open-fedspend-data

A free, daily-updated dataset of US federal contract awards, pulled directly from USAspending.gov's public search API and normalized into one schema.

Current snapshot: 28,092 contract awards, window 2026-07-29 to 2026-08-04 (rolling 7 days), totaling $436,328,109,541.33 obligated (last full run: 2026-08-04, ~135s). Numbers move daily as the window rolls forward — see data/summary.json for the current window and totals.

Data lives in data/:

• data/awards.json — every award in the window, as JSON
• data/awards.csv — same data as CSV
• data/top-recipients.json — recipients ranked by total obligated in the window
• data/by-agency.json — totals by awarding agency
• data/by-naics.json — totals by NAICS industry code
• data/summary.json — window dates, totals, run stats
• data/README.md — full schema doc and what the money fields mean

A GitHub Actions workflow (.github/workflows/update.yml) re-runs the fetcher every day and commits whatever changed. No manual updates.

Why this exists

USAspending.gov publishes every federal prime contract award since FY2008 through a public, unauthenticated JSON search API — no key, no login, no rate limit registration. It's the same API that powers usaspending.gov's own search page. This repo fetches a rolling recent window of contract awards daily, normalizes it, computes a few useful aggregates, and commits it, so you don't have to write your own pagination/retry logic against the raw API to get a clean CSV.

The window

Rolling 7 days, recalculated on every run: today back through 6 days ago (UTC, inclusive). Not a fixed calendar range — it shifts forward one day each time the fetcher runs. 7 days, not 30, because a 30-day window of every federal contract routinely runs past what USAspending's search endpoint can reliably paginate through in one run (see Limitations). A week of contract awards across every US federal agency is ~25,000-30,000 rows — substantial, but a single fetch completes in a couple of minutes.

Schema

Field	Type	Notes
awardId	string | null	USAspending's Award ID (piid/fain)
recipientName	string | null	Vendor/recipient legal name
recipientUei	string | null	Recipient's Unique Entity Identifier, when disclosed
awardingAgency	string | null	Top-tier awarding agency
awardingSubAgency	string | null	Sub-agency/component
awardAmount	number | null	Obligated amount for this award transaction — not a ceiling/potential value, see below
totalOutlays	number | null	Actual dollars disbursed to date, where reported
awardType	string | null	e.g. "DEFINITIVE CONTRACT", "PURCHASE ORDER"
naicsCode / naicsDescription	string | null	Industry classification
pscCode / pscDescription	string | null	Product/Service Code
placeOfPerformanceState / placeOfPerformanceCountry	string | null	Where the work is performed
startDate / endDate	string | null	Award's period of performance
description	string | null	USAspending's award description, as published
usaspendingUrl	string | null	Direct link to the award's public page on usaspending.gov
scrapedAt	string	ISO timestamp of the fetch that produced this row

Fields USAspending doesn't disclose come back as JSON null, never 0 or false.

Sample rows (from the live run above, verified against usaspending.gov)

A large multi-year vehicle:

{
  "awardId": "DENA0003525",
  "recipientName": "NATIONAL TECHNOLOGY & ENGINEERING SOLUTIONS OF SANDIA, LLC",
  "recipientUei": "LUJEPCRRT377",
  "awardingAgency": "Department of Energy",
  "awardingSubAgency": "Department of Energy",
  "awardAmount": 42774168231.25,
  "totalOutlays": 26301056802.91,
  "awardType": "DEFINITIVE CONTRACT",
  "naicsCode": "561210",
  "naicsDescription": "FACILITIES SUPPORT SERVICES",
  "pscCode": "M1JZ",
  "pscDescription": "OPERATION OF MISCELLANEOUS BUILDINGS",
  "placeOfPerformanceState": "NM",
  "placeOfPerformanceCountry": "UNITED STATES",
  "startDate": "2017-01-18",
  "endDate": "2027-04-30",
  "description": "IGF::CL,CT::IGF CONTRACT AWARD DE-NA0003525 TO THE NATIONAL TECHNOLOGY&ENGINEERING SOLUTIONS OF SANDIA, LLC (NTESS) FOR THE MANAGEMENT AND OPERATION OF THE DEPARTMENT OF ENERGY, NATIONAL NUCLEAR SECURITY ADMINISTRATION'S SANDIA NATIONAL LABORATORIES (SNL)",
  "usaspendingUrl": "https://www.usaspending.gov/award/CONT_AWD_DENA0003525_8900_-NONE-_-NONE-",
  "scrapedAt": "2026-08-04T04:56:47.702Z"
}

A more typical, smaller award (most rows in the dataset look like this, not the one above):

{
  "awardId": "140P4226P0018",
  "recipientName": "CARAHSOFT TECHNOLOGY CORP",
  "recipientUei": "DT8KJHZXVJH5",
  "awardingAgency": "Department of the Interior",
  "awardingSubAgency": "National Park Service",
  "awardAmount": 181429.52,
  "totalOutlays": null,
  "awardType": "PURCHASE ORDER",
  "naicsCode": "334310",
  "naicsDescription": "AUDIO AND VIDEO EQUIPMENT MANUFACTURING",
  "pscCode": "5836",
  "pscDescription": "VIDEO RECORDING AND REPRODUCING EQUIPMENT",
  "placeOfPerformanceState": "VA",
  "placeOfPerformanceCountry": "UNITED STATES",
  "startDate": "2026-07-28",
  "endDate": "2031-08-31",
  "description": "INDEPENDENCE NATIONAL HISTORICAL PARK (INDE) - URGENT - PURCHASE OF TWO (2) - FEDRAMP CERTIFIED PHYSICAL AND VIDEO TRAILERS AND SOFTWARE LICENSING",
  "usaspendingUrl": "https://www.usaspending.gov/award/CONT_AWD_140P4226P0018_1443_-NONE-_-NONE-",
  "scrapedAt": "2026-08-04T04:56:47.702Z"
}

Both URLs above were checked live and return HTTP 200 on usaspending.gov.

