How securiteconso.org Turns US Recall Notices Into a Reading List

Most people who need a recall notice find it by accident. The point of this site is to remove the accident: to take notices that are published, structured for regulators and written for manufacturers, and turn them into something you can read in a minute and act on in five.

Here is the whole machine, described honestly.

The pipeline, in five steps

Collect. Every six hours, a scheduled job pulls new records from the US Consumer Product Safety Commission recall feed and from the openFDA enforcement endpoints for food, drugs and medical devices.
Deduplicate. Records are keyed on source and identifier, so a notice is stored once regardless of how many times the feed returns it.
Summarise. Each record is summarised in English from the official text: what the product is, what the hazard is, what the agency says to do. The summary is a pointer, never a replacement — the source is linked on every card.
Illustrate where possible. If the source publishes photographs of the recalled product — CPSC notices often do — the image is copied into our own storage, served with a long cache lifetime and credited to the agency. Where the source has no image, the card has no image.
Publish and deliver. The result is a filterable list, an API, and alerts by category for people who would rather be told than remember.

What the list gives you

435 recall records as of this writing, refreshed every six hours.
Filters by source, severity, category and free-text search.
Pagination at 50 alerts per page, so a filtered result stays readable.
Photographs of recalled products where the agency publishes them — the fastest way to recognise an item you own.
A link to the official publication on every single entry.
A JSON API at /api/recalls for anyone who wants to build on the data.

What it deliberately does not do

It does not replace the notice. Summaries are generated automatically from official text; a model can be wrong. For medical devices, medicines and food, act on the official record, and when in doubt, ask a professional rather than a website.
It does not invent completeness. We track two US sources. A product recalled in another jurisdiction will not appear here unless a US agency also published it.
It does not dress up gaps. Food, drug and device records carry no product images — those cards are text. We would rather show you nothing than a stock photo implying a product we cannot identify.
It is not a government service. We are an independent information service and we are not affiliated with, or endorsed by, any authority. Every alert links back to the body that published it.
It does not sell your address. Alerts need an email address and nothing else; there is no advertising and no third-party tracking.

The two-minute version for a new reader

Open the recall list and scan the last few days. Most people recognise at least one product family they own.
Use the category filter for the things you actually buy — children’s products, household appliances, food.
If you would rather not check manually, set up an alert: choose categories, confirm your email, and notices arrive as they are published.

Where it goes next

The pipeline is deliberately narrow: two sources, one language for the summaries, no advertising. The most useful additions are the ones that shorten the distance between publication and an owner reading it — more category granularity in the alerts, and better identification fields surfaced from notices so that checking a model number takes less work.

If you spot a summary that misstates an official notice, that is a bug worth reporting: contact details are here.

Sources: CPSC recalls · openFDA · NHTSA recalls