parsimony-treasury 0.0.1

U.S. Treasury (Fiscal Data + Office of Debt Management rate feeds) connector for the parsimony framework

parsimony-treasury

US Treasury connector plugin for parsimony — federal fiscal datasets (debt, revenue, spending, securities) from the Bureau of the Fiscal Service, plus the Office of Debt Management daily rate feeds (par yield curve, bill rates). Keyless — no API credentials required.

Part of the parsimony-connectors monorepo. Distributed standalone on PyPI as parsimony-treasury.

Connectors

Name	Kind	Description
treasury_fetch	connector	Fetch any Fiscal Data API endpoint (e.g. v2/accounting/od/debt_to_penny) as a DataFrame, with optional filter, sort, and page_size.
treasury_rates_fetch	connector	Fetch one Office of Debt Management rate feed (par yield curve, real yield curve, bill rates, long-term, real long-term) for a calendar year from the home.treasury.gov OData/Atom XML feed.
enumerate_treasury	enumerator	Enumerate every addressable Fiscal Data time-series measure plus the ODM rate-feed benchmarks for catalog indexing (drives the treasury catalog).
treasury_search	connector	Semantic-search the published treasury catalog and return matching code + title + score rows. Routing: home/<feed> → treasury_rates_fetch; v<n>/<endpoint>#<field> → treasury_fetch.

Install

pip install parsimony-treasury

Pulls in a compatible parsimony-core automatically. Verify discovery:

python -c "from parsimony import discover; print([p.name for p in discover.iter_providers()])"

Quick start

from parsimony_treasury import CONNECTORS

result = CONNECTORS["treasury_fetch"](
    endpoint="v2/accounting/od/debt_to_penny",
    sort="-record_date",
    page_size=10,
)
print(result.data.head())

For multi-plugin composition:

from parsimony import discover
connectors = discover.load_all()

Catalogs

This plugin ships a treasury catalog (~914 entries) driven by enumerate_treasury. treasury_search loads a published snapshot (overridable via the PARSIMONY_TREASURY_CATALOG_URL env var) and falls back to building one in-process when no snapshot is reachable. Maintainers build and push the snapshot with scripts/build_catalog.py.

Two enumeration sources:

• Fiscal Data — self-tracking (archetype A). The live /services/dtg/metadata/ JSON (the same source the fiscaldata.treasury.gov site consumes) lists every dataset → endpoint → field in one call: 879 measure fields across 180 fetchable endpoints / 53 datasets (live count, 2026-06-09). A new dataset appears in the catalog automatically on the next build.
• Office of Debt Management rate feeds — curated registry (archetype D). The 5 daily interest-rate feeds (par yield curve, real yield curve, bill rates, long-term, real long-term) are not in the metadata endpoint; their feed set is the stable interest-rate-statistics dropdown. scripts/harvest_rate_feeds.py cross-validates the registry's benchmark maturities against the live feed columns.

Coverage

Catalog covers ALL: yes (the metadata endpoint is the authoritative, self-tracking enumeration). Connectors cover ALL accessible data: partial by design — two documented exclusions:

• 3 static-file-only datasets — Monthly Treasury Disbursements, Combined Statement, and Account of Receipts and Expenditures carry no queryable API endpoint (no endpoint_txt); they are published only as static files, so there is nothing to fetch via the JSON API.
• Treasury Coupon Issues + HQM Corporate Bond Yield Curve — a separate product distributed as binary .xls 5-year archives (monthly, for Pension-Protection-Act actuarial use), not part of the daily XML feed family. The HQM spot-rate series are already available through the parsimony-fred connector (HQMCB20YR, etc.), so this connector defers to FRED for them.

Provider

• Homepage: https://fiscaldata.treasury.gov
• API docs: https://fiscaldata.treasury.gov/api-documentation/

License

See LICENSE.