Vomit: Clean up Claude 5's token output with a separate LLM

	Vomit: Clean up Claude 5's token output with a separate LLM(github.com)
	295 points by Bluestein 1 day ago \| 290 comments
	tl;dr: Vomit is a local CLI tool that pipes Claude's terse token output through a separate local LLM (e.g., Ollama, Llama.app with GPT-OSS 20B) to translate it into readable English. It integrates via Claude hooks or can run alongside in a non-invasive tail mode, though the author notes it's slow, hallucinates occasionally since the translator LLM only sees Claude's messages, and is Mac-only and "vibe-coded."
	HN Discussion: ↑Claude's verbosity is a genuine problem and workarounds like this are needed but shouldn't be ↑Similar homegrown solutions already exist using cheap models or other tools ~If you need another vendor's model to fix Claude, just switch to that vendor entirely ↓Claude's verbose style reflects careful reasoning and shouldn't be dismissed as hateful vomit •Speculation that Claude's odd style stems from RL optimization for agent-to-agent communication