Skip to content Discuss your data requirements with our team. Get in touch ↗Discuss your data requirements with our team. Get in touch ↗

For AI agents: the documentation index is /docs/llms.txt, and this page is available as Markdown.

Find

List one type of record, such as properties, sales, permits or loans, filtered with where clauses, inside a place or around a point. Rows cost $1 per 1,000. A count is free.

GET/v1/find/{type}

POST/v1/find/{type}

Parameters

Name Type Description
type path What to list: properties, sales, permits, tax_notices, distress_filings, recorded_loans, cmbs_loans, reit_properties, sba_loans, sba_lenders, warn_notices, listings, facilities, establishments, government_comps. See types and fields.
where string, repeats One clause, such as building_sqft >= 50000. Repeat it for more; all must match. Up to 20. See the where grammar.
in string A place: a city (Oakland, CA), a county (Maricopa County, AZ), a ZIP code (60614), or a neighborhood of New York City or San Francisco (Upper West Side). Neighborhood names don't resolve elsewhere: send a ZIP code, lat and lng with radius_m, or within.
near string A property ID, or longitude,latitude (longitude first).
lat number Latitude of a point, -90 to 90. Send with lng instead of near.
lng number Longitude of a point, -180 to 180. Send with lat instead of near.
radius_m number Meters around near or lat and lng, from 50 to 50,000. Default one mile (1,609).
sort string A sort the type takes, listed under each type. A leading - means largest or newest first. Field names differ by type: properties take -last_sale_date, sales take -sale_date, and permits take no sort.
limit integer Rows per page, up to 100. Default 25.
cursor string next_cursor from the previous page.
count boolean properties and sales: the number of matches instead of rows, free. Other types don't count: page their rows with cursor until has_more is false.
stats string With count: the field to summarize with a median and quartiles. Default building_sqft.
group_by string properties only. owner ranks owners by how much they hold, with affiliates combined.
max_cost_usd number Refuse the call if a full page could cost more on your plan.

POST takes the same names as a JSON body, with where as an array and lat and lng as numbers, plus within for properties: a GeoJSON Polygon or MultiPolygon (or a Feature holding one) to search inside instead of in or near. Up to 8 polygons of up to 40 vertices each, outer rings only, longitude first. A within page's next_cursor goes back with the same within.

To search around a point, send lat and lng with radius_m, such as lat=37.8044&lng=-122.2712&radius_m=800. It's the same as near=-122.2712,37.8044; send one or the other, not both. Sales near something are comparables, so sales takes a property ID in near, not a point.

Request

curl --get "https://api.westkite.com/v1/find/properties" \
  -H "Authorization: Bearer $WESTKITE_API_KEY" \
  --data-urlencode "in=Austin, TX" \
  --data-urlencode "where=property_type = office" \
  --data-urlencode "where=building_sqft >= 50000" \
  --data-urlencode "where=held_years >= 15" \
  --data-urlencode "sort=-building_sqft"
const params = new URLSearchParams({ in: "Austin, TX", sort: "-building_sqft" });
for (const clause of ["property_type = office", "building_sqft >= 50000", "held_years >= 15"]) {
  params.append("where", clause);
}
const res = await fetch(`https://api.westkite.com/v1/find/properties?${params}`, {
  headers: { Authorization: `Bearer ${process.env.WESTKITE_API_KEY}` },
});
const { data, billing } = await res.json();
import os, requests

res = requests.get(
    "https://api.westkite.com/v1/find/properties",
    params=[
        ("in", "Austin, TX"),
        ("where", "property_type = office"),
        ("where", "building_sqft >= 50000"),
        ("where", "held_years >= 15"),
        ("sort", "-building_sqft"),
    ],
    headers={"Authorization": f"Bearer {os.environ['WESTKITE_API_KEY']}"},
)
body = res.json()

