A free MCP server and API for 95,000+ US EV charging stations

Wins Parking — Sun, 25 Oct 2026 15:52:02 +0000

68% of EV drivers hit a broken charger in a single year (Plug In America survey). Most charger data an AI assistant can reach is stale, scraped or undated.

ChargingNear.me covers 95,724 US charging stations in all 50 states and DC, including 17,028 DC fast chargers from 98 networks. It syncs daily from the U.S. Department of Energy NREL database and cross-checks Open Charge Map, Google Places and TomTom. Every station page carries the date it was last verified.

We opened that data to agents and developers.

MCP server

https://chargingnear.me/api/mcp

Remote, streamable HTTP, read-only, no key. Listed in the official MCP Registry as me.chargingnear/mcp.

Tool	What it does
`search_stations`	Filter by city, state, network, connector, speed, power or live availability
`get_nearest_stations`	Nearest stations to a latitude/longitude
`get_station_details`	Full record for one station
`find_stations_on_route`	Stations along a list of waypoints, for trip planning
`get_city_charging_stats`	Station count, network mix and top stations for a city
`get_state_charging_stats`	The same for a state

Add it to Claude Desktop, Cursor or any client that supports remote MCP:

{
  "mcpServers": {
    "chargingnearme": {
      "url": "https://chargingnear.me/api/mcp"
    }
  }
}

Each result carries a citation_url, and the server tells the model to cite it. Answers point back to a dated station page instead of a guess.

REST API

No key needed for 100 requests a day:

curl "https://chargingnear.me/api/v1/stations/nearest?lat=39.7392&lng=-104.9903&radius=5&limit=10"

The OpenAPI 3.1 spec is at https://chargingnear.me/.well-known/openapi.json. GeoJSON feeds cover map use. The base directory is CC-BY-4.0.

What people build with it

A charger finder inside a fleet or rideshare app
An EV road-trip assistant that plans stops by connector and speed
Market-gap maps for site selection: which cities have Level 2 but no DC fast

Park Graph MCP: let Claude and ChatGPT find, price and book parking

Wins Parking — Sun, 25 Oct 2026 15:51:12 +0000

Most parking lots are invisible to AI assistants. Ask Claude or ChatGPT to "find me parking near Coors Field tonight" and it can only guess from web pages.

We built Park Graph to fix that. Lot owners print one QR sign. Drivers scan and pay in 30 seconds. And every lot on Park Graph is exposed to AI agents through a hosted Model Context Protocol (MCP) server.

The endpoint

https://parkgraph.com/api/mcp

Remote, streamable HTTP. The driver flow needs no API key. It is listed in the official MCP Registry as com.parkgraph/mcp and on npm as parkgraph-mcp.

Tools

Tool	What it does
`search_lots`	Lots near a latitude/longitude, nearest first, with spaces left and the current rate
`get_lot`	Address, hours, features, pricing and live spaces for one lot
`quote`	Price for N hours at the current rate, daily max applied
`reserve`	Holds a space for a few minutes while the driver confirms
`get_session`	Status, times, amount and extensions for a parking session
`extend_session`	Adds time to an active hourly session

Connect it

Claude Desktop, Cursor or any MCP client that supports remote servers:

{
  "mcpServers": {
    "parkgraph": {
      "url": "https://parkgraph.com/api/mcp"
    }
  }
}

Stdio clients can run the npm package instead:

npx -y parkgraph-mcp

An optional Park Graph API key (PARKGRAPH_API_KEY, or Authorization: Bearer on the remote endpoint) adds live availability and pricing and turns on the session tools.

A real exchange

Driver: Need parking near Eagle airport tonight 8pm to midnight, EV.

Assistant calls search_lots, then quote, then reserve, and returns a hold with a confirm link.

The server's instructions tell the model never to invent lots. An empty result means Park Graph has no lot there, and the assistant says so.

Why we built it this way

Gates and pay stations cost $15,000 to $30,000 per lot. A QR sign costs a lamination pouch. Once payments live on a QR page, the same booking flow can be handed to an agent through a few typed tools. The owner sets the rate; the agent just finds it, prices it and holds it.

DEV Community: Wins Parking