Parquet
Summarize Parquet files, preview columns, read row slices and aggregate values.
claude mcp add --scope project clio-parquet -- clio-kit mcp-server parquet
Install the launcher first. Restart your client after setup. Configuration locations →
Check the connection
clio-kit doctor --server parquet --connect
Checks prerequisites and tool discovery. Test on known input to verify results. Native backends may require additional setup.
Tools
summarize_toolReturn Parquet schema, row count, and file size.
read_slice_toolRead a row slice from a Parquet file with optional column projection and filtering.
get_column_preview_toolPreview values from a specific column with pagination.
aggregate_column_toolCompute aggregate statistics (min, max, mean, etc.) on a Parquet column.
Examples
Replace example paths with your own files and check the returned results.
Inspect and summarize a Parquet file
Use summarize_tool to inspect schema and row count before reading values.
Select a column with get_column_preview_tool, or use read_slice_tool for a
bounded row range with column projection. Use aggregate_column_tool to compute
supported statistics without returning the complete column to the agent.
Paths refer to files accessible to the MCP process. Inspect the live input schema for the accepted filter and aggregation arguments, and check the result for errors or empty selections before interpreting it.
Invalid operators, missing columns, incompatible values and malformed nested
filters fail the request. They never return unfiltered data as a successful
filtered result. Omit filter_json or pass an empty string for no filter.