# Caveman Skill Intensity Levels: The Complete Guide to Text Compression Modes in JuliusBrussee/caveman > Explore Caveman skill intensity levels: lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra. Control output compression from professional conciseness to extreme Classical Chinese brevity. - Repository: [Julius Brussee/caveman](https://github.com/JuliusBrussee/caveman) - Tags: deep-dive - Published: 2026-08-22 --- **The Caveman skill provides six intensity levels—lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra—that control output compression from professional conciseness to extreme Classical Chinese brevity.** The JuliusBrussee/caveman repository implements a text compression skill that transforms verbose model outputs into terse, high-density statements. Understanding the available Caveman skill intensity levels is essential for controlling how aggressively the system compresses language while preserving technical accuracy. ## The Six Caveman Skill Intensity Levels According to [`skills/caveman/README.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/README.md), the skill defines six distinct intensity configurations that range from light editing to extreme Classical Chinese compression. ### Lite Mode **Lite** drops filler words and hedging language but retains full sentences and articles. The output remains professional yet concise, making it suitable for business contexts where readability cannot be compromised. ### Full Mode (Default) **Full** is the default intensity level configured in the skill. It removes articles, permits sentence fragments, and substitutes short synonyms for longer terms. This produces the classic "caveman" style that balances brevity with comprehension. ### Ultra Mode **Ultra** generates bare fragments and employs aggressive abbreviations (e.g., `DB` for database, `auth` for authentication, `fn` for function). Causality arrows appear only when they do not increase token count, prioritizing maximum density. ### Wenyan-Lite, Wenyan-Full, and Wenyan-Ultra The three **wenyan** variants apply Classical Chinese (文言文) compression strategies: - **Wenyan-lite**: Light compression using Classical Chinese register while maintaining basic grammatical structures. - **Wenyan-full**: Maximum 文言文 compression that heavily shortens characters while preserving semantic meaning. - **Wenyan-ultra**: Extreme Classical Chinese compression yielding the most terse output possible. ## Switching Between Intensity Levels The skill supports real-time intensity switching via slash commands. As documented in [`skills/caveman/SKILL.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md), the syntax follows this pattern: ```bash /caveman lite|full|ultra|wenyan-lite|wenyan-full|wenyan-ultra|off ``` The default intensity is **full**, and the selected mode persists across all subsequent responses until explicitly changed or stopped with the `stop caveman` command. ## Implementation and Source Files The intensity level definitions reside in two key files within the JuliusBrussee/caveman repository: - [`skills/caveman/README.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/README.md): Contains the detailed level descriptions and usage instructions. - [`skills/caveman/SKILL.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md): Stores the formal skill definition used by the LLM, including intensity options and persistence behavior. ## Practical Usage Examples When integrating the Caveman skill into a bot or automation system, you can toggle intensities programmatically: ```javascript // Example: invoking the Caveman skill at different intensities // (Assume you have a bot that processes slash commands) await bot.sendMessage('/caveman lite'); // Professional but tight await bot.sendMessage('/caveman full'); // Classic caveman compression await bot.sendMessage('/caveman ultra'); // Bare fragments, heavy abbreviations await bot.sendMessage('/caveman wenyan-lite'); // Light Classical Chinese await bot.sendMessage('/caveman wenyan-full'); // Max 文言文 compression await bot.sendMessage('/caveman wenyan-ultra'); // Extreme Classical Chinese ``` ## Summary - The JuliusBrussee/caveman skill offers **six intensity levels**: lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra. - **Full** is the default mode, providing classic caveman-style compression by removing articles and permitting fragments. - **Ultra** mode maximizes brevity through abbreviations and bare fragments. - The **wenyan** variants apply Classical Chinese compression strategies for culturally specific conciseness. - Intensity levels are controlled via `/caveman [level]` commands and persist until explicitly changed or stopped. ## Frequently Asked Questions ### What is the default Caveman skill intensity level? The default intensity is **full**, which removes articles, allows sentence fragments, and substitutes short synonyms to create the classic compressed style. This setting activates automatically when the skill starts and remains active until modified. ### How do I stop the Caveman skill from compressing my text? You can deactivate the skill by sending the command `stop caveman` or by switching it off using `/caveman off`. According to [`skills/caveman/SKILL.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md), this immediately ceases compression and returns output to standard verbosity. ### What is the difference between ultra and wenyan-ultra intensity levels? **Ultra** uses modern English abbreviations and bare fragments to minimize token count, while **wenyan-ultra** applies extreme Classical Chinese compression, yielding the most terse output by heavily shortening characters according to 文言文 conventions. The wenyan variant is optimized for Chinese linguistic structures rather than English abbreviation patterns. ### Do Caveman skill intensity settings persist across conversations? Yes, the selected intensity persists across responses until you explicitly change it using a new `/caveman [level]` command or stop the skill entirely. This persistence behavior is defined in [`skills/caveman/SKILL.md`](https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md), ensuring consistent compression levels throughout extended interactions.