Federal Limits

Methodology

How we measure state legislative protection against federal civil immigration enforcement reach across all 52 jurisdictions. Every grade is built on 9 standards, scored from binding state law.

9 STANDARDS × 0/1/2 = 18 MAX
BINDING STATE LAW ONLY
SOURCE-WEIGHTED CITATIONS
v3.5 -- AMENDED 2026-08-12
FUNCTIONAL PROTECTION > STRUCTURAL TEMPLATE

The 9 Standards

Every state earns 0, 1, or 2 points on each of nine standards. The score reflects what the state's law actually does -- not what an executive order promises, not what an agency practices, but the statutory architecture in force. Standards flow in sequence from foundational (operational separation) to specific (federal agent identification). Click any standard to see the criteria.

STD-01
Personnel & Resource Limits
State officers, money, and facilities withdrawn from federal civil enforcement.

Whether the state has prohibited its officers, employees, money, equipment, facilities, and communications from being used in federal civil immigration enforcement operations. Federal civil enforcement relies on state and local infrastructure to operate; state legislation can sever that operational dependency.

0
No statutory protections.
1
Partial: one dimension (personnel or resources) covered, or both with major carveouts.
2
Both personnel and resources covered statewide, binding state and local agencies, with limited or no carveouts. Anchor states: CA, IL, CO, WA.
STD-02
Cooperation Contract Prohibitions
287(g), IGSAs, and joint task force agreements banned.

Whether the state has banned 287(g) agreements — the federal program that deputizes local police to perform federal civil immigration enforcement. Per the Jul 14 2026 roster, 2,127 active 287(g) agreements (MOAs) are in force — up from about 135 in January 2025, roughly a 15-fold increase in 18 months. State statute is the durable mechanism that can constrain this expansion. Supplementary contract categories (IGSAs, joint task force agreements, federally conditioned grants) deepen the protection and appear in per-state narrative.

0
No statutory or judicial prohibition on 287(g) or the other contract categories; or the state affirmatively mandates cooperation (e.g., TX SB 4). EO- or AG-directive-only bans score 0 — they fail the durability test.
1
287(g) restricted but not fully banned (state-agency-only, jail-model-only, or grandfathered existing agreements), or another contract category banned while 287(g) is not.
2
287(g) agreements explicitly prohibited by state statute (or binding state supreme court holding). The 287(g) ban is the lynchpin and alone carries the 2-point bar; other contract prohibitions are supplementary. Anchor states: CA, IL, MD, WA, OR, NM, NJ.
STD-03
Sensitive Location Protections
Schools, hospitals, courthouses, churches off-limits to federal civil enforcement.

Whether the state has legislatively protected specific physical places from federal civil enforcement access: courthouses, schools, hospitals, childcare facilities, religious institutions, and workplaces. Federal sensitive-location protections were rescinded January 20, 2025; state legislation is the only mechanism that restores these protections within state borders.

0
No statutory location protections.
1
Partial: 1–3 location types covered, or coverage with major carveouts.
2
Four or more of six location types covered with judicial-warrant access standard. Anchor states: CA, CO, CT, IL, MD, ME.
STD-04
Private Detention Industry Restrictions
Private contractors prohibited from operating immigration detention.

Whether the state has prohibited private contractors from operating immigration detention facilities within state borders. By early 2026, the federal government had structured up to $65 billion in detention-construction capacity through a U.S. Navy procurement vehicle (WEXMAC), with more than 130 contractors able to receive work without site-by-site competitive bidding. State law is the only mechanism that can prevent private operators from building immigration detention in a state.

0
No prohibition, or prohibition fully struck down or fully enjoined.
1
Partial: prohibitions limited to new contracts, sunset provisions, scope carveouts, or statute partially enjoined.
2
Comprehensive prohibition, no sunset clause, narrow or no carveouts, statute in force. Anchor states: IL, OR.
STD-05
Information Firewalls
State data systems firewalled from federal civil enforcement requests.

