Shopify Catalog Change Monitor: Prices & Variants

Detects what changed in a public Shopify catalog between two points in time: newly listed products, removed products, added or removed variants, published price movements, compare-at price changes, stock flips and title edits. Output is one dataset row per change, not a full catalog dump.

It is built for recurring monitoring (for example a scheduled daily run), not for one-off exports. If you only need a current product list, a plain Shopify product scraper is a better fit.

How it works

1. First run per store reads the public /products.json feed and saves a compact baseline in a named key-value store (snapshotStoreName). No change records are produced, so nothing is charged.
2. Every later run reads the feed again, compares it with the saved baseline, emits change records, then replaces the baseline.
3. Offline mode (optional): supply baselineProducts and currentProducts in Shopify products.json format that you already hold. No network access happens.

Sample input

{
    "storeUrls": ["https://www.allbirds.com"],
    "maxProductsPerStore": 500,
    "minPriceChangePercent": 1,
    "changeTypes": ["product_added", "product_removed", "price_changed", "availability_changed"],
    "snapshotStoreName": "my-catalog-snapshots"
}

Replace the store with one you are entitled to monitor. Use the store's canonical domain: redirects are not followed, and the error message tells you where a store redirects to.

Input

Field	Meaning
storeUrls	1-10 public https:// Shopify store addresses.
maxProductsPerStore	1-5000, default 1000.
minPriceChangePercent	Ignore smaller price moves. Default 0.
changeTypes	Subset of the change types below. Empty means all.
maxChangeRecords	Per-store cap on emitted rows, default 1000.
resetBaseline	Replace the baseline with the current catalog; report nothing.
snapshotStoreName	Named key-value store holding baselines.
requestDelaySeconds	Pause between page requests, 0.5-30.
baselineProducts, currentProducts	Optional offline comparison.

Output

Change types: product_added, product_removed, variant_added, variant_removed, price_changed, compare_at_price_changed, availability_changed, title_changed.

Example row:

{
    "store": "https://www.example-store.com",
    "change_type": "price_changed",
    "product_id": "123",
    "title": "Example Shirt",
    "product_url": "https://www.example-store.com/products/example-shirt",
    "variant_id": "456",
    "variant_title": "M",
    "sku": "SHIRT-M",
    "old_value": "20.00",
    "new_value": "15.00",
    "price_delta": "-5.00",
    "price_change_percent": "-25.00",
    "detected_at": "2026-01-02T00:00:00Z",
    "baseline_taken_at": "2026-01-01T00:00:00Z"
}

For product_added and product_removed, variant_count and lowest_price are filled instead of old/new values. A per-store summary (status, product count, change counts, warnings, errors) is saved to the OUTPUT record of the run's key-value store.

Pricing and errors

The Actor emits one change-record event per change row, and only after the row is stored. Runs that fail, stores that error, baseline-creation runs and runs with no changes produce no charge events. The payable plan is configured separately on the platform; this repository does not set one.

Safety

• Only https:// store domains on the default port; IP addresses, credentials, local and internal host names are rejected.
• The host is resolved before use and must resolve only to public addresses.
• robots.txt is checked first; a disallow rule, or a 401/403/5xx on robots.txt, stops that store.
• Redirects are never followed; requests are sequential, delayed, size-limited and capped at 40 pages per store.
• No logins, proxies, cookies or personal data are used. Only product, variant, price and availability fields are kept.

Limitations

• Works only for stores that leave /products.json publicly enabled. Many stores disable or restrict it.
• The feed shows the storefront default prices. It does not show currency, per-market prices, discounts applied at checkout, customer-specific prices, or exact inventory counts (only an in/out-of-stock flag).
• Catalogs larger than maxProductsPerStore are truncated. A truncated current snapshot cannot prove removal, so removals are skipped; a truncated baseline cannot prove addition, so additions are skipped. Warnings are written to the summary.
• Draft, hidden or unpublished products are not visible; "removed" can also mean "unpublished".
• Change detection depends on schedule frequency: several changes between two runs collapse into the net difference.
• DNS is checked before connecting but not pinned to the connection; a hostile DNS change between check and request is not fully excluded.
• Shopify store data may be subject to the store's terms and applicable law. You are responsible for lawful use.
• Results from a few stores or a few runs are anecdotal; treat price movement statistics from small samples with caution.

Development

Offline tests: python -m unittest discover -s tests