Connection reset diagnosis
A Go service throws intermittent 'connection reset by peer' errors under load. You paste the stack traces and a code snippet into an AI agent and it concludes with high confidence: "Root cause: you're not setting a read timeout on the HTTP client, so slow upstreams pile up connections — add client.Timeout = 30s and it's fixed." The fix is plausible and easy. Before you ship it, how do you verify this is actually the root cause and not a coincidence?
rank_reset_hypothesis(samples: list[str]) → str[["reset|300|40","reset|280|45","ok|10|40","ok|12|42","ok|8|38"]]out"idle_keepalive_reuse"[["reset|10|5000","reset|12|4800","ok|11|100","ok|9|120"]]out"slow_upstream_timeout"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.