# jiskta-mcp

[![PyPI](https://img.shields.io/pypi/v/jiskta-mcp?label=PyPI&color=3b82f6)](https://pypi.org/project/jiskta-mcp/)
[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)

MCP server for the [Jiskta Climate & Environmental Data API](https://jiskta.com). Gives AI assistants direct access to historical air quality, ERA5 meteorology, water risk, geocoding, and industrial facility data — without writing any code.

## What is MCP?

[Model Context Protocol](https://modelcontextprotocol.io) is an open standard that lets AI tools (Claude Desktop, Cursor, Cline, Copilot Workspace) call external APIs in conversation. This package runs **on your machine**, not on any server — it just proxies calls to the Jiskta API using your API key.

## Quickstart

**Get an API key**: [jiskta.com/dashboard](https://jiskta.com/dashboard) — free credits included on signup.

### Claude Desktop

Add to `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):

```json
{
  "mcpServers": {
    "jiskta": {
      "command": "uvx",
      "args": ["jiskta-mcp"],
      "env": {
        "JISKTA_API_KEY": "sk_live_your_key_here"
      }
    }
  }
}
```

No Python installation needed — [`uvx`](https://docs.astral.sh/uv/) downloads and runs the package automatically.

### Cursor / Cline / other MCP clients

Use the same config format your tool supports. The command is always `uvx jiskta-mcp` with `JISKTA_API_KEY` set.

### Manual install

```bash
pip install jiskta-mcp
export JISKTA_API_KEY=sk_live_...
jiskta-mcp
```

## Available tools

| Tool | Description |
|------|-------------|
| `query_climate` | Historical air quality + ERA5 met data for a bounding box |
| `query_climate_point` | Same, for a single lat/lon point |
| `estimate_query_cost` | Dry-run — check credit cost before querying |
| `geocode` | Address → coordinates (113M housenumbers, global) |
| `reverse_geocode` | Coordinates → nearest address |
| `enrich_location` | Coordinate → NUTS3 region + WRI water risk + nearest facility |
| `water_risk` | WRI Aqueduct 4.0 water risk for a bounding box |
| `find_facilities` | Nearest E-PRTR industrial facilities within a radius |
| `get_coverage` | Check available data months before querying |
| `spatial_link` | Aggregate raster data to NUTS3 regions or countries |

## Example conversations

Once connected, you can ask things like:

> *"What was the average NO₂ level in Brussels in 2023?"*

> *"Is there any industrial pollution risk near Industrieweg 1, Antwerp?"*

> *"Compare PM2.5 trends across NUTS3 regions in the Ruhr area from 2018 to 2023."*

> *"What's the water stress level for our distribution centre at 51.5°N, 4.3°E?"*

> *"How many days did PM10 exceed the EU limit in Paris last year?"*

## Data sources

- **Air quality**: Copernicus CAMS EU reanalysis (0.1°, 2013–present) and CAMS Global (0.75°, 2020–present)
- **Meteorology**: ECMWF ERA5 reanalysis (0.25°, 2013–present)
- **Water risk**: WRI Aqueduct 4.0 (2023), global
- **Industrial facilities**: EEA E-PRTR, ~97,000 EU facilities (CC BY 4.0)
- **Geocoding**: OpenStreetMap (CC BY-SA)

## Pricing

Credits are consumed per tile scanned (geographic area × time period × variable). Use `estimate_query_cost` before large queries. Top up at [jiskta.com/dashboard](https://jiskta.com/dashboard).

## Links

- API documentation: [jiskta.com/docs](https://jiskta.com/docs)
- Python SDK: [github.com/jiskta/jiskta-python](https://github.com/jiskta/jiskta-python)
- Examples: [github.com/jiskta/jiskta-examples](https://github.com/jiskta/jiskta-examples)
- Issues: [github.com/jiskta/jiskta-mcp/issues](https://github.com/jiskta/jiskta-mcp/issues)