Using the data

curl

curl -s https://raw.githubusercontent.com/ConorsCode/open-fedspend-data/main/data/awards.json | jq '.[0]'
curl -s https://raw.githubusercontent.com/ConorsCode/open-fedspend-data/main/data/top-recipients.json | jq '.[:10]'

pandas

import pandas as pd

df = pd.read_json("https://raw.githubusercontent.com/ConorsCode/open-fedspend-data/main/data/awards.json")
df.groupby("awardingAgency")["awardAmount"].sum().sort_values(ascending=False).head(10)

JavaScript

const awards = await fetch(
  "https://raw.githubusercontent.com/ConorsCode/open-fedspend-data/main/data/awards.json"
).then((r) => r.json());

const dod = awards.filter((a) => a.awardingAgency === "Department of Defense");
console.log(`${dod.length} DoD awards this window`);

Source and attribution

All data comes from USAspending.gov, the official open data source for US federal spending, run by the US Department of the Treasury. This repo queries its public spending_by_award search API directly — no scraping, no key, no login. USAspending data is public domain; this repo's normalization/aggregation code is MIT licensed.

Field set and pagination — verified, not assumed

USAspending's fields parameter is not validated against one fixed schema the way a typed API would be, and the sort field must appear in fields or the request errors outright. The 19-field request list used here (Award ID, Recipient Name, Award Amount, Total Outlays, NAICS/PSC codes and descriptions, place of performance, dates, etc.) was confirmed live against the contracts award-type category before this repo shipped.

More importantly: USAspending's page_metadata.hasNext flag is unreliable past roughly page 100 for this endpoint when sorting by Award Amount desc — it was observed live returning hasNext: false on a page that still had a full 100 results, for several consecutive pages, with the real end of results only showing up ~180 pages later as a short page. scripts/fetch.mjs does not trust hasNext; it pages until a response comes back with fewer than 100 rows, up to a 49,900-row safety cap (USAspending's own documented ceiling for simple page-number pagination is page * limit <= 50,000). If a run ever hits that cap, data/summary.json.hitPaginationCap is true and awards.json holds only the largest awards by awardAmount, not a complete set for that window — check this field before treating a snapshot as exhaustive.

Limitations (read before relying on this for anything important)

• 7-day window only. This is not historical data and not a full-year view. For anything further back, use USAspending's own bulk download tools or query the API directly with a custom date range.
• Contracts only, no grants. Award type codes A/B/C/D (BPA calls, purchase orders, delivery orders, definitive contracts). Grant/assistance awards (block/formula grants, cooperative agreements) are out of scope.
• Obligated, not potential, amounts. awardAmount is what's been obligated on this award transaction, not a contract ceiling or ask price. A row can represent one action on a much larger multi-year vehicle — see data/README.md for detail.
• Snapshot timing. Each row reflects what USAspending returned at scrapedAt; award data can lag real contract actions by days, and USAspending occasionally revises historical records.
• No SAM.gov opportunities. This is awarded, obligated spending only — not open solicitations or pending opportunities. That's a different USAspending/SAM.gov dataset entirely.
• Sorted by awardAmount descending, capped for safety. If a window ever produces more rows than the fetcher's safety cap can retrieve, the smallest awards for that window are the ones dropped, not a random sample — check data/summary.json.hitPaginationCap.
• No personal data. This dataset contains only what USAspending publishes about the award itself — recipient organization name/UEI, never individual data.

If you need custom filters

This repo's fetcher is deliberately narrow: a fixed 7-day window, contracts only, no agency/NAICS/recipient filtering, run once a day via GitHub Actions. That's what keeps it free, fast, and easy to audit.

If you need award or grant data filtered by keyword, NAICS code, PSC code, awarding agency, place of performance, or a custom date range — run on demand instead of waiting for tomorrow's snapshot — that's a separate, paid tool: the Federal Contracts & Spending Scraper on Apify. It queries the same USAspending API plus optional SAM.gov opportunities (with your own free SAM.gov key), pay-per-result.

This repo and that actor share no code — the actor is a TypeScript project with input validation and per-source status reporting; this repo is a single plain Node script meant to be read in one sitting.

License

MIT — see LICENSE. The code is MIT licensed; the award data itself is US government public data from USAspending.gov.

Related open datasets

Part of a small set of free, daily-refreshed datasets built the same way: zero-dependency Node fetcher, GitHub Actions refresh, public endpoints only.

• Open Jobs Data — 15,919 open job postings from 90 companies across nine applicant tracking systems.
• Open Dependency Risk — 1,500 widely-used npm and PyPI packages joined with known vulnerabilities from OSV.dev.