Skip to main content
Glama
README.md
# aemet-mcp

MCP server that exposes public weather data from [AEMET](https://www.aemet.es)
(Agencia Estatal de Meteorología, Spain) to AI agents and MCP-compatible
clients.

Three tools, well-typed, backed by AEMET's official OpenData API:

- **`get_municipality_forecast`** — daily forecast (up to 7 days) for a
  Spanish municipality, by INE code.
- **`get_station_observation`** — last ~12h of hourly observation data
  (temperature, wind, precipitation…) from a weather station.
- **`get_weather_warnings`** — active weather warnings (CAP alerts) for a
  region or all of Spain.

## Prerequisites

- Node.js 20+
- A free AEMET OpenData API key: request one at
  https://opendata.aemet.es/centrodedescargas/altaUsuario

## Installation

No install needed — run it directly with `npx`:

```bash
npx -y @mmillan76/aemet-mcp
```

The server communicates over stdio, so it's meant to be launched by an MCP
client rather than run standalone in a terminal.

## Configuration

The server reads the API key from the `AEMET_API_KEY` environment variable.
It exits with an error at startup if it's missing.

### Claude Desktop / Claude Code

Add to your MCP client config (e.g. `claude_desktop_config.json`):

```json
{
  "mcpServers": {
    "aemet": {
      "command": "npx",
      "args": ["-y", "@mmillan76/aemet-mcp"],
      "env": {
        "AEMET_API_KEY": "your-api-key-here"
      }
    }
  }
}
```

Any other MCP client that supports stdio servers works the same way — point
it at `npx -y @mmillan76/aemet-mcp` with `AEMET_API_KEY` set in the environment.

## Local development

```bash
git clone https://github.com/mmillan76/aemet-mcp.git
cd aemet-mcp
npm install
cp .env.example .env   # fill in AEMET_API_KEY
npm run build
```

Test tool calls interactively with the
[MCP Inspector](https://github.com/modelcontextprotocol/inspector):

```bash
npx @modelcontextprotocol/inspector node dist/index.js
```

(the Inspector reads `AEMET_API_KEY` from your shell environment, so export
it — or source `.env` — before running the command above)

## Notes on the AEMET API

AEMET OpenData uses a two-step pattern on every endpoint: the first request
returns a JSON envelope with a `datos` URL, not the actual data; a second
request to that URL returns the real payload. This server hides that detail
— tools just return the final data. Rate limits apply per API key; if you
hit them, the tool returns a clear error instead of retrying silently.

## License

MIT — see [LICENSE](./LICENSE).

TDQS

A4.2/5.0

Scored across 3 tools

Disambiguation5/5

Each tool targets a clearly distinct data type: forecasts for municipalities, observations from stations, and weather warnings. No overlap in purpose or resource.

Naming Consistency5/5

All tool names follow the identical pattern 'get_' followed by a specific data type (municipality_forecast, station_observation, weather_warnings). This is consistent and predictable.

Tool Count5/5

Three tools is well-scoped for a weather data server, covering the core data types without unnecessary bloat. Each tool has a clear and distinct purpose.

Completeness4/5

The tool set covers the primary weather data types (forecast, observation, warnings) for the AEMET domain. Minor gaps might include historical data or air quality, but the core surface is complete for typical use.

Maintenance

ActivitySlowing
ResponsivenessNo issues