I also find it near-impossible to reliably get Claude to write text the way I want it to without continually reverting back to Claudespeak overwrought word salad. Plus the same sort of hedging/over-anticipation of future hypothetical edge cases, etc.

It's been much more reliable to just have Codex rewrite all the prose at the end of a Claude task than to keep futzing with my prompts/memory/agents.md (On certain issues I can't get Codex to be as thorough on the actual implementation, c'est la vie.)