Every licensed childcare facility across five states — New York, New Jersey, Colorado, Connecticut and Delaware — in ONE unified schema. Daycares, licensed centers, family child-care homes: name, type, address, county, capacity, license status, and published compliance records where available. Each facility carries its inspection count when the state publishes one and, for Delaware, cited-deficiency counts with the top regulations cited. The Actor's advantage is a stable API/Apify handoff across these five official state sources, not a claim that no other service normalizes childcare data.

Sister actor to our single-state ../childcare-licensing-tx — same buyers, wider map.

Coverage (New Jersey re-verified 2026-08-24; remaining endpoints verified 2026-07-12)

State	Facilities	Inspection data	Deficiency data
NY New York	16,798	—	—
NJ New Jersey	4,075 active centers at the latest check	—	—
CO Colorado	4,508	—	—
CT Connecticut	16,188	inspection_count (97,283-row feed, join verified 107/107)	—
DE Delaware	1,249	inspection_count (visits)	deficiency_count + top regulations cited (11,853-row feed, join verified 165/165)

Unified record

{
  "state": "DE",
  "name": "YMCA Of Delaware / Central Branch",
  "type": "Licensed Early Care and Education",
  "address": "11 W 11TH ST",
  "city": "WILMINGTON", "county": "New Castle", "state_code": "DE", "zip": "19801",
  "license_number": "27399", "license_status": "Licensed",
  "capacity": 120, "issue_date": null, "expiration_date": null,
  "inspection_count": 6,
  "deficiency_count": 14,
  "deficiency_details": { "non_compliance_count": 9, "top_regulations_cited": [ { "desc": "Ratios, Group Size, and Supervision", "count": 3 } ] },
  "raw": { "...": "full original source row" }
}

Compliance fields are null, never fabricated, for states/facilities where the source publishes no inspection data. NY suppresses the street address for 655 family day-care homes (address_omitted) — those are honest nulls too.

New Jersey records also carry source_attribution, source_metadata_url, and source_disclaimer. Those fields preserve NJDEP's required credit, metadata path, and notice that this secondary product has not been verified or authorized by NJDEP.

Who uses this

• Compliance and quality teams — review published licensing and deficiency history across states.
• Childcare SaaS & marketplaces — market sizing and facility research spanning states in one shape.
• Franchise & site-selection teams — capacity and competitor gaps by county across states.
• Journalists & researchers — cross-state compliance analysis.

Input examples

• Low-cost default sample (Connecticut, no inspection merge): {}
• All five states, full compliance merge: { "states": ["NY","NJ","CO","CT","DE"], "includeInspections": true }
• Just the inspection states: { "states": ["CT", "DE"] }
• One county: { "states": ["DE"], "county": "Sussex" }
• Compliance screen (DE deficiencies): { "states": ["DE"], "minDeficiencies": 5 }
• Fast licensing-only pull: { "includeInspections": false }

Input reference

field	meaning
states	subset of ["NY","NJ","CO","CT","DE"] (default: Connecticut)
county	one county; server-side where the feed has a county column (CT has none)
licenseStatus	case-insensitive substring on license status
includeInspections	merge inspection/deficiency feeds (CT, DE). default false
minDeficiencies	only facilities with ≥ N cited deficiencies (DE only)
maxRecordsPerState	cap facilities per state
maxInspectionRows	cap inspection rows read per state before building the lookup

Proof

npm run smoke (or node test_smoke.mjs) pulls real rows from every state against the live endpoints, asserts content + state tagging, and proves the CT/DE compliance joins produce non-null counts on real facilities. No account, no cost — all sources are public state open-data portals accessed logged-out.

Source & legality

Official state open-data portals (data.ny.gov, mapsdep.nj.gov, data.colorado.gov, data.ct.gov, data.delaware.gov), public and accessed logged-out through Socrata SODA or ArcGIS REST. New Jersey uses NJDEP's monthly active-center layer because its former Socrata table stopped updating in 2019. Publisher cadence and source-health gates remain distinct from the time an Actor run is observed.

Completeness and subsidy semantics (September 13, 2026 source update)

CT/DE inspection joins expose inspection_evidence.status and observed counts. A capped feed cannot establish complete totals: inspection_count and deficiency_count remain null until that detail query is exhausted. An exact cap is conservatively partial. Unavailable detail still fails the existing pipeline instead of becoming zero. Connecticut county filters are rejected before retrieval because the source has no county field.

Colorado license status comes from identifiable membership in the published licensed facility cohort, never CCCAP subsidy authorization. The original subsidy value is separate in subsidy_authorization_status; recognized booleans map to accepts_subsidies, otherwise null. An empty source record cannot imply licensing. Run node test_completeness.mjs and node test_unit.mjs for local regressions.

Connecticut capacity uses a finite nonnegative published regularcapacity when available, including zero. Otherwise, a valid maximumcapacity is retained with capacity_basis: maximumcapacity. capacity_evidence preserves both normalized values and their missing/null/blank/invalid states, and flags differing published values without asserting that the two source concepts are interchangeable. Capacity is not attendance or available places. The source's literal license number is preserved; a weak value such as DCCC is not replaced with an invented license ID. Run node test_ct_capacity.mjs for boundary and material-field regressions.