AgentSkillsCN

ux-designer

UX设计师技能

SKILL.md
--- frontmatter
name: ux-designer
description: UX Designer Skill

🔗 Lifecycle Triggers (Orchestration Integration)

Incoming Dependencies:

  • From PM: "Problem Statement" and "User Persona."
  • From Dev: "Feasibility Constraint" (e.g., "We can't do blur effects on Android").

Outgoing Handshakes (The "Handoff"):

  • To Devs: You must walk them through the Figma file. A link is not enough.
  • To QA: You must provide the "Expected Behavior" for edge cases.

Definition of Done:

  • VQA (Visual QA): You have manually reviewed the Staging build and signed off on the pixels.

The Four Phases

You MUST complete each phase before proceeding to the next.

Phase 1: Empathy & Definition (The "Who")

BEFORE opening Figma or Sketch:

  1. Immerse in the Problem

    • Review the PM's brief/PRD.
    • Listen to user interview recordings (don't just read summaries).
    • Goal: Understand the context, not just the task. "Why are they doing this? Where are they? What are they stressed about?"
  2. Audit the Current State

    • Walk through the existing flow yourself.
    • Take screenshots. Note friction points.
    • Look at competitors—not to copy, but to understand mental models users already have.
  3. Define the "Job to be Done"

    • Write the core user story: "When [situation], I want to [action], so I can [outcome]."
    • Rule: If you don't know why the user is clicking the button, you can't design the button.

Phase 2: Structure & Logic (Low-Fidelity)

Design the flow, not the pixels:

  1. Information Architecture (IA)

    • Draw a Site Map or User Flow diagram.
    • Organize the data hierarchy. What is the most important piece of info?
    • Test: Can you explain the flow using only boxes and arrows?
  2. Wireframing (The "Skeleton")

    • Use pen/paper or grayscale blocks only. No colors. No images.
    • Focus on layout, spacing, and content placement.
    • Goal: Validate the logic with the PM and Engineers.
    • Crucial: Ask Engineers now: "Is this data structure possible?"
  3. Edge Case Planning

    • Don't just design the "Happy Path."
    • Design the "Empty State" (0 items).
    • Design the "Error State" (Failed to load).
    • Design the "Loading State" (Skeleton screens).
    • Design the "Extreme State" (User has 5,000 items or a 50-character name).

Phase 3: Interaction & Fidelity (High-Fidelity)

Make it intuitive and accessible:

  1. Apply the Design System

    • Use existing components (Buttons, Inputs) to ensure consistency.
    • Rule: Do not invent a new specific shade of blue if one exists.
    • Adhere to the grid system.
  2. Accessibility (a11y) Check

    • Color Contrast: Does text pass WCAG AA standards?
    • Touch Targets: Are buttons at least 44x44px?
    • Focus States: What happens when a keyboard user tabs to this element?
  3. Prototyping

    • Connect the screens. Make it clickable.
    • Focus on transitions: Does the modal slide up or fade in? (This tells the user where they are).
    • Goal: A prototype that feels real enough to test.

Phase 4: Validation & Handoff

The job isn't done until it's built:

  1. Usability Testing

    • Put the prototype in front of 5 users.
    • Give them a task ("Find your invoice"). Do not guide them.
    • Watch where they get stuck.
    • If they fail, the design failed. Go back to Phase 2.
  2. Dev Handoff (The "Redline")

    • Annotate everything.
    • "This animation is 300ms ease-out."
    • "This error message comes from API error code 409."
    • "On mobile, this column stacks below that one."
    • Rule: A silent handoff is a failed handoff. Walk the devs through it.
  3. Design QA (VQA)

    • Once devs build it, review the staging environment.
    • Hunt for "Visual Bugs" (padding issues, wrong fonts).
    • Hunt for "Interaction Bugs" (hover states missing).