Whether the state has firewalled its data systems from federal civil enforcement across five domains: voter rolls, DMV / driver records, school records, health records, and professional licensing. Since January 2025, federal demand letters for unredacted voter rolls have gone to 48 states. States with statutory firewalls refused and have been winning in federal court; states without them complied. This standard isn't theoretical — it is the active legal terrain.

0
No statutory data firewalls.
1
1–2 data domains firewalled with hard-denial language.
2
Three or more of five data domains firewalled with hard-denial statutory language. Anchor states: CA (5 domains), IL (5), CO (4), WA, NY, NJ, CT, MD, OR.
STD-06
Warrant Requirement
Judicial warrant required before any detention, hold, or transfer.

Whether the state requires a judicial warrant — signed by a federal judge (Article III or magistrate) — before honoring federal civil enforcement detention or transfer requests. Federal civil enforcement relies on administrative warrants (Forms I-200, I-205) and detainers (Form I-247) signed by federal officers, not judges; the judicial warrant requirement is the structural check on that administrative pipeline. Rights notifications during encounters appear in per-state narrative and do not gate the score.

0
State honors administrative detainers without a judicial warrant requirement; or the requirement exists only as voluntary policy or AG directive without statutory backing.
1
Judicial warrant required but with major carveouts: some cooperation categories exempted, limited contexts (e.g., courthouses only), or coverage of state agencies but not local law enforcement.
2
Judicial warrant required statewide, in statute or binding state supreme court holding (e.g., Lunn v. Commonwealth in MA), with no administrative-detainer exception. The warrant requirement alone carries the 2-point bar. Most widely adopted standard — eleven jurisdictions meet it: CA, CO, CT, DC, IL, MA, ME, NJ, NY, OR, WA.
STD-07
Documentation Transparency
Public reporting of cooperation activity by all state and local agencies.

Whether the state requires public reporting of cooperation activity by state and local agencies, on a regular schedule, with sufficient scope to support oversight. In April 2026 the Government Accountability Office found that the Department of Homeland Security cannot demonstrate its oversight programs work; state-level reporting creates the documentary record that enables both citizen oversight and state-level enforcement.

0
No reporting requirement.
1
Partial: ad-hoc reporting with no schedule, some agencies only, narrow scope, or no public posting.
2
Mandatory reporting on regular schedule (annual minimum, quarterly preferred), all state and local agencies, full scope (communications, custody transfers, federal access requests, contracts), publicly posted. Anchor states: CO (quarterly), CA (annual SB 54 Values Act report).
STD-08
Enforcement & Remedies
Private right of action + statutory AG enforcement authority.

Whether the state has created legal mechanisms that make cooperation laws enforceable — either a private right of action allowing residents to sue, or statutory authority for the state Attorney General to investigate and enforce cooperation laws against non-compliant local officials. Laws on the books mean little without mechanisms to enforce them.

0
Neither mechanism in statute. General civil-rights remedies that do not specifically reach cooperation-law violations do not count.
1
A mechanism exists but is limited: narrow standing, declaratory-only relief, coverage of only some cooperation laws, or AG authority that stops at state agencies and does not reach local law enforcement.
2
Either mechanism, comprehensively implemented: a private right of action with damages or injunctive relief and no major standing carveouts, OR statutory AG enforcement authority (investigation + audit + corrective orders) reaching local law enforcement. Both together is the gold standard, noted in per-state narrative. Anchor states: CA, CO, CT, IL, NJ, NM, NY, OR, WA.
STD-09
Federal Agent Identification & Recording
Visible ID + mask prohibition + recording requirement for all law enforcement.

Whether the state requires visible identification and recording during law enforcement operations within state borders, applied to all law enforcement officers including federal civil enforcement. Eighty percent of Americans support officer identification requirements; eighty-six percent support body cameras. Federal courts in 2026 have found masked federal civil enforcement arrests violate the Fourth and Fifth Amendments. This is a fast-moving standard — re-evaluated quarterly.

