Caveman Compress

Compresses markdown memory files into caveman-speak to cut token usage while preserving all code blocks, URLs, and technical terms exactly. Overwrites the original with compressed version and saves a readable backup as FILE.original.md. Removes articles, filler words, and verbose phrasing but keeps structure intact. The compression process runs through a Python CLI that validates output and retries on errors. Good for trimming bloated CLAUDE.md files or todo lists without losing technical substance. Works only on natural language files, never touches actual code.

Install to Claude Code

npx -y skills add juliusbrussee/caveman --skill caveman-compress --agent claude-code