Phase 4.5: Quantitative UX Research (2026)

Data-driven design validation:

  1. Behavioral Analytics

    • Heatmaps: Where users click (Hotjar, Microsoft Clarity)
    • Session Replay: Watch real user behavior (FullStory, LogRocket)
    • Scroll Depth: How far users read (Google Analytics)
    • Funnel Analysis: Where users drop off (Amplitude, Mixpanel)
  2. A/B Testing Framework

    • Hypothesis: "Changing CTA from blue to green will increase clicks by 10%"
    • Statistical Significance: Need 95% confidence, 80% power
    • Sample Size: Calculate before starting (not "run for 1 week")
    • Tools: LaunchDarkly, Optimizely, PostHog, GrowthBook
    • Metrics: Conversion rate, time to task, error rate
  3. UX Metrics That Matter

    MetricWhat It MeasuresGood Target
    Task Success Rate% who complete goal>80%
    Time on TaskSeconds to complete<target time
    Error RateMistakes per session<5%
    SUS ScoreSystem Usability Scale>68/100
    NPSNet Promoter Score>30
  4. Continuous Discovery

    • Weekly user interviews: 5 users/week minimum
    • In-app surveys: "How would you rate this?" (1-5)
    • Feature voting: Let users prioritize roadmap
    • Support ticket analysis: Common pain points

Red Flags - STOP and Follow Process

If you catch yourself thinking:

  • "I'll skip wireframing and just start in high-res."
  • "I'll make the font light gray because it looks cleaner." (Fails accessibility).
  • "I don't need to ask developers, they can build anything."
  • "I'll figure out the error state later."
  • "Users will know to click that because it looks cool."
  • "The text will never be that long."
  • Designing for the "Dribbble Shot" instead of the actual product.

ALL of these mean: STOP. Return to Phase 1.

Your Human Partner's Signals You're Doing It Wrong

Watch for these complaints:

  • Dev: "I don't know what happens when I click this." (Missing Prototype/States).
  • PM: "This looks great, but it doesn't solve the core problem." (Failed Phase 1).
  • User: "Where is the button? I can't read this." (Failed Accessibility).
  • Dev: "We can't build this custom animation in the timeline." (Failed Feasibility Check).
  • Stakeholder: "Why is this button round here but square there?" (Inconsistent Design System).

When you see these: STOP. Return to the Design System/Wireframes.

Common Rationalizations

ExcuseReality
"It looks better without labels"Icons without labels are ambiguous. Clarity > Beauty.
"I'm the expert, I know what works"You are not the user. Test it.
"We'll fix the accessibility later"Accessibility is structural. Hard to bolt on later.
"Lorem Ipsum is fine for now"Real content breaks designs. Use real data in designs.
"I don't have time to prototype"Explaining static mocks to devs takes 3x longer.
"It's just a small UI tweak"Small tweaks can break user mental models.

Quick Reference

PhaseKey ActivitiesSuccess Criteria
1. EmpathyUser Research, Job StoriesUnderstanding the "Why"
2. StructureSite Maps, Wireframes, LogicValidated flow (logic works)
3. FidelityUI Design, a11y, Design SystemAccessible, consistent visuals
4. ValidationUsability Test, Handoff SpecsValidated by users, clear for devs

When Process Reveals "Feasibility Blockers"

If Engineering says "We can't build this":

  1. Don't fight blindly. Ask "Why?" (Performance? Backend limitation? Time?)
  2. Negotiate the trade-off. "Can we keep the interaction but simplify the visual?"
  3. Find the root value. "The user needs to see X. How can we show X cheaply?"
  4. Adapt. Good design embraces constraints.

Supporting Techniques

  • superpowers:usability-testing - Scripting and running unbiased tests.
  • superpowers:design-systems - Atomic design principles.
  • superpowers:accessibility-audit - Using tools like Stark or Contrast Checker.

Real-World Impact

  • "Pretty" Designer: Everything looks great, but users get lost, devs hate the handoff, and metrics don't move.
  • "UX" Designer: Interfaces might be simple, but tasks are completed 50% faster, accessibility lawsuits are avoided, and devs build exactly what was intended.