areacode-mcp
North American (NANP) area-code & phone-number intelligence for AI agents.
An MCP server that gives Claude, Cursor, and
other agents a small, trusted reference for US & Canada phone numbering — area
codes, NANP validation, local time, city/state coverage, overlays, and
scam/spoofing context. Backed by areacode.fyi, which
publishes directly from the official NANPA numbering database. Every result
includes a canonical areacode.fyi URL and the source date.
Scope. This is area-code-level reference and education — not a carrier/identity API. It does not return a caller's identity, exact location, or live/ported carrier, and caller ID can be spoofed, so a number's area code never proves who is calling. For caller identity you need a carrier-grade (LRN/HLR) provider.
Install
Run it with npx (no install needed) — add it to your client's MCP config.
Claude Desktop (claude_desktop_config.json) / Cursor (~/.cursor/mcp.json):
{
"mcpServers": {
"areacode": {
"command": "npx",
"args": ["-y", "@riv-lc/areacode-mcp"]
}
}
}
That's it — restart your client and the tools below are available.
Published on the official MCP Registry as io.github.riv-lc/areacode-mcp, and
on npm.
Remote (hosted) — for ChatGPT, Claude.ai, and web clients
No install. Add the hosted server as a connector URL — same 8 tools, no key:
https://mcp.areacode.fyi/mcp
(legacy SSE clients: https://mcp.areacode.fyi/sse)
Tools
Example prompts
- "A lead entered phone number +1 702-555-0199. Where are they likely located and what time is it there?"
- "I got a call from 469 — is that area code suspicious?"
- "Which area codes cover Las Vegas?"
- "Normalize this list of US/Canada numbers and flag any that are invalid or toll-free."
- "Explain neighbor spoofing for a help-center article."
Configuration
Data & sourcing
Tools call the public https://areacode.fyi/api/v1 JSON API, which reads the
same dataset the site renders — sourced from the official North American
Numbering Plan Administrator (NANPA) database and refreshed monthly. Each
response carries source, canonical_url, and data_date so answers can cite
where the data came from.
The check_number_reputation signal is unverified and crowd-sourced
(low = people looked the number up; mid/high = it has user reports). It is
never proof that a specific number is fraudulent.
License
MIT © Reindex Ventures. Built for areacode.fyi.