Overview
Product recalls are scattered across at least three federal agencies, each with its own site, format, and feed quality. WatchRecall aggregates CPSC, NHTSA, and FDA recall announcements into one searchable, browsable feed — so "was this stroller ever recalled?" is one search instead of three government websites.
How it works
- Daily automated ingestion — a pipeline pulls each agency's published recall data every day, normalizes the wildly different schemas into one shape, and dedupes cross-postings.
- Search-first — recalls are indexed by product, brand, and hazard so the site works as a lookup tool, not just a news feed.
- Boring on purpose — no editorializing; the value is normalization and a permanent, linkable page per recall.
Stack
A Python ingestion pipeline on a daily cron writing to Postgres, with a Next.js front end on Vercel serving the browse and search pages.