mcp-madagascar
# mcp-madagascar
MCP server for Madagascar humanitarian, weather, risk, and open-data source discovery.
## Scope
Madagascar-focused source discovery for humanitarian data, weather/cyclone context, administrative data, and risk indicators.
## Tools
- `madagascar_get_sources`
- `madagascar_search_data_gouv`
- `madagascar_get_dataset`
- `madagascar_fetch_source_excerpt`
- `madagascar_explain_scope`
- `madagascar_list_reference_items`
- `madagascar_search_hdx_datasets`
- `madagascar_get_gdacs_alerts`
- `madagascar_get_weather`
- `madagascar_get_weather_forecast`
## Install
```bash
npm install
npm run build
npm test
npm run dev
```
## Claude Desktop
```json
{
"mcpServers": {
"madagascar": {
"command": "npx",
"args": ["@hug0x0/mcp-madagascar"]
}
}
}
```
## Sources
- HDX Madagascar API search: https://data.humdata.org/api/3/action/package_search?q=Madagascar&rows=1
- HDX Madagascar group: https://data.humdata.org/api/3/action/group_show?id=mdg
- Météo Madagascar climate maproom: https://map.meteomadagascar.mg/index.html
- Open-Meteo API: https://open-meteo.com/
- GDACS RSS: https://www.gdacs.org/xml/rss.xml
## Example Prompts
- "Find humanitarian datasets for Madagascar."
- "List official or widely used Madagascar risk-data sources."
- "Get a source excerpt from the climate maproom."
## Safety
This MCP helps agents discover and summarize public sources. It is not an official authority. For emergency, legal, or administrative decisions, follow the competent public service.
## Glama / Docker
The repo includes `Dockerfile` and `glama.json`.
Publishing notes: [`docs/publishing.md`](docs/publishing.md).
Build steps:
```json
["npm install", "npm run build"]
```
CMD arguments:
```json
["node", "dist/index.js"]
```
## License
MIT
TDQS
Scored across 9 tools
Most tools have clearly distinct purposes, with clear separation between searching different portals (HDX vs data.gouv.fr), listing curated sources, fetching excerpts, and retrieving alerts/weather. Minor confusion could arise between get_sources and list_reference_items, but their descriptions clarify the difference.
All tool names follow a consistent 'madagascar_<verb>_<object>' pattern in snake_case. Verbs like search, get, fetch, explain, and list are used predictably, and the shared madagascar_ prefix makes the family identity strong.
Nine tools is well-scoped for a country-specific data discovery and retrieval server. Each tool serves a distinguishable purpose without redundancy, and the count feels appropriate for the domain rather than minimal or bloated.
The server covers the main workflows: searching multiple data sources, inspecting datasets, fetching excerpts from curated sources, and retrieving live alerts/weather. A notable gap is the absence of an HDX dataset detail endpoint, leaving only search for that portal, but most core exploration needs are met.