Finish a rough-but-working change set by running it through two independent AI reviewers — first an iterative `codex review` loop on OpenAI's frontier model, then a final `claude -p` pass on Claude's frontier model — applying only the fixes that stay inside…
Quellsprache: Englisch