Skip to main content

MCP Server

AI Assistant Integration

Connect Claude, Cursor, Codex, and ChatGPT to our geocoding API using the Model Context Protocol (MCP).

What is MCP?

The Model Context Protocol (MCP) is a standardized way for AI assistants to interact with external tools and services. Our MCP server allows AI assistants like Cursor, Claude, and others to use our geocoding API directly.

✨ Modern Protocol: Uses Streamable HTTP transport (MCP 2025-03-26 specification)

🔑 Sign in with Google: No API key to copy or paste. Same account, same rate limits and credits as the REST API.

Server Information

MCP Server URL

https://mcp.addresscanvas.com/mcp

Authentication

Remote MCP access uses OAuth only. Sign in with Google and approve access through your client.

Available Tools

Free (count against your monthly quota):

  • geocode_address - Coordinates and match provenance; FIPS, Census block, and APN where available
  • standardize_address - Parse and standardize an address

Property & location reports (spend prepaid credits; nothing charged when no data resolves):

  • demographic_report - Census demographics (20 credits)
  • demographic_trend_report - Change over time (20)
  • school_report - Nearby ranked schools (40)
  • crime_report - Agency-level crime & safety (25)
  • hazard_report - FEMA flood + National Risk Index (25)
  • environmental_report - Nearby EPA facilities (25)
  • climate_report - Severe-weather history (20)
  • neighborhood_report - Livability score (30)
  • broadband_report - FCC broadband availability (15)
  • property_intel_report - All eight sections in one (up to 150)

Transport Protocol

Streamable HTTP (MCP 2025-03-26 specification)

Quick Start

1

Nothing to set up first

Add the server to your client below and sign in with Google when prompted. For new users, the first authenticated tool call creates an account with a one-time bonus of 150 free credits.

2

Configure your AI assistant

Choose your client below for connection and sign-in instructions.

3

Start Geocoding

Ask your AI assistant to geocode addresses!

Setup Instructions

Add AddressCanvas as a remote custom connector in Claude. This works in Claude Desktop and on claude.ai.

1 Add a custom connector

Open Customize → Connectors → + Add → Add custom connector. Name it AddressCanvas and enter this server URL: 

https://mcp.addresscanvas.com/mcp

2 Sign in and enable the connector

Follow the OAuth sign-in prompts and approve access with Google. If asked for an OAuth client method, choose Register automatically. Finish adding the connector, then enable it from the + → Connectors menu in your conversation. Workspace users may need an owner to add the connector first.

Alternative: use the Claude Desktop config file

Install Node.js with npm, then open Settings → Developer → Edit Config. Merge this entry into claude_desktop_config.json, keeping any existing servers: 

{
  "mcpServers": {
    "addresscanvas": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.addresscanvas.com/mcp"]
    }
  }
}

Save, fully close and reopen Claude Desktop, and complete Google sign-in when prompted. This config-file method uses the mcp-remote bridge.

Claude custom connector instructions · Desktop bridge instructions

Example prompts to try (the tools take street addresses; there is no landmark lookup):

  • "What are the coordinates for 1600 Pennsylvania Ave NW, Washington, DC?"
  • "Geocode the address: 350 5th Ave, New York, NY"
  • "Standardize this address: 782 n lakewood ave ocoee fl"
  • "Geocode 782 N Lakewood Ave, Ocoee, FL 34761 and give me its census block"

💡 Pro Tip: Once configured, Claude will automatically use the geocoding tools when you ask location-related questions!

Cursor IDE natively supports MCP servers via Streamable HTTP. Simple URL-based configuration - no Node.js required!

1 Edit MCP Configuration File

Open or create ~/.cursor/mcp.json and add: 

{
  "mcpServers": {
    "addresscanvas": {
      "url": "https://mcp.addresscanvas.com/mcp"
    }
  }
}

💡 That's it! Just the URL. Cursor handles the Streamable HTTP protocol and the Google sign-in automatically.

2 Restart Cursor

Completely close and reopen Cursor IDE to load the new MCP server.

3 Verify Connection

Check Cursor's MCP logs to verify the connection:

~/.cursor-server/data/logs/*/exthost*/anysphere.cursor-mcp/

🔓 Signing in: The first time your client connects, a browser window opens to sign in with Google and approve access. There is no key to paste.