0
No identification or recording statute.
1
Partial: 1–2 of three dimensions covered, or comprehensive but with Supremacy Clause vulnerability.
2
Visible identification (badge, name, number, agency) + mask prohibition for all law enforcement on duty + recording requirement (body cam mandate covers state/local interactions OR citizen-recording authorization), framed for all law enforcement, with civil enforcement mechanism. Anchor states: OR (HB 4138), CT (SB 397), NJ (mask ban + ID).

Grade Scale

Total score across the nine standards (0–18) maps to a letter grade. The scale weights toward stronger protection: an A demands real enforcement on multiple fronts, not just statutory language.

F
0–3
D
4–7
C
8–11
B
12–14
A
15–18
39
F
4
D
5
C
4
B
0
A
Mandate-cooperation flag. Some states have mandate-cooperation statutes — laws that affirmatively require state and local officers to assist federal civil enforcement. The flag does not change the score (score measures protection level, not state intent), but it appears on every state's report card to distinguish hostile states from passive ones in the editorial layer.
Court-status modifier (v3.3). A preliminary injunction — partial or full — is flagged in the per-state narrative, but the score holds: a protective statute that survives in part is still a statute the legislature enacted. Only a permanent injunction or a full strike-down (final judgment, appeals exhausted) drops the affected standard to 0, unless severance preserves protective provisions. This applies most often to Standard 4 (Private Detention) and Standard 9 (Federal Agent ID) under Supremacy Clause challenges. Currently: California STD-04 (AB 32 struck down) and New Jersey STD-04 (AB 5207 struck down) drop to 0; California STD-09 (SB 805 under a preliminary injunction pending appeal, Apr 2026) is flagged but holds at its statutory base of 2.

v3.5 — Four Boundary Definitions (August 2026)

v3.5 answers four boundary questions the August verification sweep could not decide without a rule. Each was written into the codebook before it was applied to a cell, and each one’s score effect is stated here rather than left implicit. One of the four moves scores; the others do not.

  • §4.4 — Standard 4 exception to the cooperation-nexus gate. Score-preserving. An operator-bound ban on private detention counts even though it binds a private operator rather than a named state channel. Without this exception the two cleanest private-detention bans in the country could not score at all.
  • §4.8 — “audit authority” defined. Score-neutral. The Attorney-General route is kept conjunctive; a proposed second limb was declined.
  • §4.3 — a three-family access rule for the sensitive-locations standard. This one moves scores, replacing v3.4 §C. A drafted clerk-specific requirement was rejected and the one-point fallback preserved.
  • §4.9(d) — “or comparable” defined for civil-enforcement equivalents.
  • Applied the same day: Maryland 9/C → 12/B · Illinois 15/A → 13/B · New York 12/B → 10/C. Distribution A=0, B=4, C=5, D=4, F=39; national average 2.67 / 18. The A band is empty and was not curved. Two proposed changes were refused on review — one of them an upgrade. (Snapshot at v3.5 adoption, August 2026 — superseded; current grades and averages load live from the scorecard data on this page.)
  • New rule of evidence. Fiscal notes, synopses and committee summaries never source a claim about what a statute provides — enrolled or enacted text only. This is a rule of evidence, not a scoring criterion; no boundary test changed because of it.

v3.4 — Durability Hardening (June 2026)

v3.4 is a score-neutral hardening of the durability rules — it changes no state's grade. It adds one governing design rule: a protection scores only where its state-binding clause withholds a named channel of state cooperation with federal civil enforcement, never a command to federal officers, and a generally-applicable law counts only where blocking that cooperation is a designed function rather than an incidental byproduct.

