precedent

Federal award history and pass-through finder. Find out who actually wins a federal program before you advise a client to apply for it, and find the state agencies, universities, and large nonprofits that pass federal money down to organizations too small to win it directly.

---

The problem

You are sitting with a client. They have a $600,000 operating budget, three program staff, and no federal award history. Someone forwards them a Notice of Funding Opportunity that looks like a perfect programmatic fit, and they want to know whether to spend six weeks writing to it.

The thing you need in order to answer that question is not in the NOFO. You need to know who has won this program before. If the last five years of awards under that Assistance Listing went to state health departments and university medical centers at a median of $4 million each, your client is not a marginal candidate. They are not a candidate. If the median award is $180,000, four in ten winners each year are organizations that had never won the program before, and the awardee list is full of community-based groups in eleven states, that is a completely different conversation.

Today that answer is either invisible at the moment of decision or it costs $179 to $899 a month to look up. So most of the time the question does not get asked, and consultants advise clients into pursuits they realistically cannot win.

There is a second, larger problem hiding behind the first.

Most nonprofit leaders think of federal funding as something you apply for on Grants.gov. For small organizations that is usually wrong. A very large share of federal money reaches community nonprofits as a subaward: the federal agency awards to a state agency, a county, a university, or a large intermediary nonprofit, and that entity passes the money down under a subrecipient agreement. Head Start, victim services, aging nutrition programs, workforce development, community development block grants, and most block grant funding in general move this way.

If you are a $600,000 organization, the pass-through layer is where your actual odds live. And essentially no commercial grant research product surfaces it, because the data lives somewhere nobody in this market looks: the Schedule of Expenditures of Federal Awards inside single audit filings at the Federal Audit Clearinghouse.

precedent does both halves. It reads the direct award history from USAspending, and it reads the pass-through layer from single audits, and it tells you what it does not know.

---

Quickstart

Sixty seconds, no account, no API key, no database.

The package is federal-precedent on PyPI and the command it installs is precedent. They differ because precedent on PyPI is a name reservation for an unrelated architecture-decision-record project. uvx federal-precedent ... below runs without installing anything; if you install it properly, the command is just precedent.

# Award history for a program you already have in mind
uvx federal-precedent history 93.243

# Or find the program first
uvx federal-precedent programs "opioid"

# Machine-readable
uvx federal-precedent history 93.243 --json

The history command needs no credentials at all. The passthrough command needs a free Federal Audit Clearinghouse API key, which takes about two minutes to get by email. See .env.example.

export FAC_API_KEY="..."          # free, from https://www.fac.gov/api/signup/
uvx federal-precedent passthrough --state OH --program 93.045

An MCP server for Claude and other MCP clients is next on the roadmap and is not in this release. precedent --help lists what ships today.

---

Credits

precedent is a thin layer over public infrastructure that other people built and maintain. Naming them is not a courtesy, it is the accurate description of where the value comes from.

• USAspending and the usaspending-api team at the Bureau of the Fiscal Service and the Federal Spending Transparency community. The API is open, well documented, free, and requires no key. The request contracts in that repo are the reason this tool could be built correctly rather than by guesswork.
• The Federal Audit Clearinghouse team at GSA Technology Transformation Services, who moved the FAC from Census to GSA, put a real PostgREST API in front of single audit data, published a field dictionary and a legacy-field crosswalk, and give the key away for free. The pass-through half of this tool exists because they did that work.
• The Nonprofit Open Data Collective and GivingTuesday, whose Form 990 tooling and Master Concordance File underpin the sibling funder-graph repo that precedent cross-references by EIN.
• ProPublica Nonprofit Explorer for organization lookups used in gap-filling.

Full attribution and licenses are in NOTICE. Our contribution posture, and the issues we intend to file upstream rather than patch locally, are in docs/research/prior-art.md.

---

Worked example 1: award history

Assistance Listing 93.243, Substance Abuse and Mental Health Services Projects of Regional and National Significance. Awarding agency: Department of Health and Human Services, Substance Abuse and Mental Health Services Administration.

