Strategic deployment to minimize build credits (~$25-40 target)
1. Core Game Logic (Must Be Perfect)
GuessWhat.jsx
Complete game with all states
- βMenu screen with instructions
- βGame loop: select idiom β show hints β guess β score
- βHint system (3 hints: cryptic, example, first letter)
- βScoring: base 100 - penalties for hints/time
- βVictory/Game Over screens
- βStreak tracking
- βLocal state management working
2. Famous.AI Integration Components
Copy from Scrabble (Already Working)
- βFamousAIAuthGate.jsx - JWT auth + guest mode
- βFamousAIPearlTracker.jsx - Pearl balance + awards
- βShareScoreModal.jsx - Social sharing with card
- βProgressMigrationModal.jsx - Guest β User migration
- βFamousAIGameTemplate.jsx - Boilerplate wrapper
3. Visual Assets (Prepare Before Upload)
Game Logo
Size: 512x512
π Guess What! logo with question mark
Background
Size: 1920x1080
Subtle idiom-themed pattern (lightbulbs, speech bubbles)
Hint Icons
Size: 64x64
π Magnifying glass, π‘ Lightbulb, π °οΈ Letter
Victory Badge
Size: 256x256
π Gold trophy with confetti
Character Avatar
Size: 128x128
Snappy wearing detective hat
Sound Effects
Size: <100KB each
hint.mp3, correct.mp3, wrong.mp3
4. Database Schema (Test on Base44 First)
-- Test this entity on Base44, ensure it works
CREATE TABLE guess_what_scores (
id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
created_at TIMESTAMPTZ DEFAULT NOW(),
user_email TEXT NOT NULL,
user_name TEXT NOT NULL,
final_score INTEGER NOT NULL,
games_played INTEGER NOT NULL,
hints_used INTEGER NOT NULL,
streak INTEGER NOT NULL,
time_elapsed INTEGER NOT NULL,
idioms_guessed TEXT[] DEFAULT '{}',
difficulty_level TEXT DEFAULT 'medium'
);
CREATE INDEX idx_gw_user ON guess_what_scores(user_email);
CREATE INDEX idx_gw_score ON guess_what_scores(final_score DESC);
CREATE INDEX idx_gw_created ON guess_what_scores(created_at DESC);5. Edge Functions (Write & Test on Base44)
guess-what-complete function
Handles game completion, pearl awards
- βValidates JWT token
- βSaves score to database
- βAwards 10 pearls + streak bonus
- βSends email receipt via SendGrid
- βReturns updated pearl balance
guess-what-leaderboard function
Fetches top 50 scores
- βQueries guess_what_scores table
- βOrders by final_score DESC
- βReturns JSON with rankings
Credit-Saving Strategy
Be EXTREMELY specific in your prompts. General requests like "make it look better" burn credits with back-and-forth. Each message exchange = credits. One perfect prompt = one credit charge.
Prompt #1: Visual Upgrade (Estimated 8-12 credits)
Upload the complete GuessWhat.jsx game. The logic is perfect and working. I need ONLY visual polish - do NOT change any game logic, state management, or data flow. Visual upgrades needed: 1. **Hint Reveal Animation** - When user clicks hint button, animate hint text sliding in from right - Add particle effect (sparkles) around newly revealed hint - Use framer-motion with spring physics 2. **Correct Answer Celebration** - Trigger confetti burst using canvas-confetti - Animate score counter incrementing with easing - Show "+X Pearls" floating up and fading out - Play success sound effect 3. **Wrong Answer Feedback** - Shake the input field on wrong guess (vibration animation) - Flash red border briefly - Show encouraging message "Try again! Hint available" 4. **Streak Indicator Enhancement** - Add flame emoji π₯ that grows bigger with higher streak - Pulsing glow effect for streaks >5 - Streak break shows "broken chain" animation 5. **Loading States** - Skeleton loaders for idiom cards while fetching - Shimmer effect on pearl balance during sync - Smooth transitions between game states Do NOT modify: game logic, scoring, state management, API calls, data structures. Keep all existing functionality intact.
Prompt #2: Responsive Polish (Estimated 5-8 credits)
Optimize GuessWhat.jsx for mobile devices (320px - 428px width). Changes needed: 1. **Touch-Friendly Targets** - Increase hint button size to min 48x48px - Add 12px spacing between interactive elements - Make guess input field larger on mobile (min-h-14) 2. **Typography Scaling** - Scale idiom text: clamp(1.25rem, 4vw, 2rem) - Hint text: clamp(0.875rem, 3vw, 1rem) - Ensure readability on small screens 3. **Layout Adjustments** - Stack hint buttons vertically on mobile (<640px) - Reduce padding: p-4 on mobile, p-6 on desktop - Collapse header logo to icon-only on <400px 4. **Performance** - Lazy load background images - Debounce input validation (300ms) - Optimize animations for 60fps on mobile Test on: iPhone SE (375px), iPhone 14 Pro (393px), Samsung Galaxy S23 (412px).
