Where the risk is.
Instead of making you read tables, the platform reads them for you. Every real award is scored with transparent, printed factors — structural exposure, program fraud history, size, entity signals — and sorted so the highest-risk money surfaces first. A score measures where to look, never guilt.
Risk register — every award, scored & ranked
| Risk | Recipient | Program | Obligated | Why this score |
|---|
The scoring method — printed, not hidden
How to read this honestly
High scores cluster on state-agency pass-throughs — not because agencies are suspect, but because this open channel can't see below them, and history says that's where schemes live. The score measures structural blindness + documented program risk + exposure size. It's the same logic auditors use to plan engagements, applied continuously and in public. Named entities are real records with links; no score implies wrongdoing.
The real audit, running now.
Federal awards are the one fully open, keyless data channel today — and the channel Feeding Our Future exploited. Every row is an actual USASpending.gov record with a link to the official filing. REAL SNAPSHOT
Largest federal grants — FY2026 window
| # | Recipient | Awarding Sub-Agency | Program | Risk | Obligated |
|---|
Child nutrition money, watched.
The $250M fraud ran through CACFP (10.558) and the Summer Food Service Program (10.559), passed through Minnesota. Real obligations recorded in USASpending: REAL SNAPSHOT
Screens on real data.
Deterministic screens computed on the real pulls. A screen hit is a reason to look, not an accusation. Rules are printed beside results. REAL SNAPSHOT
Screen S1 · Large FY26 grants to LLC-type recipients
proxy for the "new-entity vendor" rule until the SOS formation-date join lands| Recipient | Program | Sub-Agency | Start | Obligated |
|---|
Screen S2 · Concentration in child-nutrition pass-throughs
computed from the Nutrition Money pullSources & Methods
What's real, what's proxy, what's locked. An audit platform that isn't transparent about its own limits isn't one.
| Layer | Source | State | Detail |
|---|---|---|---|
| Federal awards to MN | api.usaspending.gov | REAL — snapshot + live | Verified snapshot pulled 2026-07-08 (16:33 CT) via the API from usaspending.gov's own origin; the page re-fetches live on load and replaces the snapshot when the browser allows. |
| Risk scores | Computed in this page | Transparent model | Additive factor model printed on the Risk Board. Measures structural exposure + documented program risk + size + entity signals. Not a fraud determination. |
| Why live fetch can fail | Browser CORS/preflight | Explained | Opening the file directly (file://) makes some browsers block cross-origin POSTs. Snapshot keeps the page truthful; serving it (python3 -m http.server) or Chrome usually goes fully live. |
| Single audits (FAC) | api.fac.gov | Needs free key | Powers the missing-audit screen (R5). Key from api.data.gov — 2 minutes. |
| Subawards (the fraud layer) | /api/v2/subawards/ | Next build | Prime awards show the state agency; sponsors like Feeding Our Future live in subaward records. The single highest-value addition. |
| MN state checkbook (SWIFT) | TransparencyMN / MAT viewer | No API — the gap | Updated nightly, locked in a viewer. This gap is the core of the upgrade proposal. |
| Entity formation dates | MN SOS bulk file | File order | Free for researchers/non-commercial; turns S1's name heuristic into the real formation-date rule. |
| Local gov finances | OSA CTAS/SAFES | Inside OSA today | Collected under Minn. Stat. § 6.74/6.745 — publishing is a decision, not a build. |
Methods, plainly
Queries: state recipient profile (GET); FY26 grants to MN sorted by obligation (POST, top 100); nutrition programs 10.558/10.559 by fiscal year and by prime recipient (POST). Screens and risk scores are deterministic rules computed on those responses, printed beside results. Named entities appear exactly as published on USASpending.gov with links to official records; a hit or high score means "look here," never "guilty."
Known limits
Prime awards ≠ subawards — the pass-through blindspot is displayed, not hidden. Obligations ≠ outlays. Recent fiscal years lag. "LLC in name" is a weak proxy pending the SOS join. If live refresh fails, the page says so and shows the timestamped snapshot — this platform never fakes data.
What OSA 2.0 adds.
The Live Audit section works with zero cooperation from the state. This is what becomes possible with the auditor's statutory data — SWIFT, CTAS, SOS, subawards — running the same engine.
R1 · New-Entity Vendor
Vendor LLC registered days before a large public payment — run on state and local payments, not just federal.
R2 · Duplicate Payment
Same vendor, same amount, short window — the highest-recovery rule in any payment system.
