Question
You ask an AI to summarize what this Python order-processing function does before you reuse it elsewhere. The agent gives a tidy three-bullet summary: "(1) validates the order, (2) charges the card, (3) returns a confirmation object." It reads as pure and reusable. What's the risk in that summary, and how do you confirm there isn't a critical side effect it omitted?
def process_order(order): validate(order) charge = gateway.charge(order.card, order.total) db.execute("UPDATE inventory SET qty = qty - %s WHERE sku = %s", (order.qty, order.sku)) cache.invalidate(f"product:{order.sku}") emit_event("order.completed", order.id) return Confirmation(charge.id)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 coding: describe the solution in plain language (or narrate it) and the coach grades your approach. Generating runnable code from your description is coming next.