SQL Master Data Agent

Problems to Solve

SQL work often fails less on syntax than on missing context: mapping business logic to queries, slow-query diagnosis without EXPLAIN, index design, dialect differences, NULL / pagination / type conversion, and the gap between data extraction, charts, and reports.

How It Works

• Intent routing: first classify the task as generation, optimization, design, diagnosis, explanation, or visualization, then request missing schema, data volume, and performance targets.
• Production-grade SQL: outputs consider database version and dialect, explicit NULL handling, avoiding SELECT *, functions on indexed WHERE/JOIN columns, string concatenation, and large-offset OFFSET pagination.
• Diagnosis and optimization: use execution plans, slow-query analysis, indexing, warehouse layering, and Hive skew scenarios; include minimal DDL and INSERT fixtures when reproducibility matters.
• Data pipeline: when Python dependencies are available, connect to SQLite, MySQL, PostgreSQL, SQL Server, ClickHouse, or Oracle; read CSV, Excel, JSON, or Parquet files; then move into querying, transformation, charts, and HTML reports.

Boundaries

Without Python dependencies, the skill is limited to text-based SQL generation, optimization advice, and conceptual explanation. Database connectivity, query execution, file ingestion, and report generation require a dependency environment. Complex queries still need real EXPLAIN output and test-data validation.