Why This Setup is Better

  • No Node.js or npm installation required
  • Direct HTTP connection - more reliable
  • Simple configuration - a URL, then sign in with Google
  • Uses modern Streamable HTTP transport (MCP 2025-03-26 spec)
  • Faster connection establishment

How to use in Cursor:

  1. Open the AI chat panel in Cursor (Cmd+L or Ctrl+L)
  2. The geocoder tools will be available to the AI assistant
  3. Ask questions like:
    • "What are the coordinates for 1600 Pennsylvania Ave NW, Washington, DC?"
    • "Geocode: 350 5th Ave, New York, NY"
    • "Standardize the address: 782 n lakewood ave ocoee fl"
  4. The AI will automatically use the geocoding tools to answer

💡 Pro Tip: You can use the geocoding tools while coding to validate addresses, get coordinates for testing, or standardize address formats in your codebase!

Codex CLI and the IDE extension connect directly over Streamable HTTP and support Google sign-in through OAuth.

1 Add the server and sign in

Run these commands in your terminal:

codex mcp add addresscanvas --url https://mcp.addresscanvas.com/mcp
codex mcp login addresscanvas

Complete Google sign-in in the browser and approve access.

Alternative: edit config.toml

Add this table to ~/.codex/config.toml, keeping your existing settings, then run codex mcp login addresscanvas: 

[mcp_servers.addresscanvas]
url = "https://mcp.addresscanvas.com/mcp"

2 Verify the connection

Run codex mcp list to check the saved configuration. Start a new Codex session and use /mcp to check that the tools are available.

The IDE extension shares this configuration. Restart the extension after adding the server; select Authenticate in its MCP server settings if sign-in is still needed.

Official Codex MCP documentation

Connect ChatGPT to AddressCanvas as a custom MCP plugin on the web. Availability depends on your account and workspace permissions.

1 Add a custom MCP server

Open Plugins in ChatGPT, select the + button, then Add custom MCP server. Enter AddressCanvas as the name and use this Server URL: 

https://mcp.addresscanvas.com/mcp

2 Configure OAuth and sign in

Choose OAuth authentication and Dynamic Client Registration (DCR) if asked for a client registration method. Review ChatGPT's connection notice and select Create as a plugin. Complete Google sign-in and approve access when prompted.

3 Install and use AddressCanvas

Find the resulting plugin in your personal or workspace plugins and install it. Start a new conversation, type @, and select AddressCanvas. Try: “Use AddressCanvas to geocode 350 5th Ave, New York, NY.”

Official ChatGPT custom MCP setup guide

Custom GPT Actions and REST integrations

For a Custom GPT Action, use the REST API at https://addresscanvas.com/api/v1/geocode. Configure a custom API-key header named X-API-Key with a key from your dashboard. See our REST API documentation for request schemas and endpoints.

Applications can also call the REST API from an OpenAI function-calling handler.

REST API Usage

The REST API accepts JSON requests authenticated with your API key. MCP clients use the OAuth setup above.

REST API base URL: https://addresscanvas.com. See the API Documentation for details.

Geocode an Address (Main API) 

curl -X POST "https://addresscanvas.com/api/v1/geocode" \
  -H "X-API-Key: gck_your_api_key_here" \
  -H "Content-Type: application/json" \
  -d '{"address":"1600 Pennsylvania Ave NW"}'

📚 Learn More: For complete API documentation including response formats, rate limits, and advanced features, visit the Interactive API Documentation.

Features

  • Google sign-in through OAuth
  • Respects your rate limits
  • Two geocoding tools and ten property reports
  • Production-ready and secure
  • Full error handling

Security

  • OAuth authentication required for MCP
  • Rate limiting enforced
  • Provider-verified sign-in required
  • HTTPS encryption
  • Usage tracking & monitoring

Rate Limits

Geocoding and standardization share your account's monthly allowance across REST and MCP. Reports spend prepaid credits instead of this monthly allowance. Burst limits still apply.

Developer

5,000

requests per month

Property Pro

50,000

requests per month

Growth

500,000

requests per month

Enterprise

1,000,000+

requests per month

Support & Resources

API Documentation

Complete API reference and examples

Get 150 free credits

Sign in with Google for MCP — 5,000 free requests a month

Contact Support

Get help from our team

Server Status

Check API server health

Ready to Get Started?

Connect your AI assistant and sign in with Google to start using our MCP server.

Get 150 free credits View API Docs