Provider coverage audit (April 2026)

[lineoffligbot]

Currency expansion

All providers with hardcoded currency lists have been verified against their source APIs. None have additional currencies to add:

Provider	Currencies	Status
TCMB (Turkey)	21	All available — only XDR/SDR omitted (intentional)
BOE (England)	26	All active GBP spot rates covered
DNB (Denmark)	30	All active currencies covered (HRK/RUB defunct)
HKMA (Hong Kong)	17	All active currencies — legacy EUR predecessors return null
SARB (South Africa)	24	All daily series — only RUB excluded (sanctions)
BCB (Brazil)	10	PTAX API only offers these 10
FRED (US)	22	H.10 is a curated policy publication
BOI (Israel)	14	Already dynamic — fetches all available
BOB (Botswana)	6	CSV source only has these columns
BOJ (Japan)	2	BOJ only publishes USD/JPY and EUR/USD

Historical depth

Several providers had artificial coverage_start limits. Updated in April 2026:

Provider	Old start	New start	Extra years
BAM (Morocco)	2016	1999	+17
BOTA (Tanzania)	2015	1999-07	+15.5
BNM (Malaysia)	2021	2006	+15
RBA (Australia)	null	2023	Set correctly

Other providers checked and confirmed at correct limits:

• BCEAO (West Africa): 2014 is the actual API limit
• CBA (Armenia): 1999 is correct
• BI (Indonesia): Currently 2024, likely goes back to 2021+ (needs prod testing)
• BOC (Canada): Has legacy noon rates 2007-2017 with ~65 currencies under LEGACY_NOON_RATES group — needs adapter work to access

Known blockers

• Banguat (Guatemala): TLS/geo-blocked from non-LATAM IPs. GTQ is covered by NBP (Table B, weekly). Not worth pursuing.
• BOK (Korea): ECOS API requires Korean phone verification (issue Add BOK (Bank of Korea) provider #273, wontfix)
• BCRP (Peru): Only 2 currencies, EUR data has quality issues, PHP errors in responses. PEN already covered by 6 providers. Not worth adding.

New provider candidates

Tracked in individual issues:

• Add National Bank of Romania (BNR) provider #286 BNR (Romania) — RON, XML, no auth
• Add Central Bank of Uzbekistan (CBU) provider #287 CBU (Uzbekistan) — UZS, JSON, no auth
• Add National Bank of Kazakhstan (NBK) provider #288 NBK (Kazakhstan) — KZT, XML, no auth
• Add Magyar Nemzeti Bank (MNB) provider #289 MNB (Hungary) — HUF, SOAP, no auth
• Add National Bank of Moldova (BNM_MD) provider #290 BNM_MD (Moldova) — MDL, XML, no auth

Single-pair providers

These providers intentionally publish only 1-2 pairs — this is a source limitation, not an adapter issue:

• BANREP: USD/COP only
• BCCR: USD/CRC only
• BOJ: USD/JPY and EUR/USD only

[lineoffligbot]

Update — April 2026

New provider candidates — all shipped

All five candidates from the original audit are now live:

Issue	Provider	Status
#286	BNR (Romania)	Shipped
#287	CBU (Uzbekistan)	Shipped
#288	NBK (Kazakhstan)	Shipped
#289	MNB (Hungary)	Shipped
#290	NBM (Moldova)	Shipped

Additional providers added

Since the original audit, these providers were also integrated:

Provider	Currency	Notes
HNB (Croatia)	EUR	Mid rates via v3 API. Direction bug found and fixed — adapter was storing base/quote inverted (see #297)
LB (Lithuania)	EUR (historical LTL)	Historical rates, EUR era uses cross rates
BDI (Italy)	EUR	Banca d'Italia, ECB-sourced reference rates
BOJA (Jamaica)	JMD	Counter rates
SBI (Iceland)	ISK	Official rates
HKMA (Hong Kong)	HKD	End-of-day rates, 17 currencies
MAS (Singapore)	SGD	Daily exchange rates
BCU (Uruguay)	UYU	Arbitrage and interbank rates
NBRM (North Macedonia)	MKD	Daily mid-market, ~31 currencies
CBK (Kenya)	KES	Indicative rates
RBA (Australia)	AUD	Daily mid rates

Removed

Provider	Reason
Banco de Portugal	Added then removed — data was ECB-sourced with no unique value

Data quality

• HNB direction bug (Weekend rate spikes from inverted HNB base/quote #297): adapter stored rates inverted (base/quote swapped). Caused ~9% weekend spikes in range queries. Fixed in dd64854, production data corrected.
• Added rate sanity check step to the provider skill — new adapters now cross-reference against ECB and Wise mid-market rates before shipping.

