# Weather MCP Server

An [MCP](https://modelcontextprotocol.io/) server that exposes US weather data from the [National Weather Service](https://www.weather.gov/documentation/services-web-api) as tools for Cursor (or any MCP client).

## Tools

| Tool | Arguments | Description |
| --- | --- | --- |
| `get_alerts` | `state` (e.g. `CA`, `NY`) | Active weather alerts for a US state |
| `get_forecast` | `latitude`, `longitude` | Next five forecast periods for a US location |

NWS coverage is US-only. Use a two-letter state code for alerts and US coordinates for forecasts.

## Requirements

- [uv](https://docs.astral.sh/uv/)
- Python 3.14+ (this project pins `>=3.14`)

## Setup

```powershell
git clone <your-repo-url>
cd weather
uv sync
```

## Run

The server speaks MCP over stdio:

```powershell
uv run weather.py
```

## Use with Cursor

1. Copy [`.cursor/mcp.json.example`](.cursor/mcp.json.example) to `.cursor/mcp.json`.
2. Replace the `uv` path and `--directory` path with absolute paths on your machine.
3. Open **Cursor Settings → MCP**, enable the `weather` server, and reload the window if it does not connect.

Example prompts after it is connected:

- Get weather alerts for CA
- What’s the forecast for 37.7749, -122.4194?

On Windows, `uv` is often not on Cursor’s PATH, so use the full path from `where.exe uv`. Keep `--directory` as an absolute project path.

## Project layout

```
weather.py          # MCP server (tools + NWS client)
src/weather/        # Package entrypoint used by `uv run weather`
pyproject.toml      # Dependencies (`mcp[cli]`)
.cursor/            # Cursor MCP config (local `mcp.json` is gitignored)
```