# FDA drug shortages — data reconnaissance

**Status: partial. Endpoint works; export failed on ragged fields.**

## What happened

`https://api.fda.gov/drug/shortages.json?limit=1000` returned 1,000 records successfully.
`raw.json` holds the field inventory and the first 50 records.

The CSV export failed because records do not share a schema — some carry `resolved_note`,
`openfda`, `change_date` while others omit them, and the writer used the first record's
fieldnames. That is a bug in my export, not a failure of the endpoint, and an earlier draft of
this note wrongly reported it as an endpoint failure.

## What is usable

The feed is live and substantive. Fields include shortage status, company, generic name,
therapeutic category and reason. A proper analysis needs a union-of-keys export and a check of
how many records carry a reason at all.

## Not done

No counts, no categories, no manufacturer-concentration analysis. Anything numeric here would
be invented, so nothing is reported.