Prompt #3: Leaderboard Enhancement (Estimated 4-6 credits)
Enhance the leaderboard display in GuessWhat.jsx. Current: Basic table with names and scores. Target: Engaging, competitive visual. Add these elements: 1. **Rank Badges** - Top 3: Trophy icons (π₯π₯π₯) with glow effect - 4-10: Medal icons - User's row: Highlight with blue border + "You" badge 2. **Stat Cards** - Show: Score, Streak, Hints Used, Time - Mini progress bars for each stat - Color-coded: green (excellent), yellow (good), red (needs improvement) 3. **Animations** - Fade in each row sequentially (stagger 50ms) - Hover effect: lift + shadow on desktop - Pull to refresh on mobile 4. **Filters** - Tabs: All Time | This Week | Today - Difficulty filter: Easy, Medium, Hard - Search by username Keep: Existing leaderboard API call and data structure.
Prompt #4: Sound & Accessibility (Estimated 3-5 credits)
Add audio feedback and accessibility features to GuessWhat.jsx. 1. **Sound Effects** (use WebAudioSounds class if available) - Hint reveal: soft "pop" sound - Correct guess: celebration chime - Wrong guess: gentle "bonk" - Streak milestone: triumph fanfare - Volume controls in settings menu 2. **Accessibility** - Add aria-labels to all interactive elements - Keyboard navigation: Tab, Enter, Escape - Screen reader announcements for score changes - High contrast mode option - Reduced motion mode (disable animations) 3. **Settings Panel** - Toggle: Sound ON/OFF - Toggle: Animations ON/OFF - Toggle: High Contrast Mode - Persist preferences in localStorage Follow WCAG 2.1 AA standards.
Upload Sequence
- Upload all Base44-built files to Famous.AI project
- Upload visual assets to File Buckets (scrabble-gang-quiz-thats15points)
- Deploy Edge Functions (guess-what-complete, guess-what-leaderboard)
- Run database migrations (create tables)
- Set environment variables (JWT_SECRET, GATEWAY_API_KEY)
- Test in Famous.AI preview
- Submit polish prompts ONE AT A TIME
- Test after each polish iteration
Final Polish Prompt (Estimated 3-5 credits)
Final integration check for GuessWhat.jsx: 1. Verify all Famous.AI components working: - FamousAIAuthGate shows login/guest options - FamousAIPearlTracker displays balance - ShareScoreModal generates proper card - ProgressMigrationModal triggers for guests 2. Test complete flow: - Guest plays β sees temporary pearls - Guest converts β pearls migrate - User plays β pearls save to database - Email receipt sends on game complete 3. Performance check: - Initial load <2 seconds - Hint reveal animation smooth (60fps) - Leaderboard fetch <500ms - No console errors or warnings If any issues found, report them specifically and I'll fix.
Visual Upgrade Prompt
8-12 credits
Most expensive - complex animations
Responsive Polish
5-8 credits
Mobile optimization
Leaderboard Enhancement
4-6 credits
UI improvements
Sound & Accessibility
3-5 credits
A11y + audio
Final Integration
3-5 credits
Testing & fixes
Buffer for Revisions
5-10 credits
Expected iterations
Compared to $232 for Scrabble (built from scratch on Famous.AI)
β 82% cost reduction by pre-building on Base44
Be hyper-specific in prompts. Don't say "make it look modern" - list exact animations, colors, sizes.
Test after each prompt. Don't chain 5 prompts assuming they'll all work - you'll waste credits on cascading errors.
Include screenshots/examples. "Make leaderboard like this [screenshot]" saves credits vs. describing in words.
Lock working code. Explicitly state "DO NOT modify game logic" in every visual prompt.
Use preview mode liberally. Famous.AI preview is free - test thoroughly before asking for changes.
Keep a changelog. Track what you've asked for to avoid re-requesting same polish.
Batch similar requests. Group all animation requests in one prompt rather than separate messages.
Build on Base44
Complete GuessWhat.jsx with all logic, test thoroughly, prepare assets
Upload to Famous.AI
Deploy files, functions, database schema, test basic functionality
Visual Polish Round 1
Submit visual upgrade + responsive prompts, test results
Visual Polish Round 2
Submit leaderboard + accessibility prompts, test results
Integration & Testing
Final polish, test complete user flow, fix any issues
Launch
Keep in SLEEP MODE initially, wake when validated