Data

Every attribute carries the law that created it. Not a comment, not a wiki page — a citation on the attribute itself, generated from the compiled model and impossible to drift away from it.

The shape of the model

Rules, and the shapes that facts have to take.

3,039
Attributes
1,536
Concluded by rules
1,504
Declared inputs
51
Rule documents
0
Anonymous attributes
0
Double-concluded

Nothing in the model is nameless, and nothing has two rules fighting to conclude it. Both are enforced by the gate rather than by review discipline — an undeclared attribute is a build failure, not a warning.

The generated dictionary

Produced by one command, so it cannot drift from the model it describes.

The model ships a generated data dictionary: every attribute with its text, its type, the rule document that owns it, its legislative citation, its linkage to the recorded readings, and the cross-program consumption map. It is generated from the compiled data model, never hand-written — which is exactly why it can be trusted to be current, and exactly why it is able to find our own weak spots.

The coverage numbers, published whichever way they fall

At the generated run that measured it — census 3,040 attributes at that moment — 2,499 carried a legislative citation. Of the attributes concluded by a rule, 1,368 of 1,536 carried one: close to nine in ten. Of those citations, 2,443 of 2,499 resolve to a cached, hashed source we hold.

168 concluded attributes carry no citation. Most are assembling a provision rather than stating one — consolidation, model setup, intermediate results — and the run classifies each. 25 remain unclassified and worth a look, and are named as such.

The tool that would let an auditor find the soft spots is the tool that found them. We would rather publish that number than be asked for it.

A sample, not the model

Forty attributes, across five program families, each with the provision behind it.

Below is a curated sample chosen to show the range — income standards, age tests, deductions, allowances, exclusions, look-back periods. It is a window, not the model: the full attribute set is the substance of the work and is shared under engagement, not published. What the sample demonstrates is the property that matters: you can point at any row and read the law that put it there.

Medicaid and CHIP

AttributeTypeProvision
the member's current monthly MAGI-based incomecurrency42 CFR 435.603(e)
the income is excluded under the American Indian or Alaska Native exceptionsboolean42 CFR 435.603(e)(3)(i)
the member is under the age specified in 42 CFR 435.603(f)(3)(iv)boolean42 CFR 435.603(f)(3)(iv)
the member meets the income standard 42 CFR 435.116(b) requiresboolean42 CFR 435.116(b)
the member is an infant under age 1 as 42 CFR 435.118 identifies that age groupboolean42 CFR 435.118
the member meets the income standard 42 CFR 435.118(b) requires for children age 6 through age 18boolean42 CFR 435.118(b)
the member is age 19 or older and under age 65 as 42 CFR 435.119(b)(1) requiresboolean42 CFR 435.119(b)(1)
the member is under age 26 as 42 CFR 435.150(b)(1) requiresboolean42 CFR 435.150(b)(1)

The Exchange, premium tax credits and the Basic Health Program

AttributeTypeProvision
the member is within the household income bounds 26 CFR 1.36B-2(b)(1) statesboolean26 CFR 1.36B-2(b)(1)
the member and the member's spouse file a joint return for the taxable yearboolean26 CFR 1.36B-2(b)(2)(ii)
the member is granted Temporary Protected Status in accordance with 8 USC 1254aboolean8 USC 1254a
the member resides in the service area of the Exchange as 45 CFR 155.305(a)(3)(ii)(A) describesboolean45 CFR 155.305(a)(3)(ii)(A)
the member is eligible to enroll in a standard health plan under 42 CFR 600.305boolean42 CFR 600.305
the member falls within the principal coverage band the State's basic health program setsboolean42 CFR 600.305(a)(2)

