Question
A new engineer on your team keeps getting subtly wrong code from the agent — it invents helper functions that already exist, ignores your Result-type error convention and throws instead, and edits a generated file it should never touch. The agent isn't the problem; the context it's given is. As the senior on the team, what would you put in place so any agent run in this repo starts with the right codebase context — conventions, the interfaces it must use, working examples, and an explicit 'do not touch' list?
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.