User scrolls Twitter
      ↓
┌──────────────────────────────────────┐
│   Musashi Browser Extension (Client)│
└──────────────────────────────────────┘
              ↓
┌──────────────────────────────────────┐
│     Musashi API (Vercel Backend)    │
│  POST /api/analyze-text              │
│  - Keyword Matcher                   │
│  - Synonym Expansion                 │
│  - 100+ Markets Database             │
└──────────────────────────────────────┘
              ↓
┌──────────────────────────────────────┐
│     JSON Response                    │
│  {                                   │
│    markets: [...],                   │
│    confidence: 0.87,                 │
│    matchedKeywords: [...]            │
│  }                                   │
└──────────────────────────────────────┘
       ↓                    ↓
   Extension          External AI Agents
   (Human UI)         (Programmatic)

Musashi/
├── api/                           # NEW: Backend API
│   ├── analyze-text.ts           # Main endpoint
│   └── lib/
│       ├── analysis/
│       │   └── keyword-matcher.ts # Matching logic
│       ├── data/
│       │   └── mock-markets.ts    # 100+ markets
│       └── types/
│           └── market.ts          # TypeScript types
│
├── src/                           # Browser extension
│   ├── api/                      # NEW: API client
│   │   └── musashi-api-client.ts # Fetch wrapper
│   ├── content/
│   │   └── content-script.tsx    # UPDATED: Now calls API
│   ├── sidebar/
│   ├── analysis/                 # Fallback local matcher
│   └── data/
│
├── public/
│   └── api-docs.html             # NEW: HTML documentation
│
├── API_DOCUMENTATION.md           # NEW: Full API docs
├── API_DEPLOYMENT_GUIDE.md        # NEW: Deployment instructions
├── test-api.js                    # NEW: API test script
└── vercel.json                    # NEW: Vercel config

export class MusashiApiClient {
  async analyzeText(text: string, options?: {
    minConfidence?: number;
    maxResults?: number;
  }): Promise<MarketMatch[]>
}

// Singleton instance
export const musashiApi = new MusashiApiClient(
  process.env.MUSASHI_API_URL || 'http://localhost:3000'
);

export default async function handler(
  req: VercelRequest,
  res: VercelResponse
): Promise<void> {
  // CORS headers for extension access
  // Validate request
  // Run keyword matcher
  // Return structured JSON
}

async function initializeMusashi() {
  // Test API connection
  try {
    await musashiApi.analyzeText('test');
    console.log('✓ API connected');
  } catch {
    console.warn('Falling back to local matcher');
    useLocalMatcher = true;
  }

  // Process tweets
  for (const tweet of tweets) {
    const matches = useLocalMatcher
      ? localMatcher.match(tweet.text)
      : await musashiApi.analyzeText(tweet.text);

    if (matches.length > 0) {
      injectCard(tweet.element, matches);
    }
  }
}

import requests

# Agent monitors user's feed
def analyze_user_interest(text):
    response = requests.post(
        'https://musashi-api.vercel.app/api/analyze-text',
        json={'text': text, 'maxResults': 5}
    )

    markets = response.json()['data']['markets']

    # Agent decides what to trade
    for market in markets:
        if market['confidence'] > 0.7:
            trade_on_polymarket(market['market'])

# Install dependencies
npm install

# Run API locally
vercel dev
# API available at http://localhost:3000

# Test API
node test-api.js

# Build extension
npm run build

# Load extension in Chrome
# chrome://extensions → Load unpacked → select dist/

# Deploy to Vercel
vercel --prod

# You'll get a URL like:
# https://musashi-api-abc123.vercel.app

# Update extension API client
# Edit src/api/musashi-api-client.ts:
export const musashiApi = new MusashiApiClient(
  'https://musashi-api-abc123.vercel.app'
);

# Rebuild extension
npm run build

# Reload extension in Chrome

# Start local API
vercel dev

# Run test script
node test-api.js

# Expected output:
# ✅ Success (87ms)
# Found 2 market(s):
# 1. Will the Fed cut interest rates in March 2026?
#    Confidence: 87.0%
#    YES: 72% | NO: 28%

# Agent watches user's timeline
def on_new_tweet(tweet):
    markets = musashi_api.analyze(tweet)
    if markets[0].confidence > 0.8:
        execute_trade(markets[0])

# Agent scrapes headlines
for headline in fetch_news():
    markets = musashi_api.analyze(headline)
    check_arbitrage(markets)

# User asks: "Markets about AI?"
markets = musashi_api.analyze("AI regulation artificial intelligence")
display_markets_to_user(markets)

# Cross-platform price comparison
for market in musashi_api.analyze(text):
    poly_price = get_polymarket_price(market.id)
    kalshi_price = get_kalshi_price(market.id)
    if abs(poly_price - kalshi_price) > 0.05:
        log_arbitrage(market)