Run DuckDB SQL queries over CSV, Parquet, JSON/NDJSON, Excel (.xlsx), and database sources (Postgres, MySQL, SQLite, S3) for data exploration and analytics. Use when the user wants to: (1) Explore, filter, or aggregate data from tabular files, (2) Join multiple datasets, (3) Perform analytical SQL over local files or databases, (4) Get precise numeric/statistical results from large datasets. Prefer this over in-context reasoning for datasets with thousands of rows or complex joins/aggregations.