explore-data

The 'explore-data' skill is designed to help users profile and explore datasets to understand their structure, quality, and patterns. It simplifies the process of examining new tables or files, allowing users to quickly identify null rates, column distributions, and potential data quality issues such as duplicates or suspicious values. By providing a comprehensive overview of a dataset before analysis, this skill ensures users have a solid understanding of the data they are working with, reducing errors and improving the efficiency of downstream analytics.

The main features of this skill include automatic table or file detection, column classification, and detailed profiling metrics. Users can explore the dataset at both the table and column levels, including total row and column counts, data types, null rates, cardinality, and common or rare values. Numeric, string, date/timestamp, and boolean columns are each profiled with specialized metrics such as statistical summaries, length and pattern analysis, time distributions, and true/false ratios. The skill also identifies potential data quality issues, including high null rates, unexpected cardinality, negative or future values, and duplicates. The output is presented as a clean summary table grouped by column type, making the insights accessible and actionable.

This skill is particularly useful for data analysts, data engineers, business intelligence professionals, and anyone who needs to familiarize themselves with unfamiliar datasets quickly. It supports files in CSV, Excel, Parquet, and JSON formats, as well as tables from connected data warehouses. Typical use cases include initial data exploration, validation before reporting or analytics, identifying data quality concerns, and guiding which metrics and dimensions to analyze in more detail. By streamlining these tasks, the skill saves time, reduces the risk of overlooked anomalies, and supports informed decision-making based on the dataset's true characteristics.