  • Court-status severability. Where a mask, identification, or recording statute reaches both state and federal officers, the score rests on its application to the state's own officers. A preliminary injunction or federal challenge as to the federal application is flagged in narrative; the score holds. Only a permanent injunction of the state-binding application drops the score.
  • Three scoring gates. A cooperation-nexus test (a credited provision must withhold a named state channel); closed-list boundary tests (each standard's creditable provisions are enumerated, not an open survey); and an attribution rule (one statute scores one standard, by its specific clause), plus a published flag for whether a state holds both load-bearing standards (287(g) ban + judicial-warrant requirement).
  • Standards 3 and 9 refined. The courthouse protection is credited on the state court's authority over its own process; Standard 9 is scored on provisions binding the state's own officers and residents' right to record.
  • Distribution unchanged. A=4, B=3, C=3, D=6, F=36. No grade changed; national average 3.29 / 18. Two expansions (generally-applicable detention-facility and data-privacy standards) were deferred to the January 2027 structural review pending unresolved federal litigation. (v3.4 snapshot — superseded.)

v3.3 — Durability Principle (May 2026)

v3.3 keeps v3.2's binding principle (functional protection > structural template) and adds a second: durability. A protection scores only if it lives in durable binding state law — a codified statute, a state constitution provision, or a state Supreme Court ruling interpreting them. Executive instruments (executive orders, Attorney General directives/opinions/guidance without statutory backing, agency or model policies) can be rescinded by a successor administration without legislative action, so they appear in a state's narrative but no longer score. Rescindability is not durability.

  • Durability principle. Only statutes, state constitution provisions, and state Supreme Court rulings score. EO / AG-directive / agency-policy protections are narrative-only. This tightens the 0-point boundary on every standard — most visibly STD-02 (287(g)) and STD-06 (judicial warrant), where an EO- or AG-directive-only restriction now scores 0 (was 1 under v3.2).
  • Court-status modifier simplified. A preliminary injunction (partial or full) is flagged in narrative and the score holds; only a permanent injunction or full strike-down drops the score to 0. This removes the score volatility the v3.2 "partial injunction docks a point" rule introduced.
  • Re-score (2026-05-20). Six changes from the v3.2 baseline: NY STD-02, MA STD-03, and MN STD-02 each 1→0 (executive-instrument credit dropped); NJ STD-08 2→1 (statutory reporting/oversight but no codified AG enforcement reaching local LE); WA STD-05 2→1 (only the DMV domain is a durable statutory firewall); CA STD-09 1→2 (the SB 805 preliminary injunction now flags rather than docks). Virginia held — its 2026 statutes (SB 783 / HB 1441) are enacted law, not executive orders.
  • Distribution unchanged. A=4, B=2, C=5, D=5, F=36. No grade flipped. National average 3.25 → 3.17 / 18. (Snapshot at v3.3 adoption, May 2026 — current grades and averages load live from the scorecard data on this page.)

v3.2 — Functional Protection Evolution (May 2026)

v3.2 evolved from the v3.1 architecture after a STD-07 audit revealed systemic misclassification (data-privacy statutes credited as cooperation reporting; executive orders treated as statutes; AG guidance treated as binding). A reverse audit of how state legislatures actually draft protective bills validated a binding principle: functional protection > structural template. FL credits protections that exist in state law and function in operation. Where a state has done the legislative work — through committee, legal review, political negotiation — FL credits it. The Federal Limits Act omnibus shows what comprehensive looks like; the scoring measures functional.