R3 · Threshold Splitting
Purchases sliced to duck competitive-bid limits.
R4 · Shared-Agent Network
Vendors sharing registered agents winning from the same entity.
R5 · Missing Federal Audit
$750k+ federal recipient, no FAC single audit on file.
R6 · Subaward Sweep
The Feeding-Our-Future killer: every sub-recipient joined to SOS formation dates and meal-count reasonableness.
Prototype: the statewide live ledger
SAMPLE ROWS — illustrative only| Payee | Entity | Fund | Status | Amount |
|---|---|---|---|---|
| Anderson Bridge & Iron Co. sample | MnDOT — District 3 | Trunk Highway | Cleared | $1,284,300 |
| Northstar Catering LLC sample | Dept. of Education | Fed. Child Nutrition | R1 — entity 6 days old | $412,900 |
| Gopher Office Supply Inc. sample | City of St. Cloud | General Fund | R2 — duplicate | $233,150 |
| Twin Pines Consulting sample | Hennepin County | Human Services | R3 — 4× just under $25k | $99,400 |
Why the upgrade matters
The live section sees federal primes only. Minnesota's own ~$60B local layer, the SWIFT checkbook, and the subaward layer where the last fraud actually lived are where the auditor already has legal reach. Same engine, 50× the coverage.
Deployment path
Day 1: publish CTAS/SAFES data + modern § 609.456 intake. Day 30: ledger beta (counties + first-class cities). Day 60: engine on 100% of in-scope transactions. Day 90: first Flag Report + Recovery Scoreboard. Statutory mapping on the Legal Playbook page.
The Legal Playbook
What the State Auditor can do under current law, starting day one — real citations linked to the official Revisor of Statutes — and the honest short list needing the legislature. Verified against the 2025 Minnesota Statutes, July 8, 2026.
Roughly 80% of the upgrade requires zero new legislation. The power is already on the books — the will and the software are what's missing.
01 · Get the data
Examine political subdivision records — incl. by citizen petition
Core examination authority over subdivisions' books and affairs; local voters can petition the auditor in directly.
Data already flows to the auditor by law
Financial and summary budget data from every local government is already collected. Publishing it as open machine-readable data is a decision, not a statute.
Testimonial powers; refusal is punishable
Compel testimony in examinations; § 6.53 penalizes refusing to assist. Data requests from this office are not optional.
Fraud reporting TO the auditor is mandatory
Local public employees discovering theft or unlawful use of funds must promptly report in writing to the State Auditor — including not-public data. The pipe exists in law; it needs a modern intake.
02 · Follow the money — into private hands
Examine grantees & contractors of political subdivisions
The sleeper power. Authority to examine private organizations receiving subdivision money, to the extent needed to account for those funds. Public money doesn't turn invisible inside an LLC.
Audit of federal money
Express authority over federal funds in political subdivisions — the exact channel the $250M scheme exploited, and the exact data the Live Audit pages watch.
Full jurisdiction: 4,300+ entities, ~$60B
Counties, first-class cities, other subdivisions, special districts — the whole local footprint is already inside the perimeter.
Hospitals, nursing homes, fire relief associations
Explicit reach into high-dollar, historically under-watched corners.
03 · Set the standards
Prescribe accounting systems & forms
Drive local reporting toward a uniform, machine-readable chart of accounts — the one change that makes statewide detection trivial.
Auditor already ships accounting software (CTAS)
Statutory precedent for giving every local government the free detection stack.
Prescribe minimum procedures for private CPA audits
The auditor sets minimum procedures for the CPA firms doing most local audits; § 6.67 obliges misconduct reporting. Add analytics steps → every private audit becomes a sensor.
Best-practices reviews + performance measurement
Hooks for cost-per-outcome dashboards and the open 50-state playbook.
04 · What genuinely needs the legislature
State agency jurisdiction
State agencies belong to the Legislative Auditor (Minn. Stat. § 3.971 et seq.). Real-time SWIFT screening needs an MMB/OLA data-sharing MOU now, statute later.
Monthly machine-readable reporting mandate
§ 6.47 gets most of the way; hard statutory deadlines make it successor-proof.
Platform funding + recoveries mechanism
Cheap by government standards; recoveries-fund-the-office makes it self-sustaining.
Whistleblower rewards
§ 609.456 mandates reporting but pays nothing. A False-Claims-style bounty would supercharge intake.