osrmroute
MCP · AI Agents

Location tools for your AI agents

Give any AI agent geocoding, routing, distance matrix, isochrones and route optimization — through one Model Context Protocol server. Works with Claude, Cursor and any MCP client.

Agent workflow

One server, every location skill

The Model Context Protocol (MCP) is how modern AI agents plug into external tools. Our MCP server exposes selected OSRMRoute routing, geocoding and optimization capabilities as ready-to-call tools, so your agent can turn an address into coordinates, plan a route, or optimize a delivery run — without you wiring up any APIs by hand.

15 tools out of the box

Every tool runs on our own maps, with your API key.

geocode_address

Address or place name → coordinates

reverse_geocode

Coordinates → nearest address

autocomplete_address

Partial text → ranked address suggestions

get_route

Driving, cycling or walking route with distance & time

distance_matrix

Travel time & distance between every pair of points

get_isochrone

Reachable area within a time budget

optimize_route

Multi-stop, multi-vehicle route optimization

get_timezone

Timezone & UTC offset for a coordinate

natural_search

Find places using a short English query.

suggest_addresses

Build place search with suggestions and a secure retrieval step.

retrieve_place

Read source attributes from a retrieved suggestion or route search.

place_details

Read source attributes from a retrieved suggestion or route search.

list_styles

Manage map styles, datasets and publication from your application.

list_datasets

Find nearby features in your own uploaded geographic datasets.

preview_style

Describe palette and layer changes, review the map, then apply.

Set up in one minute

1

Get your API key

Create a key in your OSRMRoute dashboard — the free tier is enough to start.

2

Add the server to your MCP client

Drop this into your Claude Desktop, Cursor or Cline config and restart. That's it — the tools appear in your agent.

3

Ask your agent

"Optimize this delivery route" or "how far is it from here?" — your agent now calls real maps.

{
  "mcpServers": {
    "osrmroute": {
      "command": "npx",
      "args": ["-y", "https://osrmroute.com/downloads/osrmroute-mcp-1.1.2.tgz"],
      "env": { "OSRMROUTE_API_KEY": "YOUR_API_KEY" }
    }
  }
}

Why OSRMRoute for agents

Worldwide coverage

Driving routes and address search worldwide. Walking, cycling and detailed local data vary by service and region; see Coverage for availability.

Compare your workload

A matrix costs one credit per ten source–destination cells, rounded up, with a minimum of one. A 10 × 10 matrix costs 10 credits. It uses your allowance first, then purchased credits.

AI Agents · MCP

No juggling separate geocoding, routing and optimization vendors — one server covers the whole location stack.

Questions

Which MCP clients are supported?

Any client that speaks the Model Context Protocol — Claude Desktop, Cursor, Cline and others. The server runs over stdio via npx, so there is nothing to host.

How is it billed?

Tool calls use your normal OSRMRoute API key and count against your plan, exactly like direct API requests. No separate MCP fee.

Give your agent a sense of place

Install the MCP server and start calling real maps in minutes.

Integration / migrationUsage calculatorSandboxMeasured evidence

OSRMRoute Support

Talk directly to our team

HERE TO HELP

How can we help?

Send a message. Ticket replies are also sent to your verified email.

Your chats

Your chats will appear here.