Model Context Protocol server for querying statistical datasets with AI assistants.
Add to your Claude Desktop config:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"subsets": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/subsetsio/subsets-mcp-server.git",
"mcp-server",
"--api-key",
"YOUR_API_KEY"
]
}
}
}Get your API key at subsets.io/settings
Requires uv installed.
uvx --from git+https://github.com/subsetsio/subsets-mcp-server.git mcp-server --api-key YOUR_API_KEYSearch and browse available datasets with semantic search.
Parameters:
q(string): Search querylimit(integer): Max results (default: 10)min_score(float): Min relevance score threshold (0.0-2.0)
Example:
list_datasets(q="unemployment europe", limit=5)
Get detailed information about a specific dataset including schema, statistics, and preview.
Parameters:
dataset_id(string): Dataset identifier
Returns: Full metadata, column descriptions, row counts, data preview, and query usage stats
Example:
get_dataset_details("eurostat_unemployment_2024")
Run SQL queries on datasets using DuckDB.
Parameters:
query(string): SQL SELECT statement
Example:
execute_sql_query("SELECT * FROM eurostat_unemployment_2024 LIMIT 10")The package also includes CLI commands for managing local datasets:
# Install globally
npm install -g @subsetsio/mcp-server
# Add datasets to local collection
subsets add eurostat_unemployment_2024
# Download datasets
subsets sync
# List local datasets
subsets list
# View status
subsets status# Clone repository
git clone https://github.com/subsetsio/subsets-mcp-server
cd subsets-mcp-server
# Install dependencies
uv sync
# Run MCP server
uv run python src/server.py --api-key YOUR_API_KEY
# Or run CLI
uv run python src/cli.py --helpFull documentation at subsets.io/docs
MIT