Before building Product Recall Finder, we looked at how the official US government recall sites actually work, and at what the underlying FDA API does to developers who build on it.
FoodSafety.gov and Recalls.gov are built to browse, not to search by product name
These are the official sources, and they're genuinely reliable — but their core interface is a list widget you scroll through, compiled from FDA and USDA feeds. There's no single box where you type "peanut butter" and immediately see whether it's ever been recalled. This tool leads with exactly that: type a product name, get a merged, dated list back immediately.
The FDA's own API has a classification trap
We found a detailed developer write-up on openFDA's enforcement API that flags a real gotcha: most implementations model recall severity as a 3-value enum — Class I, II, III — but a fourth value, "Not Yet Classified," actually shows up in the data and doesn't fit that enum. A tool that only handles three values either crashes on that record or silently mislabels it. This tool treats "Not Yet Classified" as its own distinct state, not a stand-in for the least-serious class.
Food, drug, and device recalls, checked at once
openFDA actually splits recalls across separate endpoints for food, drugs, and medical devices — and they don't share a consistent shape (the device recall endpoint doesn't even carry a classification field at all). This tool queries the three enforcement endpoints that do carry classification, side by side, and merges the results by date so you don't have to know which database your product might be in.
An honest limit, stated up front
FDA's own documentation says this data isn't meant to track a recall's live status — it's a historical enforcement record, not a real-time feed. Every result here links back to the official FDA record so you can confirm current status directly at the source, and the page says so plainly rather than implying more certainty than the data actually supports.
