I'll investigate data quality issues systematically by examining the quality_issues table and profiling key columns for outliers, impossible values, and inconsistencies. ```json { "action": "tool_call", "reasoning": "Start by checking the quality_issues table which likely catalogs known data quality problems in this dataset.", "tool": "query_sql", "args": { "sql": "SELECT issue_type, severity, COUNT(*) as issue_count, COUNT(DISTINCT table_name) as affected_tables FROM raw.quality_issues GROUP BY issue_type, severity ORDER BY issue_count DESC", "purpose": "Get overview of cataloged quality issues by type and severity" } } ```