Response

{
  "data": {
    "type": "properties",
    "properties": [
      {
        "id": "property:3f2b9c1e-…",
        "address": "6836 Austin Center Blvd, Austin, TX 78731",
        "property_type": "Office",
        "building_sqft": 150220,
        "year_built": 1984,
        "held_years": 18
      }
    ],
    "next_cursor": "cur_41b0…"
  },
  "missing": [],
  "billing": {
    "operation_id": "op_7c20…",
    "operation": "find",
    "charged_usd": 0.001,
    "balance_usd": 9.96
  },
  "request_id": "req_2a9f…"
}

The charge is per row returned: a full page of 25 rows is $0.025. Pages come back whole, so an answer can hold a few rows more than limit, but a call never costs more than limit rows. To size a market before you pay for rows, ask for a count:

curl --get "https://api.westkite.com/v1/find/properties" \
  -H "Authorization: Bearer $WESTKITE_API_KEY" \
  --data-urlencode "in=Oakland, CA" \
  --data-urlencode "where=property_type = office" \
  --data-urlencode "count=true" \
  --data-urlencode "stats=last_sale_price"

For a whole market, use an export instead of paging: it returns one file with an exact total.

The where grammar

Each clause is one string: field op value. It reads the same in a query string, a JSON array or the CLI's --where.

Operator Means Example
= Equals zoning = C-2
!= Leaves one out. property_type and owner_type only. owner_type != government
~ Contains owner ~ Prologis
>= At least last_sale_price >= 1000000
<= At most year_built <= 1960
between Inclusive range units between 2 and 10
in Any of property_type in office,retail
is known Has a value flood_zone is known
is unknown Has no published value last_sale_date is unknown
  • Bounds are inclusive. > and < are refused rather than guessed, so a strict bound is never widened without you knowing.
  • Numbers may carry commas: 1,000,000. true and false are typed. Quotes around a value are optional.
  • Dates are YYYY-MM-DD. A bare year means its first day in a lower bound and its last day in an upper one, so last_sale_date between 2024 and 2025 covers both years.
  • Field names are snake case, as results name them. Camel case, like lastSalePrice, is read the same.
  • A field the type doesn't have is refused with the list of fields it does have.

Types and fields

A field a type doesn't take is refused with invalid_request and the list of fields it does take, so an agent can correct itself. Fields lists every field below as data, with its type, unit, operators, values and how much of the data carries it. The API reference lists them all, generated from the API itself.

properties

Parcels and buildings. Takes in, near, radius_m, sort, cursor, count with stats, and group_by=owner.

Field Operators Notes
property_type =, in, != Office, Industrial, Retail, Multifamily, Land, Residential, Hospitality, Mixed use, Civic / institutional. != leaves one type out.
owner ~, = ~ matches the owner of record. = is a company with its affiliates, such as owner = Blackstone, and needs in.
last_sale_date >=, <=, =, between, is known YYYY-MM-DD
last_sale_price >=, <=, =, between, is known Dollars
building_sqft, lot_sqft >=, <=, =, between, is known Square feet
year_built, stories, units >=, <=, =, between, is known
bedrooms, bathrooms >=, <=, =, between, is known
held_years >=, <=, =, between, is known Years since the owner of record took title
market_value, assessed_value, tax_amount >=, <=, =, between, is known From the assessor. Market value is the assessor's, not an appraisal.
zoning, flood_zone, subtype =, ~, is known As the publisher writes them
street = The street the property's address is on, such as Willis Ave
zoning_class = commercial, industrial, residential, mixed use. Read from the zoning code's opening letters: a reading, not a legal finding.
owner_type =, != government, nonprofit, llc, corporation, trust, estate, individual. Read from the owner name. != leaves one class out; repeat it for more.
owner_occupied = yes: the owner runs a business registered at the address
signal = Long-held, Underbuilt, Recent permit, Tenant change
building_to_lot_ratio <= Building area over lot area, for underbuilt sites. 0.5 is a common screen.
buildable_sqft, unused_buildable_sqft >=, <=, between New York City estimates from lot area and zoning
wetland_pct, floodplain_pct >=, <=, =, between, is known, is unknown Percent of the parcel, 0 to 100: mapped wetland, and the 100-year floodplain. flood_sfha_pct, the record's name, reads the same as floodplain_pct.
upland_acres, largest_upland_acres >=, <=, =, between, is known, is unknown Acres outside the floodplain, wetlands and water, and the largest contiguous piece of it
road_frontage_ft, water_frontage_ft >=, <=, =, between, is known, is unknown Feet of parcel edge along public roads, or along water
site_access = landlocked (no public road within 20 m) or road access. landlocked = true and landlocked = false read the same.
waterfront = any, or one of bay, canal, coast, lake, pond, river, stream, swamp. waterfront_type, the record's name, reads the same.
aerial_land = built, cleared, wooded, wet, mixed. Read from the newest public aerial photo.
aerial_parking_spaces, aerial_loading_docks >=, <=, between Counted from the air
aerial_building_pct, aerial_vegetation_pct >=, <=, between Share of the lot
aerial_roof = good, fair, poor
aerial_feature = construction, pool, solar

