JuliusBrussee's Caveman Plugin Reduces AI Token Usage by 65%
JuliusBrussee has developed a plugin called Caveman, designed to significantly reduce the number of tokens used by AI coding agents. The plugin is compatible with various AI models such as Claude Code, Codex, Gemini, and others. By adopting a 'caveman-speak' style, the plugin cuts down on unnecessary words, maintaining the technical accuracy of responses while reducing token usage by an average of 65% for prose and 8.5% for coding tasks. The plugin is easy to install across multiple platforms and can be activated or deactivated with simple commands. It offers different levels of compression, allowing users to choose the degree of brevity in responses. 
