Vomit

Vomit converts Claude's token vomit into English by piping it through a local LLM. It's fully local (no telemetry) and has no external dependencies.

I wrote a small blog post about this project: https://zachahn.com/posts/1787191554

Disclaimer:

• The local LLM can only see what Claude tries to communicate (no access to any actions or files), so it hallucinates a bit
• It's pretty slow
• This is totally vibe-coded, only tested on Mac
• There is a possibility you'll completely miss Claude's message. You can use something like AgentsView to get the original messages, as Vomit does not touch anything during runtime (except technically it writes files to your TMPDIR)

Install

# Install the binary to your GOPATH
go install github.com/zachahn/vomit@latest

# Setup connection details to your LLM
vomit init

# Instructions on how to replace Claude's output via hooks
vomit scrub -claude

Usage

In addition, there's a non-invasive mode if you want to run Vomit on the side.

• vomit list. List Claude session identifiers
• vomit tail [<session_identifier>]. Translate Claude's tokens for the specified session, or follow the latest one
• vomit help. See for more commands

This should work with:

• Llama.app
• Ollama
• Anything that uses the OpenAI API?

If you don't have a local LLM set up already, I think I recommend using Llama.app, then downloading GPT-OSS 20B through it. Run the init subcommand after you set this up.

License

GNU GPLv3