Current provider count: 50

Still not pursuing

• Banguat (Guatemala): TLS/geo-blocked
• BOK (Korea): requires Korean phone verification (Add BOK (Bank of Korea) provider #273)
• BCRP (Peru): quality issues, already covered

[lineoffligbot]

Evaluated and decided against

For the record, these providers were researched and intentionally not pursued:

Provider	Reason
BIS (Bank for International Settlements)	Publishes NEER/REER indices (trade-weighted baskets), not bilateral spot rates
SNB (Switzerland)	Only 4 currency pairs vs CHF, all redundant with ECB coverage (#245)
SAMA (Saudi Arabia)	No API — ASP.NET page with __doPostBack pagination. High value (Gulf currencies) but engineering cost prohibitive (#248)
RBNZ (New Zealand)	CSV/Excel only, no programmatic API (#236)
CBUAE (UAE)	No usable API (#243)
Banguat (Guatemala)	TLS/geo-blocked from non-LATAM IPs. GTQ covered by NBP
BOK (Korea)	ECOS API requires Korean phone verification (#273)
BCRP (Peru)	Only 2 currencies, EUR data has quality issues. PEN covered by 6 other providers
Banco de Portugal	API covers 1987-1998 only (pre-euro PTE). Frozen dataset with no ongoing value

Note: SBI (#234), MAS (#232), BI (#244), and BOJ (#239) were initially marked wontfix but later implemented after finding workable API paths.

[lineoffligbot]

Global central bank audit (April 2026)

Researched 105 central banks not yet integrated. Each was checked for publicly accessible exchange rate APIs.

New issues created

Issue	Provider	Currency	Format	Notes
#298	NRB (Nepal)	NPR	JSON REST	22 currencies, historical, no auth
#299	CBM (Myanmar)	MMK	JSON REST	38 currencies, no auth, current only
#300	BCN (Nicaragua)	NIO	SOAP	Daily since 2012, adds NIO depth
#301	NBS (Serbia)	RSD	SOAP	Unique currency, needs license
#302	BCH (Honduras)	HNL	REST	Unique currency, needs API key
#303	BCRD (Dominican Rep)	DOP	REST	Swagger API, needs token

EUR-only providers (skipped)

These eurozone NCBs have APIs but only republish ECB data with no unique currencies:

BBK (Germany, SDMX), BdF (France, Webstat), OeNB (Austria, SDMX), NBB (Belgium, SDMX-JSON), BOF (Finland, API portal), BS (Slovenia, JSON REST), BdE (Spain, JSON REST), DNB_NL (Netherlands, JSON REST), BNB (Bulgaria, XML — BGN is pegged to EUR)

No usable API (75 central banks)

HTML/PDF only, inaccessible, restricted, or pegged currencies with no value:

Europe: Cyprus, Estonia, Greece, Ireland, Latvia, Luxembourg, Malta, Slovakia, Albania, Bosnia, Turkmenistan, Tajikistan, Azerbaijan

Asia-Pacific: Bangladesh, Brunei, Laos, Macau, Fiji, Mongolia, Papua New Guinea, Sri Lanka, Vietnam, Maldives (needs token), Pakistan (Excel only), Philippines (Excel only)

Middle East: Bahrain, Egypt (blocked), Iran, Iraq, Jordan, Kuwait (Excel only), Lebanon (Excel only), Libya (restricted), Oman, Qatar, Tunisia, Yemen

Africa: Algeria, Sudan, Nigeria (undocumented), Ghana (monthly), Gambia (PDF), Sierra Leone, Liberia, Guinea, Cabo Verde, Mauritania, BEAC/CEMAC, Burundi, Djibouti, Comoros, Seychelles, Angola, DRC, Mozambique, Namibia, Eswatini, Lesotho, Sao Tome

Americas: Bolivia, Guyana, Paraguay (403), Suriname (PDF), Venezuela, Belize (500 error), Haiti

Caribbean/Pacific: ECCB, Bahamas, Barbados, Bermuda, Cayman, Curacao, Trinidad, Aruba, Tonga, Samoa, Vanuatu, Solomon Islands, IEOM

OpenData portals (deferred)

Uganda, Rwanda, Zambia, Malawi, Madagascar, Ethiopia, Cambodia — have data via opendataforafrica.org or similar third-party aggregators. Not pursuing until we decide on the aggregator model.

Methodology

Dispatched 10 research agents to survey all 105 CBs, then ran a second pass with endpoint verification on the 12 strongest candidates. Only CBs with working, documented programmatic endpoints (JSON/XML/SOAP) that return actual rate data were promoted to issues.

[JJLizarazu]

Technical insights on BCB (Bolivia) and potential scraper approach

Great audit, @lineoffligbot. Regarding the 'No usable API' list, specifically Bolivia (BCB):

As a developer based in Bolivia, I’ve been monitoring the Banco Central de Bolivia portal. You are correct that they don't offer a REST/JSON endpoint, but they do have a consistent structure in their daily HTML tables (UFVs and Exchange Rates).

If the project is open to it, I could:

1. Verify if there's a hidden SOAP/WSDL endpoint (sometimes used for national interbank systems here).

2. Propose a lightweight scraping adapter using the existing provider skills to parse their daily HTML, as the BCB site is quite stable in its DOM structure.

Also, for the BOC (Canada) legacy rates mentioned, I'd be happy to help look into the LEGACY_NOON_RATES group logic if you need an extra pair of hands for that adapter work.

Amazing work on the HNB direction bug fix, those weekend spikes were trick

[lineoffligbot]

Thanks for the offer.

BOC legacy noon rates: yes please. The current adapter only pulls FX_RATES_DAILY (2017+); the LEGACY_NOON_RATES group with ~65 currencies covering 2007 to 2017 is untouched. Real historical depth, well-scoped, go for it.

BCB Bolivia: worth a look if there is a clean programmatic endpoint behind the page. We already publish BOB via BDI (ECB-sourced), so it is not a coverage gap, but a second voice with longer history would be welcome. DOM-only scraping we would rather not take on.