$ uvx federal-precedent history 93.243 --since FY2020 --until FY2024

93.243  Substance Abuse and Mental Health Services Projects of Regional and
        National Significance
        Department of Health and Human Services / SAMHSA
        Assistance awards with a base obligation date in FY2020-FY2024

  1,058 awards to 726 distinct recipients          $3,570,889,654 obligated

  NEW-ENTRANT RATE                                              39.1%
  284 of 726 recipients in FY2020-FY2024 had won no award under this
  program in the FY2015-FY2019 lookback window.

  Award size
    minimum                                                      $397
    25th percentile                                          $186,013
    MEDIAN                                                   $305,161
    75th percentile                                          $854,761
    90th percentile                                        $2,615,811
    maximum                                              $205,604,128
    mean                                                   $3,375,132   (skewed)

  Distribution
    under $100k        153  ###############                      14.5%
    $100k - $250k      287  ############################         27.1%
    $250k - $500k      221  #####################                20.9%
    $500k - $1M        164  ################                     15.5%
    $1M - $5M          162  ################                     15.3%
    $5M and over        71  #######                               6.7%

  Repeat winners                                                 28.7%
  208 of 726 recipients took 2 or more awards inside the window.
  Top 10 recipients by dollars hold 48.1% of all dollars obligated.

  Geography          57 place-of-performance state and territory codes
    CA 85    AK 79    OK 67    NY 58    WI 40    MI 39    MT 31    SD 30

  Most frequent recipients
    Great Plains Tribal Leaders Health Board          7 awards    $7,634,611
    The Mental Health Association of New York City    6 awards  $322,895,770
    New Jersey Department of Human Services           5 awards  $117,821,539
    MaineHealth                                       5 awards    $1,817,853

  Caveats for this program
    256 of 1,058 awards (24.2%) report more than one Assistance Listing, so
    the dollar figures include money that is not 93.243. Treat award-size
    percentiles for this program as an upper bound.

  Source: USAspending API, POST /api/v2/search/spending_by_award, retrieved
  2026-08-30. Award amounts are total obligations across the life of each
  award, not fiscal-year outlays.

  This is informational only, derived from public data on the dates shown. It
  is not an eligibility determination, and not legal, tax, or accounting
  advice. Verify against the official source before relying on it.

Those figures are real. They were computed against the live USAspending API on 2026-08-30 and the command above reproduces them, subject to the small drift that comes from USAspending restating prior-year records.

What a consultant does with this. A median of $305,161 with 41.6% of awards under $250,000 means a mid-sized community organization is inside the size band. A new-entrant rate of 39.1% means this program genuinely takes newcomers, which is the opposite of what the presence of $205 million awards at the top of the range would suggest on its own. The new-entrant rate is the number that changes the advice, which is why it is the headline and not a footnote.

---

Worked example 2: pass-through finder

State: Ohio. Assistance Listing 93.045, Special Programs for the Aging, Title III, Part C, Nutrition Services. This is a program almost no community organization receives directly. Federal money goes from the Administration for Community Living to the state unit on aging, which subawards to regional area agencies on aging, which subaward again to local senior centers and meal providers.

$ export FAC_API_KEY="..."
$ uvx federal-precedent passthrough --state OH --program 93.045

The output below is illustrative. The structure it shows is real and verifiable: Ohio's federal aging nutrition money does move through the Ohio Department of Aging to regional area agencies on aging, and the organizations named are real Ohio area agencies on aging. The dollar figures and counts are shaped from typical Schedule of Expenditures of Federal Awards magnitudes rather than pulled live, because the pass-through query needs an API key. Run the command to get current figures.

93.045  Special Programs for the Aging, Title III, Part C, Nutrition Services
        Pass-through entities reaching organizations in OHIO
        Audit years 2019-2024, 1,214 Ohio single audits scanned

  WHO PASSES THIS MONEY DOWN
  Ranked by the number of distinct Ohio organizations that named this entity
  as their pass-through entity in their own single audit.