Site facts and aerial readings match only parcels that have been measured. So wetland_pct = 0 finds measured parcels with no wetland, and wetland_pct is unknown finds the ones not measured. What each site fact means is on Properties.

Site facts are measured county by county. Where no parcel in the searched counties has been measured for a site fact, its clause isn't applied: rows match your other clauses, summary says which clause was left out, and missing lists the field as not_covered.

Sorts: held_years, market_value, assessed_value, last_sale_price, building_sqft, lot_sqft, year_built, last_sale_date, each with - for descending.

sales

Recorded sales. With near set to a property ID, these are its comparables: closed sales from deeds, with nominal deeds left out and portfolio deeds flagged. With in, each property's latest sale.

Field Operators Notes
sale_date >=, between, = Near a property: from that date. In a place: also <=.
price >=, <=, between In a place only
property_type =
units, lot_sqft, building_sqft >=, <=, between
portfolio = false drops multi-parcel deeds. Near a property only.

Sorts: -sale_date and distance near a property; sale_date and -sale_date in a place.

permits

Building permits linked to properties. Takes in, near, radius_m and cursor. It takes no sort.

Field Operators Notes
date >=, <=, =, between Defaults to the last 12 months
date_kind = activity, filed, issued, completed
description ~ A keyword or permit number
status =
valuation >= Dollars
property_type =, in

tax_notices

Parcels on published tax-distress notices. Notices are dated, so taxes may have been paid since. Sorts: -amount, -notice_date, -times_listed.

Field Operators Notes
kind =, in delinquent, tax_sale, lands_available
tax_year =
all_years = true includes earlier years
amount >= Dollars owed
times_listed >=
property_type =, in

distress_filings

Court and lien filings on a parcel: mortgage foreclosures and lis pendens, notices of default or of a trustee's sale, foreclosed properties registered with a city, sheriff's sales, tax liens filed by a city, county or school district, eviction cases and probate estates. Each property is listed once, with its newest matching filing and how many it has. Coverage is the counties whose courts and recorders publish these filings; when nothing matches, filingsCoverage says whether West Kite could check where filings are loaded. With checked, filingsLoadedFor lists those counties in the states searched (an empty list: none loaded there yet). With unavailable, because the scope names no state or the check did not finish, there is no list and missing names filingsLoadedFor: zero matches is then not evidence that none exist. A filing may have been paid, dismissed or resolved since. filedOn is the filing's date (for a sheriff's sale, the day of the sale) and is null where the publisher gives none. Amounts differ in meaning by kind (a claim, a lien, a sale price), and one claim can cover several parcels, so they are never totalled. Names of private parties (defendants, tenants, individual lenders, heirs) are never kept; a lender or lien buyer is named only when it is an organisation. Sorts: -filed_on (default), -amount.

Field Type Operators Values Notes
kind enum = in foreclosure, foreclosure_registration, sheriff_sale, tax_lien, eviction, probate
filed_on date >= <= between
status enum = open, any
amount number, usd >=
property_type enum = in Office, Industrial, Retail, Multifamily, Land, Residential, Hospitality, Mixed use, Civic / institutional

recorded_loans

New York City only: each building's latest recorded financing (mortgage, CEMA or consolidation) with lender and amount. Records carry no maturity date. Sorts: recorded_on, -recorded_on, -amount.

