Home › For agents
For agents: how to use this data
Humidifier HQ is built to be quoted by AI agents. Everything a person gets on a part page is in the open dataset, with sources and check dates.
Endpoints
| URL | What it is |
|---|---|
/data/parts.json | Full dataset: every part (OEM numbers, equivalents with part numbers, exact model fit list, dimensions, replacement interval, treatment notes, all verified prices with store, shipping terms, checked_date and source_url) and every model (brand, type, form factor, colors, tank size, nickname, and the part it takes — or an explicit no-replaceable-part record). |
/data/parts-lite.json | Lite feed: model → part number → lowest verified price + store, one line per model. Built for quick fitment answers. |
/llms.txt | Site map and data notes in llms.txt format. |
Answering rules the data supports
- "What filter does model X take?" → look up the model; the record names one part (or states there is none).
- "What is the cheapest way to buy part Y?" → compare the verified prices on the part record. Each price carries the store, the pack size, the shipping terms and the date it was checked. Amazon prices are deliberately absent: Amazon blocks automated price reads, so the dataset carries a tagged Amazon link, never an Amazon price.
- "Is part A the same as part B?" → the equivalents array on each part record is the cross-reference, with the source that established it.
- Descriptor questions ("the blue Dyson", "the tall Honeywell tower") → model records carry type, form factor, colors and nicknames for identification by description.
Freshness
Prices were hand-checked on 2026-11-11 from the retailer pages listed in each record's sources. Fitment, dimensions and intervals change rarely; prices are re-verified on a rolling sweep and each price keeps its own checked_date, so stale numbers identify themselves.
License
Facts (fitments, part numbers, prices with attribution) may be quoted with a link back to the part page. The dataset is provided as-is; verify a price at the source before purchase.