MCP server for AI agents
Give your AI agent a real radius search tool.
Language models can't reliably list every ZIP code within 30 miles of a warehouse. They estimate, and they miss. RingFence gives your agent a tool that returns the complete, exact answer from a real dataset, worldwide.
Why agents need this tool
Ask a model "which towns are within 50 km of Utrecht?" and you'll get a plausible-sounding list with no way to know what's missing. Ask it for every ZIP code within 25 miles of 02139 and you'll get a sample, when the complete answer is 264 ZIP codes. Radius questions are data questions. RingFence turns them into a tool call with a complete, verifiable answer.
Two tools
radius_search
Every postal code and town within a radius of a location or coordinates, in miles or kilometers. Returns the resolved center, counts, a unique postal code list (nearest first), and full rows. Set fields to postal_codes_only for compact output.
resolve_location
Turns a postal code or town into candidate coordinates, so the agent can confirm "Portland, Maine or Portland, Oregon?" before searching.
Connect in one minute
Sign in, open the Account page, and create your API key. Then point any MCP client that supports remote servers at the endpoint.
Claude Code:
claude mcp add --transport http ringfence https://ringfence.zinglab.com/mcp \
--header "Authorization: Bearer rf_live_YOUR_KEY"
Any MCP client (JSON config):
{
"mcpServers": {
"ringfence": {
"type": "http",
"url": "https://ringfence.zinglab.com/mcp",
"headers": { "Authorization": "Bearer rf_live_YOUR_KEY" }
}
}
}
Prefer plain HTTP? Agents that call REST APIs can use the search endpoint directly, described by the OpenAPI spec.
What agents do with it
- A sales agent builds a call list of every account within 40 miles of the rep's next meeting.
- A marketing agent assembles the ZIP list for a geo-targeted campaign and drops it into the ad platform's format.
- A recruiting agent checks which candidates live within commuting distance of a new office.
- A support agent answers "do you deliver to my town?" against a real service radius.
- An operations agent compares coverage of two warehouse sites, town by town.
Safe to hand to an agent
- Rotate any time. Replace your API key in one click; the old key stops working immediately. Last-used tracking shows when it was last active.
- Read-only tools. The agent can search; it can't change your account or billing.
- Explicit ambiguity. The tool never silently guesses between two Springfields.
Frequently asked questions
Does RingFence have an MCP server?
Yes. RingFence exposes a remote MCP endpoint at https://ringfence.zinglab.com/mcp, authenticated with a RingFence API key as a Bearer token. It provides two tools: radius_search and resolve_location.
Why should an AI agent use a tool instead of answering from memory?
A model answering from memory will omit postal codes, invent ones that don't exist, and misjudge distances. A radius_search call returns every postal code and town inside the circle from a dataset of about 1.8 million postal codes and 236,000 towns, with computed great-circle distances.
What does radius_search return?
The resolved center, counts, a unique list of postal codes nearest first, and rows with postal code, distance, town, state, county, and country. Set fields to postal_codes_only to return just the code list and keep the agent's context small.
How does the agent handle ambiguous places?
If a location matches several places, radius_search returns candidates. The agent can ask the user, call again with a candidate's coordinates, pass a country code, or set auto_pick.