{
    "action": "tool_call",
    "reasoning": "Profile the categorical columns (company_name, stock_name, and the ingestion metadata columns) in one query: top values, their frequencies, and distinct counts per column.",
    "tool": "query_sql",
    "args": {
        "sql": "WITH u AS (SELECT 'company_name' AS col, company_name AS val FROM raw.semi_conductor_se UNION ALL SELECT 'stock_name', stock_name FROM raw.semi_conductor_se UNION ALL SELECT '_source_system', _source_system FROM raw.semi_conductor_se UNION ALL SELECT '_source_file', _source_file FROM raw.semi_conductor_se UNION ALL SELECT '_batch_id', _batch_id FROM raw.semi_conductor_se), c AS (SELECT col, val, COUNT(*) AS n FROM u GROUP BY col, val), r AS (SELECT col, val, n, ROUND(100.0 * n / SUM(n) OVER (PARTITION BY col), 2) AS pct, COUNT(*) OVER (PARTITION BY col) AS distinct_values, ROW_NUMBER() OVER (PARTITION BY col ORDER BY n DESC, val) AS rk FROM c) SELECT col, val, n, pct, distinct_values FROM r WHERE rk <= 6 ORDER BY col, rk",
        "purpose": "Top values and frequencies for each categorical column",
        "source": "raw"
    }
}