Skip to content

Korzy MCP

The property data in this catalog answers who owns a parcel, what it is zoned for and what it is worth. None of it answers whether the site can actually be served: whether water and sewer reach it, which treatment plant takes the load, how much transfer capacity the regional grid interface has left, what environmental layers cross the boundary, and which data-center projects are already being built next door. That is the question a developer, a facilities planner or an expanding operator asks before an option payment, and Korzy answers it at parcel level from public records.

Server type: Remote, Streamable HTTP
Endpoint:    https://api.korzy.co/mcp
Auth:        OAuth 2.1 sign-in (agents) or an API key with the mcp:read scope
Version:     1.0.0 (registry co.korzy/korzy, published 2026-11-05)
Tools:       16, all read-only per the vendor
Pricing:     $49/month Professional (3,000 credits), $199/month Business
Coverage:    United States
Vendor:      Korzy, korzy.co (New York City)

The evidence tier is stated rather than implied. Every tool surface sits behind a credential: a keyless initialize against https://api.korzy.co/mcp returned HTTP 401 with {"detail":"Sign in or provide a valid Korzy credential"}, so no tool was called and nothing below is a live payload. What was measured: the vendor's own tool table at korzy.co/docs/mcp/ (200), the authentication rules at korzy.co/docs/authentication/ (200), the plan prices at korzy.co/pricing/ (200), the dataset families at korzy.co/datasets/ (200), the registry record at 1.0.0 confirming the endpoint, and the OpenAPI 3.1 spec at korzy.co/openapi.json (200, 34 routes, each carrying an x-korzy-credits block). The credit figures below are read from that spec, mapped to the tools by route semantics.

What the sixteen tools answer

Tool What comes back Credits
list_datasets datasets, geographic coverage, source dates, layer keys and query limits 0
get_state_summary parcel record counts and field coverage by state 0
get_terrain_metadata terrain source resolution, units, coverage and limits 0
search_parcels parcels at an address; include_owner=true adds owner fields and doubles the cost 1
get_parcel one parcel's attributes by korzy_id, plus a map link 1
get_site_context parcels at a point, water and sewer service, nearby treatment plants and nearby data-center projects 2 (1 to 10, per 10 records)
get_utility_record one water, sewer, treatment, gas or fiber record with standardised attributes 1
get_environment_context environmental layers at a point such as flood zones and wetlands, plus nearby cleanup sites 1 to 2
get_logistics_context nearby ports, rail, roads, airports and warehouses 1 to 2
get_data_center one data-center project with observations, capacity evidence and a dated timeline 1
get_power_plant a power plant and its generators 1
get_transfer_capacity total and available transfer capability for a regional grid interface, in MW 1
get_plant_energy_history generation, fuel, outages and generator inventory for a plant 50
get_regional_energy state prices, energy and emissions, or hourly demand and fuel mix 50
get_coal_mine_shipments reported coal receipts linking a mine to power plants 10
get_parcel_terrain slope and elevation analysis over a parcel by korzy_id 1

The three zero-credit tools are the practical entry point: list_datasets returns coverage and limits before any spend, which is how an agent should open a Korzy session.

Why the utility and grid layer is the reason for the entry

The catalog already carries strong property records. PropRaven holds 191.3M US parcels with ownership, permits, hazard scores and valuations. Placestack reads parcel, owner, assessment, sales, permit and zoning records with permitted uses and dimensional standards. Hercules reads ownership, mortgages and pre-foreclosure cases. DFX covers commercial and multifamily with debt timing. All four answer what the record says about a parcel.

Korzy answers what the ground and the networks around it can carry. get_site_context returns water and sewer service areas for a point together with the treatment plants that serve it, and get_utility_record reads the individual water, sewer, treatment, gas or fiber record with its attributes. get_transfer_capacity returns total and available transfer capability in MW for a named regional interface and date. get_power_plant returns the plant and its generators. get_data_center returns a project with its capacity evidence and dated timeline rather than a directory listing. get_environment_context returns mapped flood and wetland layers plus nearby cleanup sites, and get_parcel_terrain runs slope and elevation across the boundary.

The distinct question is the one an operator asks about a specific address: can this site be served, and who else is already building there. DC Hub covers market-level power for data-center siting across 300+ markets with the seven ISO feeds, and the catalog's energy entry reads all seven ISO interconnection queues as market data. Korzy is the parcel-level read of the same question, which is a different unit of decision.

What it costs, in credits rather than adjectives

Korzy bills tool calls as credits against the same balance as its REST API. A call can make several data requests, and each call counts toward a per-minute rate limit. The credit blocks in the OpenAPI spec are specific: a point query that returns parcels costs 2 credits by default and scales 1 to 10 with per-record pricing at 10 records; include_owner=true doubles the cost of the parcel and site calls; a single utility, plant, project, environment or logistics record costs 1 to 2; and the energy reads are the expensive ones at 50 credits for a plant's history or a region's demand and price series, 10 for coal receipts. At the Professional tier of 3,000 credits a month for $49, that is 1,500 parcel lookups or 60 regional energy reads, and extra credits are $16 per 1,000. Business is $199 a month with unlimited seats and higher limits. Annual billing is $490 and $1,990.

Limits the vendor states, and one this pass adds

  • United States only. The API is built on US public records, and the dataset page scopes coverage to them.
  • Area and list queries are not available yet. The docs say point, address and record lookups today, with area search coming, so a screening pass has to iterate points rather than pull a polygon.
  • No free tier. Three summary tools cost nothing to call, but the endpoint returns 402 with subscription_required when the account has no active subscription, so the zero-credit calls still need a paid account behind them.
  • Scopes are split. An api:read key cannot use the MCP server and an mcp:read key cannot use the REST API; a request outside a key's scope returns 403.
  • Concentration risk. Every one of the sixteen tools is a lookup or a calculation and none writes, but the whole surface is one vendor's aggregation of county, state and federal sources, so a source date that lags is invisible unless the agent reads the summary tools first.

Common questions

Does Korzy replace a property records API?

No. It reads the parcel, but the entries that answer ownership, zoning, valuation and debt sit beside it: PropRaven and Placestack for records and zoning, Hercules for LLC and owner research, DFX for commercial debt timing. Korzy is the site-feasibility read on top of them.

Can an agent use Korzy without a subscription?

Not for data. The three summary tools cost zero credits, but the endpoint answers 402 with subscription_required when the account behind the key has no active plan, so a subscription is the entry ticket and the credits are the meter.

Is every Korzy tool read-only?

The vendor states all sixteen are read-only, and the routes in the OpenAPI spec are lookups and calculations with no write verb among them; the one proposal-shaped tool in its Danish-accounting neighbour does not exist here, and nothing in this surface stores a record. The claim is the vendor's, since no call could be made without a credential.

How to start

  1. Call list_datasets and get_state_summary for the state in question: both cost nothing and both report coverage and source dates.
  2. Run search_parcels on the street address, then get_parcel for the korzy_id it returns.
  3. Run get_site_context on the parcel point for water and sewer service, treatment plants and nearby data-center projects, then get_environment_context and get_logistics_context for the layers and access.
  4. Pull get_transfer_capacity for the regional interface and get_power_plant for the generators behind it only when the site is still live, because 50-credit energy reads are where a screening budget goes.

Integration with CorpusIQ

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects

AI answers on your data. See where every number comes from.

Connect your tools and ask in ChatGPT, Claude, or Perplexity. Enterprise-grade answers in minutes. From $29.95/month.

Start free, no credit card
Contact Terms Privacy Pricing