MCP Tool Descriptions
Use the MCP Inspection tools to get all the details about each tool. Here is a summary of the tools.
- browse_folder — Browse and discover dataset contents by navigating the folder hierarchy. Use get_variable_metadata to get full options for a variable.
- calculate — Evaluate arithmetic and return the result.
- create_comparison_table — Compare how different subgroups respond to a single variable, shown side-by-side against all respondents. Use create_table for cross tabulations. Example: compare support for a policy among Men vs Women.
create_table — Create a data table from a Crunch dataset. Auto-detects the preferred table type:
- Frequency: row only. Shows distribution of a single variable.
- Crosstab: row + column. Shows how two variables relate.
- Array: row is a categorical_array. Shows rows × options (no column parameter).
For comparing subgroups side-by-side, use create_comparison_table instead.
- fetch_dataset — Fetch dataset name and description.
- get_variable_metadata — Get full metadata (options, rows, type) for a variable by its alias. Use for variables found via browse_folder. Not needed after search_variables, which returns full details.
- search_variables — Search for variables by concept. Use short queries of 1-2 words (e.g. "gender", "income"). For multiple concepts, make separate calls in parallel. Returns aliases, names, types, descriptions, and options.