I had good results with combining Claude Code with Codex, let them have back and forth sessions. Their prompts were magnitudes better than mine, also their eval

I had good results with combining Claude Code with Codex, let them have back and forth sessions. Their prompts were magnitudes better than mine, also their evaluation and criticism of the other LLM What I haven’t taken time for is finding out about how I‘d automate their back-and-forth and stop manually copy/pasting their responses.