ux
Skill0 starsChanged 8 months agoMITView on GitHub
3-phase UX workflow: business clarification, textual wireframe validation, pixel-perfect implementation. Enforces WHAT/WHY first, design prototype before code. Triggers: "ux", "/ux", "design this page", "build a UI", "create a component", "redesign", "new screen", "landing page", "wireframe", "layout".
In this repository20 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ci-watchSkill
- commitSkill
- discovery-interview-prepSkill
- doc-generateSkill
- incident-responseSkill
- opportunity-solution-treeSkill
- polarisSkill
- prd-developmentSkill
- prioritization-advisorSkill
- reviewSkill
- roadmapSkill
- security-auditSkill
- session-wrapSkill
- spec-updateSkill
- tddSkill
- tech-debt-auditSkill
- test-runnerSkill
- user-storySkill
- workshop-facilitationSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
20 rules. About 0 are common, 19 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈19
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 17
When the user says "build a form/layout/page"
- Distinctive: line 18
When the user describes a feature that will have a visible UI
- Distinctive: line 19
Never skip this workflow for UI work, even if the request sounds purely technical
- Distinctive: line 42
Before asking anything, read:
- Distinctive: line 61
Ask ALL unknowns in a single message.
- Distinctive: line 87
Ask: "Is this understanding correct before I prototype?"
- Distinctive: line 89
If PO confirms → proceed to Phase 2.
- Distinctive: line 99
Use ASCII/text layout. Structure:
- Distinctive: line 124
Every element must have a stated PURPOSE (why it's there)
- Distinctive: line 126
Never show more than 3 levels of visual hierarchy
- Distinctive: line 127
Call out any gamification elements (XP, badges, progress bars) explicitly
- Distinctive: line 166
Before writing code, list components:
- Distinctive: line 180
Check each component against the active design rules:
- Distinctive: line 196
Follow the stack conventions (Server Components by default, shadcn primitives, project API client).
- Distinctive: line 227
Never skip Phase 1 even if the request seems obvious
- Distinctive: line 228
Never prototype before receiving PO answers to Phase 1 questions
- Distinctive: line 229
Never implement before PO approves the Phase 2 wireframe
- Distinctive: line 232
Always handle loading, empty, and error states — no exceptions
- Distinctive: line 233
Use project color tokens and typography scale — never raw values
- Distinctive: line 235
Always state your recommendation when asking questions to the PO
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
UX Workflow — Clarify → Prototype → Implement
When to use
- Any new page, screen, or significant UI component
- Redesigns of existing UI sections
- When the user says "build a form/layout/page"
- When the user describes a feature that will have a visible UI
- Never skip this workflow for UI work, even if the request sounds purely technical
When NOT to use
- Pure backend work (API endpoints, DB migrations)
- Fixing a bug in existing UI (no layout change)
- Config files, scripts, documentation
- Adding a single line or small tweak to an existing component
Integration points
- Before
/tdd: UI components need UX validation before tests are written - Before
/spec-update: UX output (wireframe + decisions) feeds the spec's Behavior sections - After
/review: design review catches UX regressions in changed components
Phase 1 — Clarification Métier (WHAT/WHY)
Goal: Understand WHAT and WHY before any visual thinking. Ask the PO, never assume.
Step 1a — Read existing context
Before asking anything, read:
- The functional spec for this domain:
docs/specs/<domain>.md— check User Stories and Behavior - Project CLAUDE.md — routing conventions, component patterns
- The app's CLAUDE.md if in a monorepo — styling conventions, shadcn list
- The design system rule (
.claude/rules/design-system.md) if it exists — project-specific tokens, palette, personas
Step 1b — Identify the unknowns
From the spec and request, list what is NOT yet defined:
- User goal: what does the user want to accomplish on this screen?
- Entry context: where does the user come from? what state are they in?
- Success state: what does the screen look like when the task is done?
- Empty/loading/error states: what does the user see while waiting or on failure?
- Constraints: authenticated only? mobile-first? premium paywall? PWA offline?
Step 1c — Ask targeted questions (WHAT/WHY only, never HOW)
Ask ALL unknowns in a single message. Maximum 5 questions. Format:
Phase 1/3 — Clarification
Before designing, I need to clarify:
1. [User goal question]
2. [Entry context question]
3. [Success state question]
4. [Edge case / constraint question]
5. [Priority / constraint question — optional]
(These are WHAT/WHY questions — I won't propose any layout until you answer.)STOP — Wait for PO answers. Do NOT propose layout, components, or code at this stage.
Step 1d — Validation gate
Declare answers understood. Summarize:
- User goal in one sentence
- Key states to handle (loading, empty, error, success)
- Constraints confirmed
Ask: "Is this understanding correct before I prototype?"
