Compresses AI responses into terse fragments while preserving technical accuracy.
npx skills add https://github.com/D1ssolve/craft-agents --skill caveman-d1ssolve
Please help me install this Agent Skill. Skill: caveman Source: https://github.com/D1ssolve/craft-agents/tree/main/skills/caveman Command: npx skills add https://github.com/D1ssolve/craft-agents --skill caveman-d1ssolve
< not <=" instead of paragraph-length prose. ## Quick Start Tell the AI to use caveman mode for the rest of this session to make all responses brief and token-efficient.High-intent search queries and answers about installing and using this skill.
Enable caveman mode by saying "caveman mode", "talk like caveman", or "be brief". The AI then drops articles, filler words, and pleasantries while keeping all technical terms, code, and error messages exact, cutting token usage by roughly 75%.
Say "stop caveman" or "normal mode" and the AI immediately reverts to standard full-sentence responses. The mode otherwise persists across every turn of the session without drifting back to verbose output.
Lite removes filler and hedging but keeps full sentences. Full drops articles and uses fragments with short synonyms. Ultra adds abbreviations like DB/auth/config, strips conjunctions, and uses arrows for causality such as "new ref → re-render".
No. Code blocks, commits, pull requests, and quoted error messages remain unchanged and written normally. Only the explanatory prose around them is compressed, so technical content stays exact.
It automatically switches to clear full language for security warnings, irreversible action confirmations like deleting database tables, and multi-step sequences where fragments could be misread. Compression resumes after the critical part is done.
RDCrystalLabCompress AI responses into terse modes while preserving technical accuracy.
MomoDavilukeCompress responses to reduce token usage while preserving technical accuracy.
mscipioMonitor AI model context window usage and warn before token overflow.
JNRCHAYANCompresses assistant responses into terse, technically accurate language.