Verifying AI-generated SQL migrations
An AI agent generated a complex SQL migration that backfills a new denormalized column across a 400M-row table and rewrites three reporting queries to use it. You don't deeply understand one of the window-function rewrites, and the agent assures you it's equivalent and faster. It would be a multi-hour, hard-to-reverse migration on a production analytics warehouse. How do you verify this before running it, without just trusting the agent?
diff_result_sets(baseline_rows: list[str], candidate_rows: list[str]) → list[str][["2026-01-01|100","2026-01-02|250","2026-01-03|75"],["2026-01-01|100","2026-01-02|250","2026-01-03|75"]]out[][["a|1","b|2","c|3"],["a|1","b|9","d|3"]]out["extra:d","missing:c","value:b:2:9"]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.