01 / Start with the need
Discovery before identifiers
v2 accepts a natural-language need, resolves the economic concept, and only then presents official datasets. When coverage is missing, the response says so explicitly.
1. Resolve the need
curl --fail --silent "https://open-economics-data.knbf982hkn.chatgpt.site/api/v2/search?q=desemprego%20desde%202015"2. Inspect the official schema
curl --fail --silent "https://open-economics-data.knbf982hkn.chatgpt.site/api/v2/datasets/ibge-aggregates%3A6381/schema"The 32 v1 IDs remain stable; v2 brings together 12,886 searchable official datasets, explicit multidimensional selection, and direct access to national and subnational fiscal, Federal Public Debt, foreign-trade, fuel-price, electricity-consumption, formal-employment, and investment-fund data.
OpenAPI v2↗ OpenAPI v1↗02 / Shared core
Concept, dataset, dimension, and observation
The concept describes what a person needs independently of coverage. The dataset preserves official identity. Measures, classifications, and geographies preserve structure before observations are returned.
conceptResolved economic needdatasetOfficial identity and publisherdimensionsExplicitly selected breakdownsperiodOriginal reference periodraw_valueExact upstream valuestatusZero, suppression, and absence stay distinct03 / Semantics
Dates are not just labels
- Request dates use YYYY-MM-DD and are inclusive.
- BCB monthly series use day 01 as a reference-month marker.
- IBGE six-digit keys are parsed by frequency.
- PNAD moving quarters overlap; they are not monthly points.
04 / Revision policy
Official history can change
GDP, seasonal adjustments, and other series can be revised. History is refreshed and upserted rather than treated as append-only.