Screen a Product Catalog for the FDA Food Traceability List

Export stable product IDs, descriptions, supplier IDs, ingredient information, and available product-form fields. A row called 'MIX 12CT' contains too little context. Obtain a meaningful description before assuming an empty lookup means the product is outside the list.

Keep the original description beside any normalized text. Normalization can remove pack-size noise, but it should not discard words such as frozen, fresh-cut, or shelf-stable. Those words can change how an FDA category applies. Treat supplier aliases as additional evidence rather than replacing the product identity.

Create states such as candidate match, insufficient information, reviewed listed product, and reviewed exclusion or exemption. Preserve the candidate matches even when a reviewer rejects them. That history explains both the automated suggestion and the final decision.

For a first pass, use the search endpoint once per distinct meaningful description in your own workflow. FoodChainAPI's current public search is a single-query endpoint; do not describe this as a hosted bulk service. Control concurrency and retain errors so a partial run cannot appear complete.

Inspect a sample from each review state, including apparently easy matches. A run returning no matches for a supplier's entire produce catalog may indicate abbreviated inputs or a parsing error. Count unresolved rows separately from reviewed products.

Use the FoodChainAPI lookup to inspect examples before building automation. A catalog-level classification record belongs with the product, while receiving and shipping records belong with actual lots and events. Keep those layers linked without confusing them.