Data Sncf

What Data Sncf does

Data Sncf exposes SNCF Open Data through three MCP tools. search_datasets finds relevant datasets by keyword and returns identifiers, titles, themes, and record counts. Search coverage includes train schedules, stations, punctuality, ridership, and the rail network.

dataset_info retrieves metadata for a selected dataset, including its fields, schema, themes, and record count. This is useful for discovering the column names before writing a query. The query tool reads records using ODSQL and supports filters, grouped or aggregated results, ordering, and pagination.

The integration is independent and unofficial; it is not affiliated with, endorsed by, or published by SNCF or the upstream data provider.

How it works

A typical workflow starts with search_datasets using a keyword such as a transport topic. Pass a returned dataset ID to dataset_info to inspect the available fields, then use that schema to build a query request. Queries can narrow records with where, select or group data for aggregation, sort results with order_by, and divide larger result sets with limit and offset.

The hosted MCP endpoint is available at https://gateway.pipeworx.io/data-sncf/mcp. It exposes the Data Sncf tools along with shared Pipeworx gateway tools, including routing and discovery functions. The full gateway endpoint at https://gateway.pipeworx.io/mcp provides access to all listed data sources, while ask_pipeworx can route questions across the catalog from either gateway connection.

For direct HTTP use, the gateway provides tool endpoints under /v1/tools/; for example, data_sncf_search_datasets accepts a JSON POST request. The first calls do not require an account.

Setup and configuration

Add the hosted endpoint to an MCP client with a remote server configuration:

{
  "mcpServers": {
    "data-sncf": {
      "url": "https://gateway.pipeworx.io/data-sncf/mcp"
    }
  }
}

The package can also run locally over stdio without a Pipeworx account or gateway round trip:

{
  "mcpServers": {
    "data-sncf": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-data-sncf"]
    }
  }
}

Run npx -y @pipeworx/mcp-data-sncf directly to start the local server. The local process answers MCP initialize, tools/list, and tools/call requests and exposes only the Data Sncf tools. The hosted connection adds shared Pipeworx meta-tools, so its tool list is larger than the local package's list.

The project is licensed under MIT.