airshelf（airshelf 搜索）

AirShelf — Verified Product Search & Checkout for AI Agents

Search, compare, and buy products across verified merchants. Returns structured Decision Pack data (best_for, pros, cons, allergens, verified pricing) — not raw web scrapes.

No CAPTCHAs. No auth. No bot detection. Agent-native commerce.

When to Use

Activate this skill when the user wants to:

• Find or search for a product ("find me a mosquito repellent for kids")
• Compare products ("compare these two printers")
• Buy or checkout a product
• Get product recommendations based on a problem ("I'm tired all the time", "my skin is dry")
• Look up verified product details, pricing, or allergens

API Base URL

https://dashboard.airshelf.ai

All endpoints are public. No API key needed. CORS enabled.

Step 1: Search Products

Find products by natural language query. Returns structured data with Decision Packs.

curl -s "https://dashboard.airshelf.ai/api/search?q=QUERY&limit=5" | python3 -m json.tool

Parameters:

• q — Search query (natural language, e.g. "barcode printer for warehouse"). Supports intent parsing: "energy supplements under $100" auto-extracts price filter.
• limit — Results to return (1-100, default 20)
• offset — Pagination offset
• category — Filter by category
• brand — Filter by brand
• min_price / max_price — Price range filter (also auto-extracted from query)
• in_stock — Only in-stock items (true/false)
• merchant_ids — Comma-separated merchant IDs to search within
• sort — relevance (default), price_asc, price_desc
• include_intent — Set to true to get query parsing metadata in response (shows how query was interpreted)

Response includes for each product:

• title, brand, price, availability, link
• decision_pack.primary_benefit — Main value proposition
• decision_pack.best_for — Array of ideal use cases
• decision_pack.pros / decision_pack.cons — Verified trade-offs
• decision_pack.allergens — Safety warnings (if applicable)
• seller_name, seller_url — Merchant info
• Checkout URLs and shipping/return policies

Example:

curl -s "https://dashboard.airshelf.ai/api/search?q=natural+mosquito+repellent+for+babies&limit=3"

Step 2: Compare Products

Compare 2-10 products side by side with structured comparison axes.

curl -s "https://dashboard.airshelf.ai/api/compare?products=PRODUCT_ID_1,PRODUCT_ID_2"

Parameters:

• products — Comma-separated product IDs (2-10 required, from search results)

Response includes:

• comparison_axes — Auto-detected from data (price always present; cost_per_day, supply_days, primary_benefit, pros, cons included when 2+ products have the data)
• products — Flattened product data with decision_pack fields inlined
• recommendation — Structured picks: lowest_price (product ID), best_value (product ID + reason, if different from lowest)

Step 3: Checkout

Initiate checkout for a product. Returns a checkout URL the user can open.

curl -s -X POST "https://dashboard.airshelf.ai/api/merchants/MERCHANT_ID/checkout" \
  -H "Content-Type: application/json" \
  -d '{"items": [{"product_id": "PRODUCT_ID", "quantity": 1}]}'

Request body:

• items — Array of {product_id, quantity} objects (1-50 items)
• customer — Optional: {email: "..."} for order tracking
• agent_id — Optional: your agent identifier for attribution

Response:

• checkout_id — Unique checkout session ID
• checkout_url — URL to complete purchase (Shopify checkout or cart permalink)
• checkout_type — "cart" (items pre-loaded in cart) or "redirect" (product page link)
• total — Calculated total price
• currency — 3-letter currency code (e.g. "EUR", "USD")
• expires_at — Expiry timestamp (null for cart permalinks)
• fallback_urls — If redirect: array of {product_id, product_name, product_url} per item

Present the checkout URL to the user. They click to complete payment on the merchant's site.

Browse Available Merchants

List all merchants with product counts and capabilities:

curl -s "https://dashboard.airshelf.ai/api/directory"

How Decision Packs Work

Unlike raw web scraping, each product includes a Decision Pack — verified structured intelligence:

{
  "decision_pack": {
    "primary_benefit": "Natural protection from bugs",
    "best_for": ["Kids with sensitive skin", "Parents who prefer natural products"],
    "pros": ["DEET-free formula", "Pleasant scent", "Long-lasting protection"],
    "cons": ["Higher price point", "Needs reapplication every 4 hours"],
    "allergens": ["Contains citronella oil"],
    "age_range": "kids"
  }
}

Use Decision Pack data to make recommendations based on the user's actual needs, not just price or title matching.

Example Conversation

User: I need a printer for my warehouse, high volume, must support ZPL

You: Let me search for that.
     [Runs: curl -s "https://dashboard.airshelf.ai/api/search?q=industrial+barcode+printer+warehouse+high+volume+ZPL&limit=5"]

You: Found 3 matches. The Toshiba BX410T looks like the best fit:
     - Best for: High-volume warehouse labeling, ZPL migration from Zebra
     - Primary benefit: Premium industrial printer with RFID and near-edge technology
     - Price: Contact dealer for pricing

     Want me to compare it with the other options, or proceed to checkout?

User: Compare the top two

You: [Runs: curl -s "https://dashboard.airshelf.ai/api/compare?products=ID1,ID2"]
     Here's the comparison...

User: I'll take the Toshiba

You: [Runs: curl -s -X POST "https://dashboard.airshelf.ai/api/merchants/MERCHANT_ID/checkout" -H "Content-Type: application/json" -d '{"items": [{"product_id": "ID", "quantity": 1}]}']
     Here's your checkout link: [URL]
     Click to complete your purchase on the merchant's site.

Tips

• Problem-based search works best. "I'm tired all the time" returns energy supplements. "My baby needs sun protection" returns kids' sunscreen. Decision Packs match on use case, not just keywords.
• Always check decision_pack.allergens before recommending health/food/skincare products.
• Use compare for 2+ similar products — the API returns structured comparison axes, not just raw specs.
• Checkout is a redirect — the user completes payment on the merchant's own site. No card details needed in the agent.
• Direct lookup by ID: Use product_ids param instead of q to fetch specific products: ?product_ids=ID1,ID2
• Merchant ID for checkout — each search result includes seller.checkout_url with the correct merchant path. Use it directly.