Field Operators Notes
recorded_on >=, <=, between
amount >=
lender ~
lender_kind =, in bank, insurer, nonbank, private, government
satisfied = true includes satisfied loans
stale = true includes stale ones
lots <= Leaves out blanket loans over more lots
property_type =, in

cmbs_loans

Loans in commercial mortgage-backed securities, from servicer filings, for maturity walls and refinancing risk. in takes City, ST or a state. Sorts: maturity, -balance.

Field Operators Notes
maturity >=, <=, between Default: the next 24 months
state = Two letters
city =, in
property_type = office, retail, multifamily, industrial, warehouse, hotel, mixed use, self storage, manufactured housing or healthcare
balance >=

sba_loans and sba_lenders

Published SBA 7(a) and 504 loans, and lenders ranked by them. in takes County, ST or a state.

Field Operators Notes
borrower, lender ~
date >=, <=, between
amount >=
program = 7a or 504
status =
signal = recent, financed, fixed-assets, purchase
property_id =

warn_notices

California layoff, closure and relocation notices: possible space coming back, not confirmed vacancy. in takes a California county.

Field Operators Notes
employer ~
date >=, <=, between
signal = recent, large, closure, repeat
property_id =

listings

Advertised spaces for sale or lease, with source evidence. Takes in and limit.

Field Operators Notes
transaction = sale or lease
use = office, retail, industrial and so on
sqft >=, <=, between
keyword ~
property_id =
contacts = true adds broker contacts

facilities

Schools nationwide, from the US Department of Education, and health facilities in New York State, each matched to its lot and that lot's owner. Takes in or near, and limit up to 50; one page, no cursor.

Field Operators Notes
kind =, in private_school, public_school, nursing_home, hospital, clinic. Default: private schools, nursing homes, hospitals and clinics.
owner_operates = true keeps facilities whose operator's name matches the lot's owner
property_type =, in

establishments

Named businesses and operators at an address from public registers, such as bank branches, licensed premises, care facilities and schools, each with the property it stands on when one matches. Every answer says per kind and state whether a source is loaded, so no rows is never read as none. Takes in, near with radius_m, limit and cursor. The kinds, coverage and examples are on Establishments.

Field Type Operators Values Notes
kind enum = in bank_branch, credit_union, liquor_license, snap_retailer, pharmacy, medical_practice, care_facility, health_center, child_care, school, post_office, government_facility, nonprofit, industrial_facility, ev_charger, fuel_station, airport, port_terminal, refinery, gas_storage, petroleum_terminal, wind_turbine, solar_farm, stream_gauge, wic_vendor, casino, short_term_rental, short_term_rental_market, campground, library, mobile_home_park, oil_gas_well, water_well, water_tower, cannabis_license, vehicle_dealer, contractor, consumer_lender, lottery_retailer, lodging, billboard, water_right, workplace, layoff_notice, federal_contractor, ppp_loan, visa_employer_worksite, motor_carrier, drug_establishment, sales_tax_permit, subsidized_housing
name string = ~ (business or operator name contains) or =
category string = As published: main_office, on_premise, nursing_home
status string = As published: open, active, pending
opened_on date >= <= between YYYY-MM-DD
property_id string = Establishments matched to this property
linked boolean = = true (matched to one property) or = false

government_comps

San Francisco only: purchase prices and appraised values quoted in public government financing packets and Board legislation, near one property. Each keeps its document's status: proposed, requested, approved or completed. Takes near (a property ID, required), radius_m from 100 to 16,094 (default 3,219) and limit up to 100; one page, no cursor.

Market series have their own endpoint: Markets. Which places each type covers is on Data sources.

Errors

Code When
invalid_request An unknown type or field, an operator the field doesn't take, or a clause that can't be read. Also lat without lng (or the reverse), lat and lng with near, or degrees out of range. The message names what's allowed.
max_cost_too_low A full page would cost more than your max_cost_usd.
scope_missing The key doesn't have the data scope.
spend_limit_exceeded The key's monthly cap would be passed.
insufficient_balance Your balance can't cover the page.
rate_limited Too many requests this minute.