  1. Ohio Department of Aging                            state agency
     named by 11 Ohio subrecipients      $41.2M expended through it
     also passes: 93.044, 93.052, 93.053, 10.561
     -> this is the state unit on aging; it is the top of the Ohio chain

  2. Council on Aging of Southwestern Ohio               nonprofit, Blue Ash OH
     named by 9 Ohio subrecipients        $8.7M expended through it
     EIN 31-0896213   passes down 93.045, 93.044, 93.052
     -> funders.opengrants.io/funders/31-0896213

  3. Western Reserve Area Agency on Aging                nonprofit, Cleveland OH
     named by 7 Ohio subrecipients        $6.1M expended through it

  4. Area Office on Aging of Northwestern Ohio, Inc.     nonprofit, Toledo OH
     named by 5 Ohio subrecipients        $3.9M expended through it

  5. Central Ohio Area Agency on Aging                   public, Columbus OH
     named by 4 Ohio subrecipients        $3.4M expended through it

  OHIO ENTITIES REPORTING THEY PASSED 93.045 MONEY DOWN
  From SEFA lines where the auditee marked the award as passed through to
  subrecipients, ranked by reported pass-through amount.

    Ohio Department of Aging                             $39,800,000
    Council on Aging of Southwestern Ohio                 $7,900,000
    Direction Home Akron Canton Area Agency on Aging      $4,600,000
    Ohio District 5 Area Agency on Aging                  $2,100,000

  COVERAGE WARNING  read this before you use the list above
    Single audits are only filed by organizations that expend at or above the
    federal single audit threshold in a fiscal year. That threshold was
    $750,000 and rose to $1,000,000 for fiscal years beginning on or after
    2024-10-01. Organizations below the threshold file nothing, so they are
    absent from this data entirely, not counted as zero.

    That means this list is skewed toward larger recipients and larger
    intermediaries, and the subrecipient counts are a floor, not a total. An
    intermediary "named by 4 subrecipients" is named by 4 subrecipients that
    were themselves large enough to file. It may have fifty more.

    Pass-through entity names in this data are free text typed by the
    auditee. precedent normalizes and clusters them; run with
    --show-name-variants to see exactly which raw strings were merged.

    Amounts are federal expenditures in the audited fiscal year, not award
    amounts and not the year the award was made.

  Source: Federal Audit Clearinghouse API, general / federal_awards /
  passthrough endpoints, retrieved 2026-08-30. FAC production data refreshes
  weekly, typically Wednesdays, and covers audit years 2016 forward.

  This is informational only, derived from public data on the dates shown. It
  is not an eligibility determination, and not legal, tax, or accounting
  advice. Verify against the official source before relying on it.

What a consultant does with this. The client was never going to win 93.045 from the Administration for Community Living. The five organizations at the top of that list are the actual funders in their world, they are reachable by phone, they have subrecipient procurement processes, and no subscription product this client's consultant is paying for lists them.

---

What it computes, and exactly how

Precise definitions matter more than the numbers themselves, because two tools can produce different medians from the same data and both be defensible.

Output	Definition
Award universe	Assistance awards (award type codes 02, 03, 04, 05) reported under the Assistance Listing, assigned to a fiscal year by base obligation date, deduplicated by USAspending's award identifier
Award amount	Total obligated across the life of the award, not fiscal-year outlays
Median	50th percentile of award amounts, over awards rather than recipients
Percentiles	p10/p25/p50/p75/p90 by inclusive linear interpolation
Distribution buckets	Fixed half-open bands so programs are comparable: under $100k, $100k-$250k, $250k-$500k, $500k-$1M, $1M-$5M, $5M and over
Recipient identity	Unique Entity Identifier where present, then USAspending's recipient hash, then normalized legal name
Repeat winner	A recipient with 2 or more distinct awards inside the window
New-entrant rate	Share of distinct recipients in the window that had no award under the same Assistance Listing during the lookback window (default: the 5 fiscal years immediately before)
Geographic spread	Count of distinct place-of-performance state and territory codes, plus leaders by count and by dollars
Concentration	Share of window dollars held by the top 10 recipients
Pass-through, demand side	Distinct organizations in the state that named an entity as their pass-through entity in their own single audit
Pass-through, supply side	Reported pass-through amounts from SEFA lines the auditee marked as passed down to subrecipients

Full statistical specification, including tie-breaking and null handling, is in prompts/01-build-core.md.

---

Coverage and limitations

Read this section before you quote a number from this tool to a client.

