections in reference.md for mechanism + implementation + guardrails of every principle you apply.
  2. While building: apply interaction/perception laws (7–21) first; layer persuasion (1–6, trust) on top.
  3. Before declaring any screen done: run the Review Checklist at the bottom of this file. It is the gate.

Quick Reference

Persuasion & motivation (reference Part I)

#PrincipleRule
1Smart DefaultsPre-select the sensible choice for every field; never default anything costing money, data, or consent
2Goal GradientNever start progress at 0% — count what the user already did; ≤7 visible steps; at the end, flip to "1 step left"
3ReciprocityDeliver standalone value before any signup gate; never blur/lock results the user generated — gate depth (export, history, hi-res), not the artifact
4IKEA / EndowmentLet users build before the account wall; button says "Continue"/"Save", never "Sign up"; persist pre-signup work through account creation
5Loss AversionFrame the concrete, personal, true cost of inaction; no fake urgency, no confirm-shaming
6AnchoringNever show a price in isolation; control the reference point (tier anchor, % of related purchase, monthly-equivalent); real anchors only

Interaction laws (reference Part II)

#PrincipleRule
7Fitts's LawPrimary actions big (≥44px) and near attention; destructive actions distant, never adjacent twins of Confirm
8Hick's LawLimit choices per screen; progressive disclosure; highlight one recommended option
9Jakob's LawFollow platform conventions; innovate on value, not controls
10Miller / ChunkingGroup in 3–5; never require remembering info across screens
11Tesler's LawAbsorb complexity in code (auto-detect, parse, infer) — never push it to the user
12Doherty ThresholdAcknowledge every interaction <400ms; manage perceived speed beyond that
13Postel's LawAccept any plausible input format and normalize silently; output precise and consistent
14ParetoPrime placement and polish for the critical 20% of features

Perception & memory (reference Part III)

#PrincipleRule
15GestaltVisual grouping must match logical grouping; labels closer to their own field than the neighbor's
16Von RestorffExactly one visually distinct element per screen
17Serial PositionMost important items first and last in any sequence
18Peak-EndEngineer one peak moment; polish every ending — confirmations, errors, cancellation
19ZeigarnikVisible unfinished states pull users back — only for things they actually started
20Aesthetic–UsabilityPolish buys tolerance and trust, but masks problems in testing — measure task completion, not opinions
21Banner BlindnessCritical info must never look like an ad or sit in banner/sidebar/carousel-shaped containers

Also in reference.md: Nielsen's 10 heuristics (Part IV), perceived performance — skeletons, optimistic UI, labor illusion (Part V), forms & microcopy — outcome-stating buttons, inline validation, never blame the user (Part VI), trust & social proof (Part VII), cross-cutting principles.

Common Mistakes

  • Blurring the user's own generated result behind a signup modal. Feels like persuasion; reads as extortion. Show the real result, gate the depth (#3).
  • Onboarding progress at "Step 1 of N" with nothing pre-filled. Count signup/input already given as completed progress (#2).
  • Pre-signup work lost at the account wall. Persist state (localStorage → migrate on account creation); losing their work is the pattern's worst failure mode (#4).
  • Animation-heavy specs with no accessibility notes. Shimmer, staggered reveals, and countdowns must respect reduced-motion, keyboard, screen readers, and ≥4.5:1 contrast. Accessibility is a floor, not a feature.
  • Persuasion patterns on top of a confusing flow. Perfect anchoring on a broken checkout still loses; fix usability first.

Review Checklist (the gate)

Persuasion & motivation

  • Every form field with a knowable answer has a smart, honest default
  • Onboarding progress never starts at 0%; total visible steps ≤ 7
  • User receives standalone value before any signup/email gate
  • Users can build/customize something before the account wall; pre-signup state persists through signup
  • Action prompts name a concrete, true cost of inaction; no confirm-shaming, no fake urgency
  • No price appears without a controlled, honest reference point
  • Social proof is real, specific, and recent
  • Every persuasion pattern passes the test: "Is the underlying claim true?"

Interaction & layout

  • Primary actions are big, close, and singular; destructive actions are distant and confirmed or undoable
  • Choices per screen are limited; complexity is progressively disclosed, not dumped
  • Navigation, icons, and flows follow platform conventions
  • Content is chunked in groups of 3–5; nothing must be remembered across screens
  • Complexity the code can absorb (formats, parsing, inference) never reaches the user
  • Exactly one visually dominant element per screen; visual grouping matches logical grouping
  • Important info never looks like an ad

Feedback & performance

  • Every interaction acknowledges within ~400ms; longer work shows skeleton/progress/optimistic state
  • Skeletons mirror real layout; optimistic updates have a rollback state
  • Waits over ~5s explain what's happening
  • Undo exists for reversible actions; confirmations reserved for irreversible ones

Forms, errors & copy

  • Only necessary fields; inline validation on blur; input never lost on error
  • Errors say what happened and the specific fix, in human language, next to the field
  • Buttons state outcomes, not verbs ("Show 12 results," not "Submit")
  • Anxiety-answering microcopy sits next to every commitment point

Final audit

  • Screen passes Nielsen's 10 heuristics (reference.md Part IV)
  • Works with keyboard, screen reader, and reduced motion; contrast passes WCAG AA

What ships with it: 6 files

29.8 KB alongside SKILL.md

.claude-plugin/

Keep looking

Skills are one crate of 326,750. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.