5.3 KiB
5.3 KiB
Skill: Design Thinking - Prototype
Description
Build rapid, low-fidelity prototypes to test solutions quickly and cheaply before full implementation.
Input
- solution_idea: The idea to prototype from ideate phase (required)
- fidelity_level: low|medium|high (optional, default: low)
- components_available: Existing UI components/patterns (optional)
- constraints: Time/tech constraints (optional)
Prototyping Approaches
1. Rapid MVP Patterns
Lo-Fi (1-2 hours):
- Paper sketches
- Wireframes in Figma/Excalidraw
- Clickable mockups
- No code, just flows
Med-Fi (4-8 hours):
- Static HTML/CSS
- Component prototypes
- Fake data, real UI
- Key interactions only
Hi-Fi (1-2 days):
- Working code prototype
- Real component library
- Actual data flow
- Limited features, polished
2. Lo-Fi to Hi-Fi Progression
Stage 1: Sketch (30 mins)
Purpose: Explore ideas fast
Tool: Paper, whiteboard, Excalidraw
Output: 3-5 rough layouts
Decision: Pick one to refine
Stage 2: Wireframe (1-2 hours)
Purpose: Define structure & flow
Tool: Figma, Balsamiq, HTML
Output: Clickable wireframe
Decision: Validate flow with users
Stage 3: Visual Design (2-4 hours)
Purpose: Define look & feel
Tool: Figma with design system
Output: High-fidelity mockups
Decision: Validate aesthetics
Stage 4: Code Prototype (4-8 hours)
Purpose: Test technical feasibility
Tool: React/Next.js components
Output: Working prototype
Decision: Ready for user testing
3. Component-First Approach
Build from existing:
// Start with proven components
import { Button, Input, Card } from '@/components/ui'
// Compose into new patterns
function CampaignQuickCreate() {
return (
<Card>
<Input placeholder="Campaign name" />
<Button>Create in 5 seconds</Button>
</Card>
)
}
Component Inventory:
- List available UI components
- Identify reusable patterns
- Note what needs building
- Prioritize existing over new
4. "Good Enough" Quality Gates
Lo-Fi Checklist:
- User can see the flow
- Key interactions are clear
- No polish needed
- Can get feedback in <1 hour
Med-Fi Checklist:
- Looks like real product
- Core interactions work
- Uses design system
- Good enough for user testing
Hi-Fi Checklist:
- Production-quality code
- Accessible & responsive
- Error states handled
- Ready to ship (limited scope)
Prototype Templates
Quick Start Template
// Minimal prototype structure
interface PrototypeProps {
mode: 'demo' | 'interactive'
}
export default function Prototype({ mode }: PrototypeProps) {
// 1. State (minimal)
const [step, setStep] = useState(1)
// 2. Key interactions only
const handleNext = () => setStep(s => s + 1)
// 3. Visual feedback
return (
<div className="max-w-2xl mx-auto p-8">
<ProgressBar step={step} total={3} />
{step === 1 && <Step1 onNext={handleNext} />}
{step === 2 && <Step2 onNext={handleNext} />}
{step === 3 && <Success />}
</div>
)
}
Paper Prototype Template
┌─────────────────────────┐
│ Campaign Quick Create │
├─────────────────────────┤
│ │
│ [Campaign Name____] │
│ │
│ [Template ▼] │
│ - Black Friday │
│ - Product Launch │
│ - Newsletter │
│ │
│ [Create Campaign] │
│ │
└─────────────────────────┘
Interactions:
1. Type name → enables template
2. Select template → preview
3. Click create → done (5 sec)
Output Format
{
"status": "success",
"prototype": {
"type": "medium_fidelity",
"time_spent": "6 hours",
"tools_used": ["Figma", "React", "Tailwind"],
"url": "https://prototype-link.com",
"components_reused": ["Button", "Input", "Card", "Select"],
"components_created": ["QuickCreateFlow"]
},
"key_features": [
"1-click template selection",
"Auto-filled campaign defaults",
"5-second creation flow",
"Inline preview"
],
"limitations": [
"Only 3 template types",
"No advanced customization",
"Fake data backend"
],
"testing_scenarios": [
{
"scenario": "Create campaign from template",
"steps": ["Select template", "Enter name", "Click create"],
"expected_time": "5 seconds",
"success_criteria": "Campaign created without errors"
}
],
"next_step": "Test with users using /dt test"
}
Quality Gates
- Built in <8 hours (for med-fi)
- Core flow works end-to-end
- Uses existing components where possible
- Good enough for user feedback
- Testing scenarios defined
- Known limitations documented
Token Budget
- Max input: 1000 tokens
- Max output: 1500 tokens
Model
- Recommended: sonnet (creative problem-solving)
Philosophy
"Perfect is the enemy of good. Ship the prototype." Fast feedback beats perfect implementation.
Keep it simple:
- Build minimum to test hypothesis
- Reuse > rebuild
- Speed > polish
- Learn > perfect
- Fake it till you make it (use mock data)