Sample output
Once you generate, each question lands as a card with the competency tag and a paragraph on what to listen for. Here is the shape, with one hard-coded sample:
Q1 of 1Ownership
Tell me about the last system you owned past v1. What broke first, and what did you do?
Listen for
Strong answers name a specific service, a specific failure mode (latency, integrity, cost), and a fix that mentions trade-offs. Weak answers use adjectives like "really scalable" with no incident, no metric, no number. Probe on dates, on the team they were part of, and on what they would do differently now.