AGENT SKILL

ux-design

Load before designing or reviewing any flow, form, onboarding, navigation or content for real users, especially low-literacy, older, neurodivergent, child or other specific audiences. Covers research and usability tests, heuristic reviews, personas and jobs to be done, journeys and service blueprints, information architecture, UX laws, SUS, SEQ and HEART metrics, plain language and reading level, and microcopy and error messages.

License 许可
MIT
最近更新
2026-11-10
Stars
★ 2
Forks
0

01 前言 Front Matter

name
ux-design
license
MIT
metadata
{ "agent-attribution": "computer=NMG-MBP-M5; tool=Codex; version=codex-cli 0.160.0; timestamp=2026-10-27 07:43:23 AST", "version-timestamp": "2026-10-27 07:43:23 AST" }
description
Load before designing or reviewing any flow, form, onboarding, navigation or content for real users, especially low-literacy, older, neurodivergent, child or other specific audiences. Covers research and usability tests, heuristic reviews, personas and jobs to be done, journeys and service blueprints, information architecture, UX laws, SUS, SEQ and HEART metrics, plain language and reading level, and microcopy and error messages.

02 资源文件 8

03 SKILL.md 正文

UX design

Excellent UX work starts from a decision that evidence can change, uses the cheapest method that answers it, and ends in structures, flows and words that people complete without help. Every claim is labeled observed, inferred or assumed; nothing is invented to look complete.

Defaults

Decision Default Change when
Qualitative usability round 5 participants per distinct user group, then fix and retest (NN/g) 2+ distinct groups: 3-5 each; rare-but-severe tasks: add participants
Quantitative usability benchmark About 40 participants per group (NN/g) You need a tighter interval: size from the target margin of error
Interviews 5-8 per segment; stop at saturation, often near 12 in total Very heterogeneous segments or markets: more per segment
Card sort / tree test Open sort 15-20 people; tree test about 50 per tree variant (30 minimum) Several audiences with different mental models: per audience
Survey About 385 completes for +/-5% at 95% confidence; about 100 for +/-10% Comparing segments: that n per segment
Expert review 3-5 evaluators, Nielsen's 10 heuristics, severity 0-4 Solo or AI review: label it "expert inspection", never user evidence
Satisfaction SEQ after each task, SUS at session end (68 = average) Very short intercept: UMUX-Lite
Response time 0.1 s feels instant, 1 s keeps flow, 10 s holds attention; feedback within 400 ms Long jobs: progress, cancel and background completion
Hierarchy Broad and shallow: 5-9 top-level groups, 3 levels or fewer for most sites Large catalogs: faceted navigation plus search
Reading level Grade 6-8 for consumers; GOV.UK writes for a reading age of 9 Experts: keep precise terms, define each once
Sentences Average 15-20 words; split anything over 25 Legal text that must be verbatim: add a plain summary above it
Personas 3-5 behavioral personas; proto-personas labeled HYPOTHESIS One clear audience: one persona plus JTBD statements
Flow coverage Every flow: entries, happy path, alternatives, errors, empty, cancel, recovery, exit Never relax

Workflow

Paths in commands are relative to this skill's directory; resolve them to absolute paths before running.

  1. Frame the decision. Write what will change depending on the answer, then 3-5 research questions. A question no answer could change is cut.
  2. Use existing evidence first. Analytics, search logs, support tickets, reviews, sales notes, prior studies, competitor flows. Record each source with an ID and date.
  3. Pick the method and size the sample with the matrix in research methods. Write the plan: questions, method, participants and screener, consent, tasks or guide, analysis, timeline.
  4. Run it honestly. With no participants available, deliver the protocol plus a heuristic review labeled as expert inspection. Never fabricate participants, quotes, counts or emotions; synthetic users may draft hypotheses only.
  5. Synthesize. Observation, then insight, then recommendation, each with participant counts ("4 of 5") and severity per heuristics and laws.
  6. Model. Personas or JTBD (personas), journey map or service blueprint and user flows (journeys and flows), sitemap, taxonomy and labels (information architecture).
  7. Write the words with content design: labels, errors, empty states, confirmations. Real content goes into wireframes; no lorem ipsum.
  8. Prototype at the lowest fidelity that answers the question, test it, and measure with measurement: task success, time, errors, SEQ, SUS.
  9. Render and inspect. Draw flows and wireflows (Mermaid or FigJam), screenshot prototypes at phone and desktop widths, and review them with the design-critique skill before sharing.
  10. Hand off numbered requirements, flows with state IDs and IA to ui-design, web-design or native-app-design, each traced to a finding or a labeled hypothesis.

Decisions that change the answer

  • Research stage. Discovery (generative: interviews, field study, diary), design (evaluative: usability tests, tree tests), launch (summative: benchmarks, A/B tests). Do not run an A/B test to answer a "why" question.
  • Audience. Read the audience entry before designing (without a shell, open it directly): low literacy (grade 5-6 target, one question per screen, icons with words, 18px body, plain-language translations, ISO 24495-1), older adults, neurodivergent, children under 13; others via python3 ../../library/scripts/atlas.py query "<audience in plain words>". Older adults: longer sessions, larger text in stimuli, more recruiting lead time. Children under 13: parental consent plus child assent, short sessions. Low literacy: oral tasks, reading age 9 or lower. Expert power users: realistic data volumes and keyboard paths.
  • Industry. Health, finance and government add consent, data handling and recruiting constraints (special-category data under GDPR, HIPAA in US healthcare). Query industry.* entries for trust signals and regulation.
  • Market. Moderate in the participant's language; scale responses vary by culture (some markets rate higher or lower overall), so compare within a market, not across. Localize stimuli before testing.
  • Channel. Websites: tree tests and first-click tests are cheap and decisive. Native apps: test on real devices. Public kiosks and touchscreens: field observation at the site. Multi-touchpoint services: service blueprint.
  • B2B buying committee. Research buyers, admins and end users separately; their jobs and success criteria differ.
  • Traffic. Run the power calculation first (baseline 5%, +10% relative lift needs about 31,000 visitors per variant). If the site cannot reach that in 4-6 weeks, use qualitative testing and before/after benchmarks instead of an A/B test.

