Reduce token usage in AI responses while preserving technical accuracy.
npx skills add https://github.com/jaysoo/dot-ai-config --skill caveman-jaysoo
Please help me install this Agent Skill. Skill: caveman Source: https://github.com/jaysoo/dot-ai-config/tree/main/dot_gemini/skills/caveman Command: npx skills add https://github.com/jaysoo/dot-ai-config --skill caveman-jaysoo
Reduces token usage in AI responses while preserving full technical accuracy, enabling leaner, faster conversations.
Enable caveman mode and respond with terse, technically accurate output.
High-intent search queries and answers about installing and using this skill.
Token optimization through caveman mode reduces AI response verbosity by dropping fillers, articles, and hedging while preserving critical technical content and code formatting for leaner, faster conversations.
Caveman mode is a prompt engineering technique that reduces token consumption by adopting a terse, caveman-speaking style. It drops unnecessary fluff while preserving full technical accuracy, enabling leaner, faster conversations across interactive prompts and long-form sessions.
You can control brevity intensity using configurable modes: lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra. These modes adjust verbosity and style to suit your token efficiency needs.
Terse response mode auto-triggers on phrases like "caveman mode" or "be brief," or when token efficiency is requested. It can also be manually overridden with a command to ensure deterministic behavior throughout long-form sessions.
No, token optimization does not affect code blocks or critical technical formatting. While it drops unnecessary fluff and hedging from explanations, it explicitly preserves critical technical content and code formatting to maintain accuracy.
You should avoid brevity modes when responses require extensive explanatory context, conversational hedging, or formal article structure. The modes are designed for terse, precise tech talk and may strip necessary nuance from non-technical discussions.
sakamoto-family-smileOffer selectable token budgets to balance Claude's response length before answering.
zangjeicyLint and export Google DESIGN.md design tokens to Tailwind theme JSON.
calevansEnforces voice, tone, formatting, and structure for technical documentation files.
rlagycksEstimate prompt token counts and suggest response detail levels.
