# nyc-property-intel [Health: Active]

**Category:** 🏠 Real Estate  
**Repository:** https://github.com/ccedacero/nyc-property-intel  
**GitHub Stars:** 7  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/nyc-property-intel

## Description
MCP server giving Claude AI access to 22+ NYC public-record databases for real estate due diligence

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What nyc-property-intel does

The nyc-property-intel MCP server exposes NYC property and neighborhood records through MCP tools for real estate due diligence. It can start with an address or BBL and return a property profile containing ownership, building class, zoning, FAR, assessed value, and lot dimensions. Additional queries cover housing and building violations, complaints, litigations, registrations, permits, tax records, liens, mortgages, sales, rent stabilization, fires, 311 requests, evictions, and nearby crime data.

The data is intended to surface public-record facts and potential risk indicators. It is not an appraisal service and does not estimate a property's market value. The `analyze_property` tool combines the available sub-queries into a broader due diligence summary with property details, financial information, FAR analysis, rent stabilization data, comparable sales, and observations.

## How it works

The hosted deployment is available through an MCP HTTP endpoint. A trial token from the project website is passed as a Bearer authorization header when configuring Claude Code or Claude Desktop. The web chat provides a separate way to submit three free property queries without installing an MCP client or obtaining a token.

For self-hosting, the application runs with Python 3.12 or newer, the `uv` package manager, PostgreSQL 16 or newer, and the `nycdb` command-line tool. NYC datasets are loaded into PostgreSQL, with Socrata used as a fallback. The repository includes a database dump and a script for loading data from source in phases. After loading, the supplied SQL scripts create indexes and materialized views.

Most lookups use locally loaded records from sources including DCP, HPD, DOB, DOF, ACRIS, FDNY, 311, DOI, and NYPD. Data coverage differs by dataset: for example, rent-stabilization counts cover 2007–2017, marshal evictions begin in 2017, FDNY incidents begin in 2013, and NYPD crime data begins in 2006.

## Setup and configuration

Hosted use requires an account and trial token from the project website. Configure the MCP URL and add the token as an Authorization header. Claude Code can register the hosted endpoint with `claude mcp add`; Claude Desktop accepts the same endpoint and header in its MCP configuration.

Self-hosting requires cloning the repository, running `uv sync`, and setting `DATABASE_URL` in a `.env` file or client configuration. The documented local example uses a PostgreSQL database named `nycdb`. A database dump can be restored in about ten minutes according to the project documentation, while loading all source data with the seed script takes substantially longer. `NYC_GEOCLIENT_SUBSCRIPTION_KEY` is optional and can improve address resolution. `SOCRATA_APP_TOKEN` is also optional and provides higher rate limits for selected 311, FDNY, and NYPD requests.

## Tools and capabilities

The nyc-property-intel MCP server includes 18 tools. `lookup_property` is the documented starting point for resolving an address or BBL. The issue and history tools inspect HPD, DOB, ECB/OATH, ACRIS, DOF, and related records.

Other capabilities include:

- Searching comparable sales and neighborhood statistics.
- Reviewing permits, tax assessments, exemptions, liens, mortgages, and encumbrances.
- Checking rent-stabilized unit counts and eviction records.
- Finding FDNY incidents, 311 requests, DOB complaints, and NYPD crime within a configurable radius.
- Running `analyze_property` for a combined due diligence report.

The results reflect the underlying public records and their stated date ranges, so they should be interpreted with attention to dataset coverage and freshness.

_Full upstream README: https://allmcps.com/mcp/nyc-property-intel/readme_