Tools
claudish-to-english rewrites the agent's own output into plain language on screen, 579 stars on day one
Leutenegger/claudish-to-english was created and pushed on 2026-08-21 and reached 579 stars and 61 forks within a day, MIT, Shell. It is a Claude Code plugin that pipes each assistant message through a rewrite before display, defaulting to a local ollama model and optionally using the codex CLI, Anthropic, or any OpenAI-compatible endpoint, with an optional Markdown rewrite. It is fail-open and screen-only, so it does not touch the transcript the agent reasons over. Same-day sibling repos with the same complaint (adnanakil/nobuzz at 146 stars, LB623/no-negative-echo at 150) suggest this is a category, not one person's itch.
Source
↳ Follow the thread