civic-library-mcp

An MCP server that gives AI agents clean, token-efficient access to US civic & property data — geocoding, census tracts, Opportunity Zones, ACS demographics, and FEMA flood zones — sourced entirely from free federal open data.

No scraping, no paid keys for the core tools, no proprietary databases. Just the official sources (U.S. Census Bureau, U.S. Treasury CDFI Fund, FEMA), wrapped in a small set of well-described tools that return compact JSON instead of the bloated payloads most data MCPs dump into your context window.

Built for developers working on proptech, real-estate, fintech, civic tech, GIS, and housing research with AI agents.

Tools

Four of the five tools work with zero configuration. Only get_tract_demographics needs a free Census API key (see below).

Install

Claude Desktop / Cursor / any MCP client

Add to your MCP config (e.g. claude_desktop_config.json):

{
  "mcpServers": {
    "civic-library": {
      "command": "npx",
      "args": ["-y", "civic-library-mcp"],
      "env": {
        "CENSUS_API_KEY": "your-free-census-key-optional"
      }
    }
  }
}

Claude Code

claude mcp add civic-library -- npx -y civic-library-mcp

The CENSUS_API_KEY line is optional — leave it out and four of the five tools still work.

Census API key (only for demographics)

get_tract_demographics calls the U.S. Census ACS Data API, which requires a free key. Get one in ~30 seconds at https://api.census.gov/data/key_signup.html and set it as CENSUS_API_KEY in the server's environment.

Examples

Check whether a property is in an Opportunity Zone:

// check_opportunity_zone({ "address": "900 Camp St, New Orleans, LA 70130" })
{
  "is_opportunity_zone": true,
  "tract_geoid_2010": "22071013400",
  "designation_type": "Low-Income Community",
  "round": "1.0",
  "authority": "Tax Cuts and Jobs Act of 2017 (26 U.S.C. § 1400Z-1)",
  "resolved_from": "address",
  "source": "U.S. Treasury CDFI Fund (designated 2018-12-14)"
}

Flood risk for a coordinate:

// check_flood_zone({ "latitude": 29.9511, "longitude": -90.0715 })
{
  "flood_zone": "X",
  "zone_subtype": "0.2 PCT ANNUAL CHANCE FLOOD HAZARD",
  "in_special_flood_hazard_area": false,
  "zone_description": "Moderate-to-minimal risk (outside the 1% annual-chance floodplain).",
  "source": "FEMA National Flood Hazard Layer"
}

Tract demographics (needs a key):

// get_tract_demographics({ "tract_geoid": "22071013400",
//   "fields": ["population", "median_household_income", "poverty_rate"] })
{
  "tract_geoid": "22071013400",
  "tract_name": "Census Tract 134, Orleans Parish, Louisiana",
  "acs_dataset": "ACS 5-Year 2023",
  "population": 2276,
  "median_household_income": 41250,
  "poverty_rate": 0.231
}

A note on correctness: Opportunity Zones use 2010 census tracts

Opportunity Zone designations are keyed to 2010 census tract boundaries, but the default Census geocoder returns 2020 tracts — and tract GEOIDs changed between the two vintages. A naive lookup that matches a 2020 GEOID against the OZ list will silently return wrong answers near any boundary that moved.

check_opportunity_zone handles this correctly: it re-resolves the address/coordinate to its 2010 tract before matching. (You can se

... View full README on GitHub