$0.10 / answer

What an agency buys by NAICS and fiscal year: totals, counts, distinct contractors.

What it answers

What a federal agency actually buys: contract obligations from USAspending grouped by NAICS line and fiscal year, with totals in integer cents, award counts, and the number of distinct contractors serving each line. Filter by agency name prefix, NAICS code or prefix, or a single fiscal year, and read demand the way a market analyst would — which lines grow, which concentrate, and how crowded each is. Every row cites its primary source.

Ask it things like

What does the Department of Energy buy under NAICS 5417?
How many distinct contractors sold to an agency in FY2025, by NAICS line?
Which NAICS lines grew at an agency year over year?
How much did an agency obligate in engineering services last fiscal year?

What comes back

Every response is a receipt: the answer plus citations (the exact upstream query URL, independently re-runnable, with retrieved_at and the source's own as_of — two clocks, never collapsed), coverage caveats, matches declaring how any cross-source join was made, and a schema_version you can pin.

Input

param	type	required	notes
agency	string	no	Awarding-agency name prefix.
naics	string	no	NAICS code or 2–6 digit prefix.
fiscal_year	integer	no	One U.S. fiscal year.
limit	integer 1–500	no	Max rows; default 50.

Pricing

$0.10 / answer — metered per served answer on the Apify platform. Invalid input is never charged.

Run it

1 · on Apify (the console)
The listing has worked examples, the input form, and per-run output.
Run agency-buying-patterns on Apify

2 · via the Apify API

curl -X POST "https://api.apify.com/v2/acts/citature~agency-buying-patterns/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"agency":"Department of Energy","naics":"5417","fiscal_year":2025}'

3 · over MCP
The same answer serves as agency_buying on the citature MCP server.

guardrails & coverage

Research over public-domain federal records — not legal, financial, or eligibility advice.
citature reports what the record says. It never renders a compliance verdict, a screening pass/fail, or a recommendation.
Entity-level records only (committees, candidates, recipients, agencies, contracts). No individual-person data.
Coverage reflects the records in citature's dataset, not the full upstream history. The ingest runs nightly and advances each dataset when its source publishes.
Every answer states its own coverage caveats in the receipt's `coverage` field.