Understanding the Six Intensity Levels for Caveman Compression

Caveman compression offers six intensity levels—from lite to wenyan-ultra—that control how aggressively LLM output is condensed, with full being the default.

The Caveman project by JuliusBrussee is an LLM output compression system that strips unnecessary filler while preserving technical details and code snippets. The degree of compression is governed by specific intensity levels defined in skills/caveman/SKILL.md and skills/caveman/README.md. These levels determine whether the output remains grammatically complete or collapses into ultra-terse classical Chinese.

What Are Caveman Compression Intensity Levels?

According to the source code in skills/caveman/SKILL.md, intensity levels are persistent session settings that define exactly how aggressively prose is trimmed. The implementation in bin/lib/openclaw.js handles the slash command parsing and injects the appropriate intensity ruleset into the LLM context. Each level specifies which linguistic elements are removed—from simple filler words to entire grammatical structures.

The Six Intensity Levels Explained

Lite

Lite removes filler words and hedging language but retains articles and complete sentences. This produces professional yet concise output where sentences remain grammatically complete and easily readable.

Full (Default)

Full is the default intensity when invoking /caveman without arguments. It drops articles, allows sentence fragments, and substitutes short synonyms. This creates the classic "caveman" style—compact sentences with minimal filler that retain technical accuracy.

Ultra

Ultra strips conjunctions, uses single-word expressions, and avoids any abbreviations or arrows. This represents extreme compression where each fact is stated once using the briefest possible wording while maintaining English language structure.

Wenyan-Lite

Wenyan-lite applies a classical Chinese register (文言文) while keeping relatively light compression. It uses semi-classical phrasing but retains basic grammatical structures, making it accessible to readers familiar with technical Chinese.

Wenyan-Full

Wenyan-full fully adopts classical Chinese style, achieving 80-90% character reduction compared to standard prose. It employs maximum terseness in 文言文 with classical sentence patterns while preserving technical meaning.

Wenyan-Ultra

Wenyan-ultra pushes classical Chinese compression to its absolute limit. This mode produces ultra-terse classical Chinese output representing the most condensed linguistic representation possible within the Caveman system.

How to Switch Between Intensity Levels

The intensity level is persisted for the whole session and can be switched at any time using the /caveman slash command, as documented in commands/caveman.md and implemented in bin/lib/openclaw.js. The command parser accepts the intensity level as an optional argument.

/caveman              # → full (default)

/caveman lite         # → lighter compression

/caveman ultra        # → extreme compression

/caveman wenyan       # → classical Chinese (default wenyan-full)

stop caveman          # → back to normal prose

Practical Examples of Each Intensity Level

When processing the technical question "Why does my React component re-render?", each intensity level produces distinct output as defined in the skill specification:

Intensity	Example Output
lite	Your component re-renders because you create a new object reference each render. Wrap it in useMemo.
full	New object ref each render. Inline object prop = new ref = re-render. Wrap in useMemo.
ultra	Inline obj prop, new ref, re-render. useMemo.
wenyan-lite	組件頻重繪，以每繪新生對象參照故。以 useMemo 包之。
wenyan-full	每繪新生對象參照，故重繪；以 useMemo 包之則免。
wenyan-ultra	新參照則重繪。useMemo 包之。

These examples demonstrate how the same technical information is progressively condensed as the intensity increases, from complete sentences in lite to single-character classical Chinese in wenyan-ultra.

Summary

• Caveman compression provides six distinct intensity levels: lite, full, ultra, wenyan-lite, wenyan-full, and wenyan-ultra
• The default intensity is full, activated by the /caveman command without arguments
• Intensity settings persist for the entire session and are processed by bin/lib/openclaw.js
• Ultra strips conjunctions and abbreviations, while wenyan modes offer classical Chinese compression
• Use stop caveman to restore standard prose output according to commands/caveman.md

Frequently Asked Questions

What is the default intensity level for Caveman compression?

The default intensity level is full, which drops articles and allows sentence fragments while maintaining readability. This default is hardcoded in the command parser within bin/lib/openclaw.js and documented in skills/caveman/README.md.

How do I switch between different intensity levels?

You can switch intensity levels at any time during a session using the /caveman slash command followed by the level name, such as /caveman lite or /caveman ultra. The change is handled by bin/lib/openclaw.js and persists for the entire session until you invoke stop caveman.

What is the difference between ultra and wenyan-ultra intensity?

While ultra compresses English text to single-word expressions and strips conjunctions, wenyan-ultra applies the same extreme compression philosophy to classical Chinese (文言文), achieving the most condensed representation possible in that linguistic register according to skills/caveman/SKILL.md.

How do I disable Caveman compression and return to normal output?

To restore standard prose output, use the command stop caveman. This command immediately disables the compression ruleset for all subsequent responses, bypassing the intensity level settings defined in the skill configuration.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}