The single audit threshold makes pass-through coverage inherently partial. A non-federal entity must have a single audit only if it expends at or above the threshold in federal awards in a fiscal year. That threshold was $750,000 for many years and rose to $1,000,000 for fiscal years beginning on or after 2024-10-01 under the 2024 revision to the Uniform Guidance. An organization spending $400,000 in federal pass-through money files nothing and appears nowhere in this data. Every subrecipient count precedent reports is a floor. Every intermediary ranking is a ranking among intermediaries visible to organizations large enough to file. Treating these lists as complete is the single most likely way to be wrong with this tool, and it is why the coverage warning is printed in the output and not only here.

Schedule of Expenditures of Federal Awards reports expenditures, not awards. A SEFA line says what an organization spent under a program during an audited fiscal year. It does not say what they were awarded, when they were awarded it, or how long the award runs. Multi-year awards appear across several audits at partial amounts.

Pass-through entity names are free text. The auditee types them. "Ohio Dept. of Aging", "OHIO DEPARTMENT OF AGING", and "State of Ohio, Department of Aging" are three strings for one agency. precedent normalizes and clusters, publishes the alias table it uses, and shows you the raw variants on request. It will still occasionally split one entity or merge two, and it never merges across states.

Awards frequently report more than one Assistance Listing. In the 93.243 example above, 24.2% of awards list a second program number, and USAspending reports one obligated total per award rather than a split by program. Award-size statistics for programs with heavy multi-listing are therefore an upper bound. precedent reports the multi-listing share on every profile so you can judge it.

FAC API coverage starts at audit year 2016. Earlier single audits live in the legacy Census extracts and are not queried by this tool.

USAspending search reaches back to 2007-10-01. A lookback window that crosses that boundary is truncated, which inflates the new-entrant rate. precedent flags this on the output when it happens.

Federal spending data is restated. Agencies correct prior submissions. Numbers pulled today can differ slightly from numbers pulled last month. Every output carries a retrieval date for this reason.

For-profit subrecipients are generally outside single audit requirements, so commercial intermediaries are underrepresented in the pass-through view relative to governments and nonprofits.

This tool makes no eligibility determination. It reports what other organizations received. It does not know your client's registration status, their indirect cost rate, their audit findings, or whether the program is open.

---

Data sources

Source	What it provides	Auth	Refresh cadence
USAspending API	Direct federal assistance awards: recipient, Unique Entity Identifier, amounts, dates, Assistance Listing numbers, place of performance	None	Daily load from agency submissions; prior periods restated
Federal Audit Clearinghouse API	Single audit submissions and Schedule of Expenditures of Federal Awards, including which entity passed money to whom	Free API key by email	Production endpoint updates weekly, typically Wednesdays. Covers audit years 2016 forward
OpenGrants API	Currently open funding opportunities, optional enrichment only	Optional API key	Daily

Detailed endpoint documentation, request shapes, field semantics, pagination behavior, and observed rate limits are in docs/research/data-sources.md.

Terms of use worth reading before you redistribute anything derived from these: FAC API terms.

---

Optional: pair history with what is open right now

precedent works completely without any OpenGrants credentials. If you set OPENGRANTS_API_KEY, every program profile gains a section showing the currently open opportunities under that Assistance Listing, drawn live from the OpenGrants /grants-api index, so you get the historical awardee profile and the live application deadline in one place. Enriched lines are marked — live from OpenGrants so you always know which facts came from public bulk sources and which came from the API. If the key is missing, expired, or the network is down, the command still returns its full public-data result.

---

MCP server — planned, not in this release

The intent is to expose the same capabilities as Model Context Protocol tools for agent use: award_history, passthrough_finder, recipient_profile, find_program. The core logic already lives in the library precisely so the CLI and an MCP server can both be thin adapters over it, but the server itself is not written yet. It is not installed by this release and precedent mcp does not exist. When it lands it will be documented here with a version number.

---

Disclosure

This is informational only, derived from public data on the dates shown. It is not an eligibility determination, and not legal, tax, or accounting advice. Verify against the official source before relying on it.

---

Contributing

See CONTRIBUTING.md. Fixture-based tests against real committed samples of upstream responses are required for any change touching a data source, because mocked-shape tests do not catch schema drift and schema drift is the failure mode that actually matters here.

License

Apache License 2.0. See LICENSE and NOTICE.

---

Built and maintained by Egeria Corporation, sponsored by OpenGrants.