SNAP

AttributeTypeProvision
the percentage of the Federal income poverty level that is the SNAP net income eligibility standardnumber7 CFR 273.9(a)(1)(i)
the member is an elderly or disabled member under 7 USC 2012(j)boolean7 USC 2012(j)
the member is over the age of 15 and under the age of 60 as 7 USC 2015(d)(1)(A) statesboolean7 USC 2015(d)(1)(A)
the member is a person 7 USC 2015(o)(3)(A) excepts as under 18 or over 65 years of ageboolean7 USC 2015(o)(3)(A)
the earned income deduction 7 USC 2014(e)(2)(B) allows the SNAP household of the membercurrency7 USC 2014(e)(2)(B)
the dependent care deduction 7 USC 2014(e)(3) allows the SNAP household of the membercurrency7 USC 2014(e)(3)
the member's vehicle is excluded from the member's SNAP financial resources under 7 CFR 273.8(e)(3)boolean7 CFR 273.8(e)(3)
the maximum monthly SNAP allotment for the size of the member's SNAP householdcurrency7 CFR 273.10(e)(4)(ii)

WIC, school meals and Summer EBT

AttributeTypeProvision
the member is a pregnant woman as 42 USC 1786(b)(11) and 7 CFR 246.2 define that termboolean42 USC 1786(b)(11)
the member presents a dietary deficiency that impairs or endangers healthboolean7 CFR 246.7(e)(2)
the member is within the 90 days before the member's certification period expiresboolean7 CFR 246.7(h)(1)(i)
the member is a child who is income eligible for a reduced price meal under 42 USC 1758(b)(9)(B)(i)boolean42 USC 1758(b)(9)(B)(i)
the school the member attends offers meals at no charge to all enrolled children under 7 CFR 245.9boolean7 CFR 245.9
the member is a child who is receiving assistance under a program 7 CFR 292.2 namesboolean7 CFR 292.2

Medicare Savings Programs

AttributeTypeProvision
the specified low-income medicare beneficiary percentage of the official poverty linenumber42 USC 1396d(p)(2)(A)
the member's monthly qualified medicare beneficiary income standardcurrency42 USC 1396d(p)(2)(A)
the member's earned income qualifies for the blindness exclusion 42 USC 1382a(b)(4)(A) statesboolean42 USC 1382a(b)(4)(A)
the member meets the Medicare entitlement condition 42 USC 1396d(p)(1)(A) statesboolean42 USC 1396d(p)(1)(A)

Non-MAGI Medicaid — aged, blind and disabled pathways, spend-down, transfers and long-term care

AttributeTypeProvision
the member is aged for the purposes of 42 USC 1382c(a)(1)(A)boolean42 USC 1382c(a)(1)(A)
the member is categorically needy under 42 CFR 435.631(b)(1)boolean42 CFR 435.631(b)(1)
the medically needy income standard that applies to the member for the budget periodcurrency42 CFR 435.831(c)
the member disposed of those assets for less than their fair market valueboolean42 USC 1396p(c)(1)(A)
the disposal 42 USC 1396p(d)(3) deems the member to have made falls within the look-back periodboolean42 USC 1396p(d)(3)
the community spouse resource allowance 42 USC 1396r-5(f)(2) defines for the membercurrency42 USC 1396r-5(f)(2)
the minimum monthly personal needs allowance 42 USC 1396a(q)(2) sets for the membercurrency42 USC 1396a(q)(2)
300 percent of the supplemental security income benefit rate 42 USC 1396b(f)(4)(C) namescurrency42 USC 1396b(f)(4)(C)

Integration, measured

Proof that the programs genuinely share a model rather than merely sit next to each other.

Co-location is easy to claim. The generated consumption map counts the places where one family's conclusion is actually read by another family's rules: 174 measured edges across 28 program pairs, plus 20 shared declared inputs. Those edges are what a mixed household actually needs, and they are counted rather than asserted.

What one of those seams is worth

A 2025 food-assistance amendment moved a Medicaid figure, with no Medicaid provision amended anywhere. A long-term-care allowance imports a food-assistance standard by statutory cross-reference, and the amendment narrowed that standard. Two otherwise identical households differ only in that figure — and the protected income of the spouse in the nursing home moves by nearly two hundred dollars a month.

If those rules live in different systems maintained by different teams, that change is invisible until somebody appeals.