Prep RoomSign inStart free
Code RoomIterative test fix diagnosis
MediumPrep Room Coding #4346

Iterative test fix diagnosis

Vibe & agenticAlgorithms & data structuresEntry–Mid~12 min

You asked an AI assistant to fix a failing test in your TypeScript code. Its first fix didn't work; its second fix made a different test fail. You're tempted to just keep prompting until something sticks. Walk through how you'd diagnose what's going wrong, and when you'd stop prompting and dig into the code yourself.

Implement
next_ai_fix_action(rounds: list[list[str]]) → str
Examples
in[[["test_parse_amount"],["test_parse_amount"],["test_render_total"]]]out"debug_yourself"
in[[["test_parse_amount"],[]]]out"accept"
in[[["test_parse_amount"],["test_render_total"]]]out"reprompt_with_all_tests"
What a strong answer looks like

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.

0:00 of about 12 min

Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.

Which questions mattered is sealed until you submit. Telling you now would just be handing over the edge cases.