gvzdv/claudish-to-english ? reverse-engineered prompt

Build me a Claude Code plugin that rewrites each assistant reply into plain English so the answers are easier to skim, but keeps the original message underneath or safely available if needed.

I want it to work like a display only layer, so it changes what I read on screen without changing the actual transcript or reasoning. It should be able to use a local model by default, and also let me switch to other providers if a local one is not available. If rewriting fails for any reason, it should just show the original text and not break the session.

Please make it smart about language too, so if Claude is replying in another language, the rewrite stays in that language. I’d also like an optional setting that rewrites Markdown files into simpler language when they are saved, but keep that turned off unless I enable it.

If you need to check current docs or command details online, go ahead.