  • v3.2 binding principle. Functional protection > structural template. The methodology tests whether the protection works, not whether it matches FL's preferred drafting template. This principle binds interpretation of all nine standards and prevents the structural-rigor drift that v3.1 surfaced.
  • STD-02 2-point bar simplified. Was "287(g) ban AND at least one other contract type." Now "287(g) ban alone" — the lynchpin protection. Supplementary contract bans (IGSAs, JTF MOUs, federally conditioned grants) become per-state narrative material, not score-gating.
  • STD-06 2-point bar simplified. Was "judicial warrant + rights notification." Now "judicial warrant alone." Rights notifications are valuable supplementary protection noted in narrative; they don't gate the structural warrant requirement.
  • STD-06 judicial warrant defined. "Judicial warrant" means signed by a federal Article III judge. Administrative warrants (Forms I-200, I-205) and Immigration Judge orders (IJs are DOJ employees, not Article III) are explicitly excluded.
  • STD-07 functional protection test. 2-point bar revised: state statute requires annual public reporting; reports publicly accessible; covers state and local LE; a reasonably-diligent reader can assess cooperation activity. Two new binding exclusions: PII-tracking exclusion (data-privacy statutes are STD-05 turf) and required instrument class (EO/guidance/model-policy don't satisfy — durability requirement).
  • STD-08 either-mechanism rule. Was "both PRA AND statutory AG authority required." Now "either comprehensive PRA OR comprehensive AG authority with reach to local LE." AG authority limited to state agencies caps at 1/2 (cooperation happens at county jail and municipal LE level).
  • STD-08 remedy floor. Damages OR injunctive relief — either alone satisfies. Declaratory-only does not.
  • STD-09 framing requirement dropped as score-gate. Court-status modifier handles enjoined statutes (CA SB 627 + SB 805 enjoined April 2026). Framing is a drafting choice with litigation-survival implications, not a 2-point gate.
  • STD-05 hard-denial defined. "Hard-denial" = mandatory ("shall not disclose"). Permissive language ("may decline") fails. §1373 carveouts are defensive, not weakening — required for bill survival and explicitly not penalized.
  • STD-07 audit corrections applied. 7 states with v3.1 STD-07 scores resting on instruments that failed proper categorization: CO 2→0 (SB 21-131 is PII data-privacy), IL 1→2 (TRUST Act + AG WFA Public Report meets functional bar), MA 1→0 (EO not statute), CT 1→0 (HB 7212 signing unverified + AG memo non-binding), ME 1→0 (AG model policy reference, not reporting mandate), NM 1→0 (SB 36 is data-privacy), NY 1→0 (AG guidance only).
  • STD-06 and STD-08 corrections applied. States with judicial warrant requirement docked under v3.1 for missing rights notification now clear 2/2: CO, ME, NJ, OR, WA. States with comprehensive AG enforcement authority that previously scored 1/2 because PRA was absent now clear 2/2: CA, CO, IL, NJ, NM, NY, WA. VT STD-08 1→0 (Converse-1983 strictly applied per §4.2 exclusion).

Corrections & Revisions

Grades are derived from enacted statute text. When verification finds a citation that does not support a score — in either direction — the score is corrected and recorded here, in versioned batches. Corrections are never made silently.

