Product Discovery

You have access to a product search script that queries a catalog of millions of products across thousands of retailers. Use it whenever you need real product data to answer a user's question.

Search Script

Location: product-discovery/scripts/search.sh (relative to the skill root)

Run it via the shell. The script requires curl and jq.

Usage

search.sh [OPTIONS] "query text"

Options

Flag	Description	Example
-n NUM	Number of results (default: 5, max: 30)	-n 10
-p MAX_PRICE	Maximum price in dollars	-p 100
--min-price MIN	Minimum price in dollars	--min-price 50
-g GENDER	Gender filter (male/female/unisex)	-g male
-c CONDITION	Product condition (new/refurbished/used)	-c new
-a AGE	Comma-separated age groups (newborn/infant/toddler/kids/adult)	-a "kids,toddler"
--availability STATUS	Comma-separated availability statuses (InStock/OutOfStock/PreOrder/BackOrder/LimitedAvailability/SoldOut/Discontinued)	--availability "InStock"
-i IMAGE_URL	Search by image (visual similarity)	-i "https://example.com/photo.jpg"
-b BRAND_IDS	Comma-separated brand IDs to include	-b "brand_abc,brand_def"
-w WEBSITE_IDS	Comma-separated website IDs to include	-w "website_abc"
--categories IDS	Comma-separated category IDs to include	--categories "cat_abc"
--exclude-brands IDS	Comma-separated brand IDs to exclude	--exclude-brands "brand_xyz"
--exclude-websites IDS	Comma-separated website IDs to exclude	--exclude-websites "website_xyz"
--exclude-categories IDS	Comma-separated category IDs to exclude	--exclude-categories "cat_xyz"
--keyword-only	Use exact keyword matching instead of semantic search	--keyword-only
--next TOKEN	Pagination token from a previous search	--next "tok_abc..."

The query argument is optional when using -i for image-only search. Text and image can be combined.

Examples

# Basic text search
search.sh "wireless noise cancelling headphones"

# Price-filtered search
search.sh -p 100 -n 10 "running shoes"

# Price range search
search.sh --min-price 50 -p 200 "winter boots"

# Gendered search
search.sh -g female -p 200 "winter jacket"

# Condition filter
search.sh -c used "macbook pro"

# Kids products
search.sh -a kids -p 50 "sneakers"

# Only in-stock products
search.sh --availability "InStock" "yoga mat"

# Image-based visual similarity search
search.sh -i "https://example.com/dress.jpg"

# Combined text + image search
search.sh -i "https://example.com/jacket.jpg" "similar but in blue"

# Keyword-only (exact match, no semantic search)
search.sh --keyword-only "Nike Air Max 90"

# Paginate for more results
search.sh --next "tok_abc123..." "running shoes"

Output Format

The script outputs structured text, not raw JSON. Each product includes its ID, brands, and all merchant offers with prices and buy links:

Found 5 products (next_page: tok_abc123)

1. Nike Air Zoom Pegasus 41
   ID: prod_abc123
   Brands: Nike
   Offers:
     - nordstrom.com: $89.99 (InStock) https://buy.trychannel3.com/...
     - nike.com: $94.99 (InStock) https://buy.trychannel3.com/...

2. Adidas Ultraboost Light
   ID: prod_def456
   Brands: Adidas
   Offers:
     - adidas.com: $97.00 (InStock) https://buy.trychannel3.com/...

If no results are found, the output is: No products found.

If the API key is missing or invalid, the script prints instructions on how to get one.

Workflow Patterns

Find products

User asks "find me running shoes under $100":

1. Run: search.sh -p 100 "running shoes"
2. Present the results as a clean numbered list with product name, price, merchant, and buy link.

Compare products

User asks "compare AirPods Pro vs Sony WF-1000XM5":

1. Run two searches: search.sh -n 3 "AirPods Pro" and search.sh -n 3 "Sony WF-1000XM5"
2. Build a markdown comparison table with columns for product, price, merchants, and availability.

Best option under a budget

User asks "what's the best laptop under $800":

1. Run: search.sh -p 800 -n 10 "laptop"
2. Review the results and recommend the top picks, explaining why based on the product details.

Image-based search

User shares an image URL and says "find me something like this":

1. Run: search.sh -i "IMAGE_URL"
2. Present visually similar products with prices and links.

Get more results

If the user wants more results after an initial search:

1. Copy the next_page token from the previous output.
2. Run: search.sh --next "TOKEN" "original query"
3. Present the additional results.

How to Present Results to the User

• Synthesize the script output into a clean response. Do NOT paste the raw script output.
• Present products as a numbered list or markdown table — whichever fits the question better.
• Always include: product name, price, merchant name, and the buy link URL.
• For comparisons, use a markdown table with columns like Product, Price, Merchant, and Link.
• If multiple merchants sell the same product at different prices, highlight the cheapest option.
• Keep it concise — the user wants recommendations, not a data dump.