Compresses AI responses into terse caveman-style prose while preserving technical accuracy.
npx skills add https://github.com/leogurja/dotfiles --skill caveman-leogurja
Please help me install this Agent Skill. Skill: caveman Source: https://github.com/leogurja/dotfiles/tree/main/home/dot_agents/skills/caveman Command: npx skills add https://github.com/leogurja/dotfiles --skill caveman-leogurja
< not <=.", cutting reading time without losing technical detail. ## Quick Start Ask the assistant to talk like caveman or type /caveman to enable compressed responses for the rest of the session.High-intent search queries and answers about installing and using this skill.
Enable caveman mode with /caveman or by asking the assistant to talk like a caveman. It drops articles, filler words, and pleasantries while keeping code, error strings, and technical terms exact, and the style persists for the whole session.
Say "stop caveman" or "normal mode" and the assistant immediately reverts to standard prose. You can also switch intensity at any time with /caveman lite, full, ultra, or one of the wenyan variants.
No. The rules preserve code blocks, exact error strings, numbers, units, and negation words like not and never. An auto-clarity rule also drops compression for security warnings, irreversible actions, and ambiguous multi-step instructions.
Yes. The wenyan-lite, wenyan-full, and wenyan-ultra levels compress replies into Classical Chinese register with increasing terseness. The skill otherwise preserves the user's dominant language and only compresses style, never switching languages.
Avoid it for content read by other humans, such as commit messages, docs, issue reports, and PR descriptions, which the skill itself writes in normal prose. It is also unsuitable when fragment ambiguity could cause misread multi-step procedures.
rbaumierInvoke applicable skills at conversation start to guide AI behavior.
openharmonyinsightNormalize user-described AI usage problems into structured, taggable feedback reports.
m2ai-portfolioDecompose agent session token spend into context-driven and output-driven costs.
yashtekavadeCompress assistant replies into terse, technically accurate text.