Datos.gob.es-MCP
# Datos.gob.es-MCP. MCP integration with the Spanish Government Open Data Portal
[](README.md)
[](README_es.md)
[](https://mseep.ai/app/ancode666-datos-gob-es-mcp)
[](https://mseep.ai/app/ancode666-datos-gob-es-mcp)
**Datos.gob.es-mcp** enables querying and analyzing over 90,000 public datasets available on the [datos.gob.es](https://datos.gob.es/es/) portal directly from Claude AI and other compatible MCP clients using the **Model Context Protocol (MCP)**.
This MCP server exposes tools for LLMs to search, filter, and access open data across multiple sectors.
## Main features
- **Keyword search** across dataset titles, descriptions and tags.
- **Thematic category filtering** (environment, transportation, education, etc.)
- **Detailed metadata access** for each dataset.
- **Available distributions** listing (formats and access URLs)
- **Custom SPARQL queries** execution against the official SPARQL endpoint.
## Installation
### Install via uv
### Prerequisites
- Python 3.10 or higher
- [uv](https://docs.astral.sh/uv/getting-started/installation/) package manager
### uv Installation
First install `uv`, a modern Python package manager.
**Install from command line**:
En MAC y Linux:
```bash
curl -LsSf https://astral.sh/uv/install.sh | sh
```
En Windows:
```bash
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
```
También se puede instalar con pip:
```bash
pip install uv
```
For more information about installing **uv**, visit the [uv documentation](https://docs.astral.sh/uv/getting-started/installation/).
## Integration with clients like Claude for Desktop
Once **uv** is installed, you can use the MCP server with any compatible client like Claude Desktop. Configuration steps:
1. Go to **Claude > Settings > Developer > Edit Config > `claude_desktop_config.json`**
2. Add this configuration block under `"mcpServers"`:
```json
"datos_gob_es_mcp": {
"command": "uvx",
"args": [
"datos_gob_es_mcp"
]
}
```
3. If you have other MCP servers configured, separate them with commas `,`
For other MCP-compatible clients like Cursor, CODEGPT or Roo Code, add the same configuration block to their respective MCP server settings.
## Usage Examples
Once properly configured, you can request operations like:
```text
- "Search for public transportation datasets in Madrid"
- "List latest datasets published by Barcelona City Council"
- "Show details for dataset with URI https://datos.gob.es/es/catalogo/l01330241-padron-de-vehiculos-ano-2023-autobuses"
---
TDQS
Scored across 10 tools
Each tool has a clearly distinct purpose: listing, searching, filtering by theme/publisher, getting details/distributions, and running SPARQL queries. There is no overlap that could cause confusion.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., list_datasets, get_dataset_details). The only exception, run_custom_sparql, still fits the verb_noun style and is clear.
10 tools is well-scoped for the domain of accessing open government data, covering all essential operations without being excessive or too few.
The tool set covers dataset discovery, filtering, details, and distributions, plus listing publishers and themes. A custom SPARQL endpoint adds flexibility. Minor gaps include no drill-down into publisher/theme details, but core workflows are supported.