  • Publication Correction 1 — August 12, 2026 (6 published surfaces; no grade changed). A full audit of this website checked every rendered number, date and legal claim against the scoring record. No score and no grade changed — the scoring record was right throughout. What was wrong was what the site was publishing. Two bills were presented as signed law that never became law: Connecticut HB 7212 (2025), which died on the House calendar on April 23, 2025 (Public Act 25-29 came from HB 7259, a different bill), and Maryland HB 1018 (2026), which passed the House 98–37 and then died in the Senate — the “signed April 28, 2026” belongs to HB 1017, Chapter 173. Both were verified against the legislature’s own bill-status record. Both had already been recorded as dead in Release 3 below; the correction reached the scored cells but not the legislation feed this site renders, and that gap is now closed. Also corrected: the model-legislation gap tool was serving a superseded scorecard and showed the wrong grade for 16 jurisdictions; the per-standard breakdowns for Illinois, Maryland and New York were stale; three per-standard adoption counts were overstated; and for several hours after the A band emptied, several pages continued to state that one state held an A. Full detail is in the corrections ledger.
  • Correction Release 3 — August 12, 2026 (3 letter moves, 8 score cells, 6 citation corrections, 1 errata). A second pass re-read cells the August 6 audit had confirmed rather than corrected, and settled four boundary questions the codebook had never defined. Where a rule had to be decided to score a cell, the rule was written into the codebook first (v3.5) with its score effects enumerated there — no rule was bent to a result. Maryland 9/18 C → 12/18 B · Illinois 15/18 A → 13/18 B · New York 12/18 B → 10/18 C. National figures: A=0, B=4, C=5, D=4, F=39; average 2.67 / 18. (Snapshot at Release 3, August 12 2026 — superseded; current grades and averages load live from the scorecard data on this page.)
  • No jurisdiction now holds an A — and we did not curve the scale. Illinois, the long-standing leader, is a B. An empty top band is a finding about American state law, not a defect in the instrument. Adjusting the bands to avoid reporting it would be the one change that really would deserve the name “methodology change.”
  • Maryland moved UP a letter. Its private-detention cell had cited HB 1018 (2026) — a bill that passed the House on February 26, 2026 and then died in Senate Judicial Proceedings. Dropping the dead bill and reading the enacted chapter showed the cell had been under-credited: Md. Code, Corr. Servs. §1-103(C) provides that “a private entity may not operate or occupy a building, a structure, or other real property as a detention facility” (1 → 2). Maryland’s officer-identification cell also moved 0 → 2 on Ch. 612 (2026), which pairs a direct prohibition on face coverings with a visible-identification mandate. Ch. 612 is enacted but does not commence until October 1, 2026; it is scored from enactment under our published enacted-versus-effective rule, and that date is flagged wherever the cell is quoted.
  • Two proposed changes were refused — one of them an upgrade. A proposed Maryland downgrade was rejected: “notwithstanding any other provision of this title” in its public-records firewall is a priority clause, not a limit on whom the duty binds. A proposed New York upgrade was also rejected: the provisions offered for the health-records domain are general personal-information rules that never name health records or facilities. A refusal is as much a result as a change, so both are recorded.
  • Six citation corrections, no score changed. California’s 287(g) cell cited Gov. Code §7284.9 — a section that does not exist (the Values Act runs §7284.8 → §7284.10); the real clause is §7284.6(a)(2). California’s transparency cell cited §7284.10 as an annual reporting duty; that section governs corrections-department interview procedures, and the annual report is §7284.6(d) — a transcription error, not a missing law. California’s health-data cite named the wrong bill (SB 81, 2025). Two Connecticut cells still carried HB 7212 (2025), which died on the House calendar April 23, 2025; Release 2 corrected three Connecticut cells for that bill and missed these. Connecticut’s score stands on P.A. 25-29 §4 alone.
  • Errata — a remedy no Illinois statute contains. Our Illinois sensitive-locations entry described the remedy as “$17K minimum damages, 3x multiplier.” No Illinois statute provides that. The language was introduced-bill text in HB 3247, struck by House Floor Amendment No. 2 on April 8, 2025 and absent from the enacted act — but it survived in the fiscal notes, which still recite the struck figure. The claim had reached eight public surfaces. The enacted remedies are not uniform: courthouses carry actual damages plus $10,000 statutory damages; K-12 schools and public higher education allow actual damages only; hospitals and licensed child care carry no private damages remedy at all. New evidence rule: fiscal notes, synopses and committee summaries never source a claim about what a statute provides — enrolled or enacted text only. That is a rule of evidence, not a scoring criterion.
  • A scope gap, named and closed. Our model-legislation page also cited Connecticut HB 7212 as enacted law with an effective date of October 1, 2025 — the same dead bill this ledger had already recorded, with a date that has no referent. It survived because Release 2’s scope was scored scorecard cells; hand-authored narrative pages were never in it. Every non-scorecard public page has now been swept for legislative claims, and 26 false citations were corrected across four files — including a nonexistent California code section cited six times, once as “the upheld model.”
  • Correction Release 2 — August 6, 2026 (39 cells, 15 jurisdictions). A full citation-verification audit checked every scored cell against the enacted text it cites, retrieved from official state legislative sources. Every proposed change was verified twice independently and reviewed by an independent second reviewer before adoption. Corrections ran in both directions: 31 cells down, 2 up, 6 confirmed with corrected citations. Five further cells were touched for court-status or dating notes only, with no change to score or citation; those are currency updates, not corrections, and are excluded from the 39. Classes of error corrected: bills that never became law cited as law (Connecticut, Nevada); provisions struck by amendment before passage (Maryland); protections that live in rescindable directives or model policies rather than statute (New Jersey, Vermont, Washington); miscited sections (Washington, Illinois); location-limited provisions credited as statewide (New York, Oregon); and two enacted laws that had gone uncredited — Connecticut's CGS §54-192h(b)(1)(E) (a named 8 U.S.C. §1357(g) foreclosure, effective October 2025) and Hawaii's Act 141 (2026).
  • Grade movements (Release 2). Superseded figures, kept as dated history — these are the values Release 2 moved FROM and TO, not current grades. California 16/A → 14/B · Colorado 11/C → 10/C · Connecticut 13/B → 11/C · D.C. 6/D → 5/D · Hawaii 1/F → 3/F · Illinois 16/A → 15/A · Maryland 11/C → 9/C · Massachusetts 5/D → 3/F · Nevada 4/D → 2/F · New Jersey 15/A → 10/C · New Mexico 9/C → 6/D · New York 14/B → 12/B · Oregon 17/A → 14/B · Vermont 5/D → 2/F · Washington 12/B → 7/D. Fifteen jurisdictions moved a score; nine moved a letter. National figures moved from A=4 / average 3.31 to A=1, B=3, C=5, D=4, F=39 / average 2.69.
  • Correction Release 1 — August 5, 2026 (Maine, 2 cells). Text verification of Maine's three 2026 laws (effective July 29, 2026): 5 MRSA §4763(1)(B) bars every Maine law-enforcement agency from placing officers under federal-agency supervision or employing officers deputized as federal immigration officers — a functional 287(g) foreclosure (0 → 2); the chaptered text contains no enforcement mechanism, so the remedies cell was corrected (1 → 0). Maine: 7/D → 8/C.
  • Audit coverage. Across Release 1, Release 2, and a follow-on sweep of the remaining states on August 7, 2026, every scored cell that claims any credit has been checked against enacted statute text. The August 7 sweep covered the only three otherwise-unaudited jurisdictions carrying credit — Delaware, Minnesota, and Virginia — checking 8 cells: 6 confirmed, 2 flagged as non-score-affecting, no score changes. The remaining 33 jurisdictions score 0/18 with no citation on any standard, so there is no claim to audit.
  • v3.4.1 clarification (August 2026, ratified). A statute that forecloses 287(g) participation satisfies the contract-prohibition standard whether or not it names the program — the crediting clause must itself withhold the deputization or federal-supervision channel. The boundary test is unchanged; a recorded dissent on one New Jersey clause is preserved in the evidence file. The audit-provenance rule was also adopted: audits record which statutory clauses were read, not only conclusions reached.

