Data
The sources the skills are written against.
The skills teach a model how to handle these sources correctly — what their series identifiers encode, which figures are seasonally adjusted, which are revised and how far. Access tooling is on the roadmap and is not built; this page is the source list the guidance covers, not a running service.
Danish depth
- Statistics Denmark (StatBank) Covered by the skills
Labour market, income, wages, and prices. Covered: StatBank table IDs, register versus survey basis, and Danish revision practice.
- Danmarks Nationalbank Not yet documented
Monetary and financial series. Not yet documented in the skills — the conventions differ enough from ECB practice to need their own treatment.
- Public finance and tax Not yet documented
Government finances and taxation. Not yet documented — the accrual/cash distinction and the ESA general-government perimeter need writing up properly.
Global coverage
- FRED Covered by the skills
Federal Reserve Bank of St. Louis. The skills cover its series-ID conventions, the SA/NSA and real/nominal suffixes, and ALFRED for vintages.
- Eurostat Covered by the skills
EU-wide statistics for cross-country comparison. Covered: dataset codes, dimension filters, chained-volume non-additivity, and ESA 2010 breaks.
- World Bank Covered by the skills
Global development, trade, and income indicators. Covered: indicator codes, PPP versus market exchange rates, and constant-price base years.
- IMF Covered by the skills
International financial statistics and balance of payments. Covered: SDMX dataflow and key structure, and WEO vintages as forecast rounds.
Provenance
The rule the skills enforce: no series is used without source, identifier, units, period, transformation, and vintage. Where the adjustment status or deflator cannot be determined, the instruction is to refuse rather than guess.