caveman

Token reduction for code reviews is achieved by compressing natural language into terse phrases that strip filler and hedging. This Skill cuts token usage by roughly 75% while preserving technical accuracy, making it ideal for concise debugging explanations and mobile clients with token limits.

Configuring brevity levels for token-efficient prompting involves selecting from lite, full, ultra, or wenyan variants. You trigger these modes explicitly using commands like /caveman full, and the chosen intensity persists across subsequent chat turns until you manually disable it.

Yes, compression mode automatically suspends terse output for security warnings, destructive confirmations, and clarification requests. This safety-aware exception ensures critical multi-step instructions remain fully readable and are not obscured by extreme brevity.

Yes, token efficiency mode auto-triggers when the assistant detects token-efficiency requests in your chat interactions. This enables you to receive compressed, content-preserving responses automatically without needing to issue manual configuration commands each turn.

You should avoid terse compression when handling security warnings, destructive multi-step confirmations, or complex clarification requests. The Skill automatically bypasses brevity in these critical scenarios to ensure technical accuracy and user safety are fully maintained.