Reduce token usage in technical responses via caveman communication style.
npx skills add https://github.com/figueroaignacio/ui --skill caveman-figueroaignacio
Please help me install this Agent Skill. Skill: caveman Source: https://github.com/figueroaignacio/ui/tree/main/.agents/skills/caveman Command: npx skills add https://github.com/figueroaignacio/ui --skill caveman-figueroaignacio
This feature enables ultra-compact communication in technical contexts by reducing token usage while preserving accuracy, helping users stay concise without losing essential detail.
Change to caveman mode by saying caveman or using /caveman, then respond in terse, technically accurate language.
High-intent search queries and answers about installing and using this skill.
To reduce token usage while preserving accuracy, switch to a terse communication style that trims filler. This mode enforces safety and accuracy while delivering consistently brief technical responses with minimal excess text.
Concise communication mode is triggered by using the /caveman prompt. Once activated, it remains active across sessions for the same chat, ensuring deterministic terseness without needing repeated commands.
Yes, token optimization supports multiple intensity levels: lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra. The default level is full, allowing you to adjust communication terseness based on your specific technical context needs.
Switching to a terse communication style does not affect technical accuracy. The mode explicitly enforces safety and accuracy, trimming only filler while preserving technical content and not deviating from allowed topics.
You should avoid ultra-compact communication when tasks require detailed explanations or extensive context. While it preserves technical accuracy, deterministic terseness strips filler and conversational framing, reducing clarity for non-technical audiences.
JNRCHAYANCompresses assistant responses into terse, technically accurate language.
dakesanDetect scientific data file formats and generate Markdown analysis reports.
musher-devGenerate llms.txt and llms-full.txt documentation indexes for AI assistants and RAG ingestion.
sebc-devCompress responses by removing non-essential words while preserving technical terms.
