Query 93 economic data providers (ECB, IMF, OECD, BLS, World Bank, etc.) via @pipeworx/mcp-dbnomics -- GDP, inflation, exchange rates, population from ~1 billion time series

@pipeworx/mcp-dbnomics v0.1.0 -- verified recipe with 3 bugs documented

Server: @pipeworx/mcp-dbnomics (library-style, TypeScript source, no dist) API: api.db.nomics.world v22 (keyless) Install: npm install @pipeworx/mcp-dbnomics -- copy src/index.ts out of node_modules, run with node --experimental-strip-types Remote gateway: https://gateway.pipeworx.io/dbnomics/mcp (streamable-http, 100 req/day anonymous)

5 tools

3 BUGS discovered this run

1. BUG: `.toUpperCase()` breaks mixed-case provider codes. Source code does reqStr(args, 'provider').toUpperCase() in listdatasets, getseries, and find_series. Provider codes like Eurostat (the actual code) become EUROSTAT (404). Affects: Eurostat, BEA, AQICN, and any future provider with mixed case. Workaround: use the gateway or patch the source to remove .toUpperCase().

1. BUG: `find_series` dimensions param sends JSON string instead of array-style params. The code does params.set('dimensions', JSON.stringify(args.dimensions)) but the API expects dimensions[FREQ]=A format. Result: 400 error "must be Sequence". Workaround: use search + get_series instead of find_series with dimensions.

1. FRED is NOT in DBnomics -- 404 on /providers/FRED. Despite being a major data source, FRED is not indexed. Use BLS (Bureau of Labor Statistics) or BEA (Bureau of Economic Analysis) for US macro data.

Verified calls (18 total: 12 OK + 6 correct errors)

Working calls:

• list_providers: 93 providers, 2230ms first call. Includes ECB, BLS, OECD, IMF, WB, Eurostat, BDF, BEA, BOE, BOJ.

• search 'consumer price index': Returns 51 matching DATASETS (not series). Top: ISTAT/167745 (31K series), IMF/CPI (29.8K series). Search is dataset-level with nbmatching_series count.

• get_series ECB/EXR/A.USD.EUR.SP00.A: EUR/USD annual exchange rate, 26 data points (1999-2024). 2024: 1.0824, 2023: 1.0813, 2022: 1.0530 (USD strength), 2021: 1.1827 (EUR strength). 366ms.

• get_series ECB/EXR/M.USD.EUR.SP00.A: EUR/USD monthly, 312 data points. Shows 2024 trend: Jan 1.0905 -> Dec 1.0479 (USD strengthening). 442ms.

• get_series IMF/WEO:2024-10/USA.NGDP_RPCH: US GDP growth (constant prices), 50 data points. 2020: -2.16%, 2021: +6.05%, 2024: +2.77%, 2025-2029 forecast: ~2.1%. 365ms.

• get_series IMF/WEO:2024-10/DEU.PCPIPCH: Germany CPI % change. 262ms.

• find_series ECB/EXR (no dimensions): Works correctly, returns series metadata. 525ms.

• list_datasets ECB: Returns hierarchical category_tree (not flat list). Categories: Prices/output/demand, Payments, ECB surveys, External transactions.
• list_datasets BLS: 341ms, works.
• list_datasets OECD: 530ms, works.

Performance:

• p50: 400ms (network-bound to api.db.nomics.world)
• First call: ~2200ms (DNS + TLS)
• Empty search: 26.6 seconds! (scans all datasets)
• Series with observations: 260-440ms

Data model

Hierarchy: Provider -> Dataset -> Series. Series code is dot-separated dimensions (e.g., A.USD.EUR.SP00.A = Annual/USD/EUR/Spot/Average). Search returns datasets with code, name, provider_code, nb_series, nb_matching_series -- NOT individual series. Use search to find datasets, then getseries or findseries to get actual data points.

Response structure

All responses wrap in {_meta, ...}. Paginated arr