If PO confirms → proceed to Phase 2.
Phase 2 — Prototypage Textuel
Goal: Validate layout and UX decisions BEFORE writing any code.
Step 2a — Produce a textual wireframe
Use ASCII/text layout. Structure:
┌─────────────────────────────────────────────┐
│ [HEADER AREA — e.g., page title + subtitle] │
├─────────────────────────────────────────────┤
│ [PRIMARY CONTENT AREA] │
│ • [Element 1: type + purpose] │
│ • [Element 2: type + purpose] │
├─────────────────────────────────────────────┤
│ [SECONDARY CONTENT / SIDEBAR] │
├─────────────────────────────────────────────┤
│ [CTA ZONE — primary action + secondary] │
└─────────────────────────────────────────────┘
States:
- Loading: [describe skeleton/spinner behavior]
- Empty: [describe empty state copy + action]
- Error: [describe error message + recovery]
Mobile (<768px): [describe layout changes]Rules for the wireframe:
- Every element must have a stated PURPOSE (why it's there)
- One and only one primary CTA per screen
- Never show more than 3 levels of visual hierarchy
- Call out any gamification elements (XP, badges, progress bars) explicitly
- Specify responsive behavior for mobile
Step 2b — Justify UX decisions
For each significant design choice, state the rationale:
| Decision | Rationale | Alternative rejected |
| [choice] | [why this serves the user goal] | [what was considered] |
Step 2c — Ask for PO validation
Phase 2/3 — Wireframe
[textual wireframe here]
UX Decisions:
[table above]
Questions before I code:
1. Is the hierarchy correct?
2. Are the states handled correctly?
3. Any copy/label feedback?
→ Approve to proceed to implementation, or request changes.STOP — Do NOT write any React/Tailwind code until PO explicitly approves the wireframe.
Phase 3 — Implémentation Pixel-Perfect
Goal: Translate the approved wireframe into production-ready code.
Step 3a — Component decomposition
Before writing code, list components:
<PageName>
├── <HeaderSection> — server or client?
├── <ContentArea>
│ ├── <ItemCard> — reusable?
│ └── <EmptyState> — when list is empty
├── <LoadingState> — Suspense boundary
└── <CTASection> — primary actionStep 3b — Design checklist
Check each component against the active design rules:
- Max-width constraint respected (max-w-2xl text, max-w-5xl layout)
- Typography scale used (h1/h2/h3/body/data) — no skipped levels
- Color tokens used (never raw hex values)
- Loading state present for async data (skeleton, not spinner)
- Empty state present for zero-data case (with CTA)
- Error state present for failure case (with recovery action)
- Single primary CTA identified
- Mobile-first (tested mentally at 375px)
- Touch targets >= 44x44px
- Dark mode tokens used (no raw white/black)
- WCAG AA contrast (4.5:1 body, 3:1 large)
Step 3c — Implement
Follow the stack conventions (Server Components by default, shadcn primitives, project API client).
Step 3d — Self-review
After implementation:
- Compare output against the approved wireframe — does the code match?
- Flag any deviations from the wireframe with explanation
- Suggest
/tddfor logic-bearing components - Suggest
/reviewif 3+ files changed
Step 3e — Report to PO
Phase 3/3 — Implementation complete
Files created/modified:
- [list]
Design checklist: [X passed / Y total]
Deviations from wireframe (if any):
- [or "None — implementation matches wireframe exactly"]
Next: /commit or /uat for full validationRules
- Never skip Phase 1 even if the request seems obvious
- Never prototype before receiving PO answers to Phase 1 questions
- Never implement before PO approves the Phase 2 wireframe
- Phase 1 questions are WHAT/WHY only — never propose HOW
- One primary CTA per screen — hard rule
- Always handle loading, empty, and error states — no exceptions
- Use project color tokens and typography scale — never raw values
- Component decomposition must precede implementation
- Always state your recommendation when asking questions to the PO
Anti-patterns
| Anti-pattern | Why it's wrong |
| Starting to code from a vague request | Solves the wrong problem — Phase 1 exists for this |
| Asking "how should I implement this?" in Phase 1 | Phase 1 is about WHAT/WHY, HOW is the dev's job |
| Skipping empty/error states | Breaks real user flows — users WILL hit these |
| Two primary CTAs on one screen | Splits user attention, decision paralysis |
| Using raw hex colors or pixel values | Breaks design system consistency and dark mode |
| Writing components before decomposing them | Leads to monolithic, untestable components |
| Skipping mobile consideration | Mobile-first means design for mobile FIRST |
References
/spec-update— update functional spec with UX decisions after Phase 3/tdd— write component tests after wireframe is approved/review— design review before committing UI changes/uat— full acceptance testing including UX validation scenarios
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