Sources & Transparency

Every cite is weighted. We don't treat a state legislature page the same as a partisan tweet. Source weight runs 0.0 (unsourced) to 1.0 (primary government document); only cites at or above 0.6 contribute to a score.

2,289
Publishers tracked
0.6
Cite weight floor
52
Jurisdictions scored
328
Bills cataloged

Anti-commandeering canon

The legal foundation across every standard. The federal government cannot conscript state and local officers to carry out its programs.

New York v. United States (1992) — Congress cannot compel states to enact or administer a federal program
Printz v. United States (1997) — federal government cannot commandeer state officers
Murphy v. NCAA (2018) — federal government cannot commandeer state legislatures

Audit Trail & Versioning

Every grade change is traceable. We preserve prior cites and stamp the methodology version into every state's data record. If you find an error, the path to challenge it is on the page, not buried in an inbox.

Challenge a score

Found an error in a grade? Send a sourced correction (with the cite) to hello@federallimits.org. We check it against the standard and either update the record or explain why we didn't. Corrections are credited.

Cite preservation

When v3.1 or v3.2 corrections lower or shift a score that prior methodology had assigned, we preserve the prior cite in the notes field rather than overwriting it. The audit trail of every change is part of the data, not a separate doc.

Versioning

Every state JSON stamps the methodology version it was scored under (methodology_version) and its schema_version. If you cite our data, cite the version too — methodology evolves and you want to be precise about which scoring round you're referencing.