SQL Tool

Business data lives in warehouses and OLTP databases. A SQL tool lets agents answer 'What were Q3 sales by region?' with real numbers, not guesses.

A SQL tool is a translator who speaks both English and database — you ask a question, they write the query, and read you the answer.

Scenario

The agent generates SELECT region, SUM(revenue) FROM sales WHERE quarter='Q3' GROUP BY region and returns a formatted table.

Solution

In Tool Calling & Function Calling, apply SQL Tool to this scenario: The agent generates SELECT region, SUM(revenue) FROM sales WHERE quarter='Q3' GROUP BY region and returns a formatted table. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to SQL Tool (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where SQL Tool happens in the code.

1def run_sql(query, connection):  # define a reusable function2    if not query.strip().upper().startswith("SELECT"):3        raise ValueError("Only SELECT queries allowed")4    with connection.cursor() as cur:5        cur.execute(query)6        columns = [d[0] for d in cur.description]7        rows = cur.fetchmany(100)8    return {"columns": columns, "rows": rows}  # return the result

• Treating SQL Tool as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for sql tool

Quick recap — the most important points from this module.