Catalog Release Guard - Price and Stock Regression Audit
Pricing
$0.05 / completed catalog release audit
Catalog Release Guard - Price and Stock Regression Audit
Compare two catalog snapshots before a release. Catch mass removals, stockout spikes, currency changes and large price movements, with expected-change allowances and traceable release gates.
Pricing
$0.05 / completed catalog release audit
Rating
0.0
(0)
Developer
Mahruf Hossain
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
13 days ago
Last modified
Categories
Share
Catalog Release Guard
Compare a baseline catalog with a proposed replacement before publishing a product-feed or pricing update. Get a machine-readable PASS/HOLD decision, per-product changes, explicit review gates and SHA-256 fingerprints of both inputs.
Built for commerce operations teams and agencies reviewing catalog releases. Unlike a live-page scraper or a simple list of price changes, this Actor checks aggregate release risk and supports expected-change allowances. It does not crawl stores, submit feeds or modify products.
Quick start
{"before":[{"id":"DEMO-1","price":"19.99","currency":"USD","availability":"in_stock"}],"after":[{"id":"DEMO-1","price":"1.99","currency":"USD","availability":"out_of_stock"}]}
The example intentionally returns HOLD: a large price drop and an unexpected stockout need review. Use your own catalog snapshots, not customer/order records.
Inputs
Both snapshots accept at most 1,000 products; total JSON input must fit within 2 MB. before must be nonempty. Each row requires:
| Field | Value |
|---|---|
| id | Stable product ID, 1–100 characters |
| price | Nonnegative decimal string, up to 12 whole digits and 6 decimals |
| currency | Three uppercase currency-code letters; code membership is not validated |
| availability | in_stock, out_of_stock, preorder or backorder |
| group | Optional label, 1–100 characters; retained for input fingerprinting, not segmented gating |
Unknown fields are rejected to prevent accidentally passing customer data. No URLs, API keys, cookies or personal contact details are required. Do not put secrets or personal information in product IDs or group labels.
Optional thresholds (percentages, 0–100): maxRemovedPercent (default 5), maxOutOfStockIncreasePercent (10), maxPriceChangePercent (25). A movement equal to a threshold passes; a greater movement holds. Price movements compare absolute change against the baseline price, using integer arithmetic at six decimal places. Both increases and decreases are checked.
expectedChanges accepts { "id": "SKU-1", "field": "price" }, with fields price, availability, removed or added. Allowances prevent those changes from triggering related movement gates; changes remain visible and labeled expected. Currency switches, duplicate IDs and zero prices cannot be waived this way. Removal and stockout percentages use baseline unique product count; stockouts count in_stock → out_of_stock transitions among retained, unambiguous IDs.
Output and integration
One completed audit record is written to the default dataset and OUTPUT key-value record. It contains releaseDecision, summary, gates, changes, thresholds, baselineSha256 and candidateSha256. Export JSON for a release checklist or a CI/workflow condition. A HOLD is a successful audit result, not a failed Actor run. Invalid inputs fail the run with a safe message and no completed-audit event charge.
Pricing
One audit-completed event per successfully saved report, covering up to 1,000 products per snapshot. The Store pricing panel is authoritative. Expected initial price: $0.05 per completed audit; any platform/start charges are shown separately there. No subscription or third-party paid API is required.
Limits, privacy and interpretation
Only supplied snapshots are checked. PASS means the configured checks passed, not that a catalog is commercially correct or eligible for Google Shopping. This is not a complete Google Merchant specification validator, tax validator, currency registry, stock-level reconciliation or live website consistency check. No data is sent to third-party websites or AI models. Input/output are stored in your Apify run storage under Apify's retention/access rules; do not submit confidential or personal data. Logs contain summary counts only.
Relevant guidance: Google product data specification and price mismatch guidance. These explain the operational context, not certification or endorsement.
Development
Run node --test test.mjs. The comparison core has no external dependencies. The cloud wrapper uses the official Apify SDK in the Apify Node 24 base image. Development fixtures are synthetic and labeled DEMO.