semantic-sql-history-profiler

What problem does it solve?

Manually authoring semantic models requires guessing column business meanings, join relationships, and valid value ranges, leading to inaccurate SQL generation, broken metrics, and hours of redundant schema inspection for data teams.

Core Features & Use Cases

• Historical SQL Mining: Extracts join relationships, dimension candidates, measure candidates, and common filter templates directly from existing successful SQL queries instead of relying on guesswork.
• Bounded Column Profiling: Offers three profiling modes (sql_only for speed, lightweight for sampled distributions, deep for exhaustive analysis) with configurable resource limits to avoid overloading production systems.
• Use Case: A data team building a MetricFlow semantic model for sales data can use this skill to profile 6 months of historical sales queries, automatically identifying that order_status is a dimension with 4 distinct values and order_total is a measure with a valid range of $10 to $10,000, cutting model authoring time by 70%.

Quick Start

Use the semantic-sql-history-profiler skill to analyze the provided historical e-commerce SQL queries and generate profiling evidence for your new sales semantic model.