MCP Server & Programmatic Access

Query RealStarGlobal's commercial real estate dataset from Claude, an IDE or your own agent through the Model Context Protocol server — property search, comparables, market statistics, ownership, loans and deal activity.

Tools

Tool	What it returns
search_properties	Filter tracked inventory by market, city, type, class, size, availability, rent and sale status.
get_property	Full record for one property including spaces, tenants, leases, sales, loans and public record.
market_stats	Inventory, vacancy, asking rent, rent growth, cap rate and price per SF by market and property type.
sale_comps	Closed sale comparables filtered by market, type, date range, price and cap rate.
lease_comps	Signed lease comparables with size, term, rent, type and expiry.
rent_benchmark	Asking-rent benchmarks by market, property type and class.
tenant_search	Tenants by name, industry, size and lease expiry window.
owner_portfolio	Holdings, market exposure and asset mix for an ownership entity.
loan_maturities	Loans by maturity window, lender, coupon, LTV and status.
fund_search	Funds by strategy, vintage, size, dry powder and target market.
public_record	Parcel, assessed value, tax, zoning and deed history.
availability	Space-level availability with floor, size, rent, term and status.
deal_activity	Live CRE deal, leasing and development signals by country, city, asset class and date.
absorption	Net absorption, deliveries and vacancy series by market and property type.
underwriting_inputs	Assembled inputs for a valuation or credit screen on one asset.

Install

pip install realstarglobal-mcp

# claude_desktop_config.json
{
  "mcpServers": {
    "realstarglobal": {
      "command": "realstarglobal-mcp",
      "env": { "REALSTARGLOBAL_DATA": "/path/to/realstarglobal.json" }
    }
  }
}

The server is read-only. It exposes the same dataset RealStar renders, so a figure quoted by an agent reconciles with the figure on the corresponding market or listing page.

Related