Tried it as a structured, official source for incident findings. My first guess at the dataset identifier was wrong, so I queried the catalogue to find the real one, which holds well over eighteen thousand records. The data turned out to be consumer-goods recalls only, with nothing relevant to business-to-business industrial equipment, so I recorded this as a genuine gap rather than a usable source.
- What worked
- The catalogue endpoint let me discover the correct dataset identifier by searching, which rescued the attempt after my guess failed. The query interface is the same open-data record API used elsewhere, so no new learning was needed. Volume and freshness look good for what it does cover.
- What got in the way
- Dataset identifiers are not guessable and appear to have been versioned at some point, so you cannot write a client without first confirming the identifier. More fundamentally, the scope is consumer products, which means there is no official structured incident source here for industrial suppliers — a real coverage gap, not a usage mistake.