Common failures

  • Asking people what they want or would do. Fix: ask about the last time they did it, and observe behavior.
  • Leading or hypothetical questions ("Would you use a feature that..."). Fix: neutral, past-tense, open questions from the guide template.
  • Reporting "80% succeeded" from 5 users. Fix: report "4 of 5" and the adjusted-Wald interval (36-98%) if a number is demanded.
  • Personas made of demographics and stock photos. Fix: behavioral variables, goals and evidence links; no portraits that imply traits.
  • Journey maps with invented emotions. Fix: plot only evidenced emotions with source IDs; mark gaps as unknown.
  • Navigation that mirrors the org chart. Fix: top-tasks research, open card sort, then tree-test two candidate hierarchies.
  • NPS used as a usability metric. Fix: SUS, SEQ and task success for usability; NPS only as a relationship metric with its caveats.
  • "Submit", "OK" and "Invalid input". Fix: verb-plus-object buttons and errors that say what happened and how to fix it.
  • Testing with colleagues or the client team. Fix: recruit real target users with a behavior-based screener.

References

  • Surface placement: choose inline, popover, sheet, inspector, window or menu from task and lifetime.
  • Research methods: method matrix, interview guide, usability testing, card sorting and tree testing, surveys, diary studies, analytics and A/B tests, recruiting, consent, synthesis.
  • Heuristics and laws: Nielsen's 10 heuristics with checks, severity 0-4, UX laws with caveats, cognitive load.
  • Measurement: SUS scoring and grades, SEQ, UMUX-Lite, task metrics, NPS caveats, Google HEART.
  • Information architecture: inventories, organization schemes, breadth versus depth, labeling, taxonomy, validation.
  • Journeys and flows: JTBD, journey maps, service blueprints, user flows, wireflows, state inventories.
  • Content design: plain language, voice and tone, microcopy, error messages, readability, localization.
  • Personas: proto-personas versus research personas, building from behavioral variables, inclusive personas.
  • Atlas: python3 ../../library/scripts/atlas.py query "older adults healthcare portal in the UK" when audience, industry or market changes the plan; show <id> to read one entry.
  • Deep process for client projects: UX architecture, persona build (118-field schema), prototype evaluation, evidence research, content writing, UX deliverable contracts and the UX and content decision card.
  • Related skills: ui-design for components and states, web-design for sites, design-critique for the render-and-review loop.

Catalog IDs covered: ux, persona, prototype, research, content

04 同仓库更多 Skills

newmindsgroup
newmindsgroup/visual-design-kit

web-design

Load before planning, designing, auditing or launching any website, landing page, cinematic hero, microsite or above-the-fold layout, or a prompt for Lovable, Bolt or v0 to build one, including regulated (prescription drug, health, finance), multilingual or market-specific launches. Covers page types (home, product, pricing, about, blog), conversion and ethical persuasion, responsive and fluid layouts, Core Web Vitals, SEO and structured data, localization and language-law requirements, stack and CMS choice (Astro, Next.js, WordPress, Webflow, Framer) and launch QA.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

ui-design

Load before designing or specifying any interface screen or component, even a quick one: forms, tables, navigation, dashboards, onboarding, auth and passkeys, pricing, checkout, search and filters, settings, empty, loading and error states, notifications, consent and age-appropriate flows, and AI chat and agent interfaces, all to WCAG 2.2 AA. Also picks React UI libraries (toasts, drawers, command menus). Gives component specs, state matrices, sizes, density and keyboard and accessibility behavior.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

typography

Choose, pair, size and set type for brand, web, app, print, slides, signage and motion: modular scales, fluid clamp() type, measure, leading, tracking, OpenType features, font licensing and web font loading, and multiscript setting for CJK, Arabic, Hebrew, Devanagari and Thai. Use when selecting or pairing fonts, building a type scale or type tokens, fixing weak hierarchy or bad line breaks, setting long-form text or numbers, loading web fonts, or localizing type for another market.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

reference-reconstruction

Reconstructs the observed interface and behavior of a supplied app, site or screenshot set using a screen index, measurements, state and command maps, reusable components and matched-condition comparisons. Use when asked to reverse engineer, clone, reproduce or match a reference product, then adapt it if requested.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

print-packaging-social

Load before any print, packaging, label, barcode, social, out-of-home, slide-deck or email design question, even a quick one, and before making finished collateral files. Specs: bleed, safe zones, 300ppi, CMYK, PDF/X preflight, dielines, barcodes, social formats with 9:16 safe areas, billboard letter heights, 16:9 decks, bulletproof email. Renders token-driven brand books, sell sheets, cards, carousels and decks to PDF and PNG with a Canva editable twin.

★ 2
newmindsgroup
newmindsgroup/visual-design-kit

native-app-design

Design macOS, iOS, iPadOS and Android interfaces: native windows and surfaces, navigation, keyboard and pointer, accessibility, appearance modes, platform tokens, app icons and SwiftUI/Compose handoff. Use for native app screens, flows, store assets, Mac utilities or cross-platform choices. Mac implementation uses the optional BAMA workflow; preserve existing project OS targets and approved brand tokens.

★ 2