deal-hunt skill

---
name: deal-hunt
description: "Find the best current deals/coupons for a specific product. Searches web for deals and returns raw results for analysis."
---

# Deal Hunt

Search for deals on any product. Returns raw Tavily search results - Claude analyzes them to find the best prices.

## Prerequisites

**Tavily API Key Required** - Get your key at https://tavily.com

Add to `~/.claude/settings.json`:
```json
{
  "env": {
    "TAVILY_API_KEY": "tvly-your-api-key-here"
  }
}
```

## Usage

```bash
# Search entire web (default - no domain filter)
python scripts/deal_hunt.py "Dyson V15"

# Multi-query search (max 3, runs in parallel, deduplicates results)
python scripts/deal_hunt.py "AirPods Pro" --queries "AirPods Pro deal,AirPods Pro coupon,AirPods Pro discount"

# Limit to specific sites
python scripts/deal_hunt.py "MacBook Air" --domains amazon.com,walmart.com,bestbuy.com

# Custom single query
python scripts/deal_hunt.py "Nintendo Switch" --query "Nintendo Switch OLED bundle deal"

# Fresh deals only
python scripts/deal_hunt.py "PS5" --time-range day
```

## CLI Parameters

| Option | Short | Default | Description |
|--------|-------|---------|-------------|
| `product` | - | Required | Product name |
| `--query` | `-q` | `{product} deal price` | Single custom search query |
| `--queries` | | None | Comma-separated queries (max 3), runs in parallel with dedup |
| `--domains` | `-d` | None (search all) | Optionally limit to specific domains |
| `--max-results` | `-n` | 10 | Number of results per query |
| `--time-range` | `-t` | week | day, week, month, year, none |
| `--search-depth` | `-s` | advanced | basic, advanced, fast, ultrafast |

## Output

Returns JSON with results:

```json
{
  "meta": {
    "product": "AirPods Pro",
    "queries": ["AirPods Pro deal", "AirPods Pro coupon"],
    "domains": null,
    "time_range": "week",
    "search_time": "2026-01-13T...",
    "total_results": 15
  },
  "results": [
    {
      "title": "...",
      "url": "https://...",
      "content": "...",
      "score": 0.95
    }
  ]
}
```

When using `--queries`, results are deduplicated by URL (highest score kept, content merged).

## Output Schema for Analysis

After running the search, Claude should analyze results and structure findings as:

```json
{
  "product": "Sony WH-1000XM5",
  "best_deal": {
    "price": 279.99,
    "original_price": 399.99,
    "discount": "30% off",
    "retailer": "Amazon",
    "url": "https://amazon.com/...",
    "condition": "new",
    "in_stock": true
  },
  "all_deals": [
    {
      "price": 279.99,
      "retailer": "Amazon",
      "url": "https://...",
      "notes": "Prime shipping"
    },
    {
      "price": 169.99,
      "retailer": "eBay via Slickdeals",
      "url": "https://...",
      "notes": "Refurbished"
    }
  ],
  "coupons": [
    {
      "code": "AUDIO10",
      "discount": "10% off",
      "retailer": "Best Buy",
      "expires": "2026-01-31"
    }
  ],
  "summary": "Best new price is $279.99 at Amazon (30% off). Refurbished available for $169.99."
}
```

Claude extracts prices from content, compares deals, and presents the best options with purchase links.