Skip to content
Back to search

caveman-distillate

SkillWorks with Claude CodeGitHub

Caveman Distillate — ultra-compressed AI output mode. Distill text to its core. ~90 tokens, ~65% output savings. Claude Code skill.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/dlepold--caveman-distillate/badge.svg)](/listings/dlepold--caveman-distillate)

HTML

<a href="/listings/dlepold--caveman-distillate"><img src="/api/listings/dlepold--caveman-distillate/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

npx skills add dlepold/caveman-distillate

Or manually: copy `skills/caveman-distillate/SKILL.md` into your `.claude/skills/` directory.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 62 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Caveman Distillate
  2. > Why use many token when few token do trick?
  3. **Caveman Distillate** is a Claude Code skill that distills AI output to its core — cutting ~65% of output tokens while keeping full technical accuracy.
  4. Inspired by [JuliusBrussee/caveman](https://github.com/JuliusBrussee/caveman), distilled down to ~90 tokens of system prompt (vs ~450 in the original). Same effect, less overhead.
  5. ## What It Does
  6. Strips articles, filler words, pleasantries, and hedging from Claude's responses. Code blocks, error messages, commits, and security warnings stay untouched.
  7. **Before (69 tokens):**
  8. > The reason your React component is re-rendering is likely because you're creating a new object reference on each render cycle. When you pass an inline object as a prop, React's shallow comparison sees it as a different object every time, which triggers a re-render. I'd recommend using useMemo to memoize the object.
  9. **After (19 tokens):**
  10. > New object ref each render. Inline object prop = new ref = re-render. Wrap in `useMemo`.
  11. ## Install
  12. ```bash
  13. npx skills add dlepold/caveman-distillate
  14. ```
  15. Or manually: copy `skills/caveman-distillate/SKILL.md` into your `.claude/skills/` directory.
  16. ## Usage
  17. **Activate:** Say any of these:
  18. - `caveman`, `terse`, `kurz`, `kurz&knapp`, `kurz bitte`, `less tokens`, `weniger text`
  19. **Deactivate:**
  20. - `normal`, `normal mode`, `verbose`, `ausführlich`
  21. ## Multilingual Triggers
  22. Supports both English and German activation — because developers aren't all English-speaking.
  23. | Language | ON | OFF |
  24. |----------|-----|-----|
  25. | English | terse, caveman, less tokens | normal, verbose |
  26. | German | kurz, kurz&knapp, kurz bitte, weniger text | ausführlich, normal mode |
  27. ## How It Differs From the Original
  28. | | [caveman](https://github.com/JuliusBrussee/caveman) | caveman-distillate |
  29. |---|---|---|
  30. | System prompt size | ~450 tokens | **~90 tokens** |
  31. | Intensity levels | 3 (lite/full/ultra) | 1 (full — the useful one) |
  32. | Multilingual triggers | English only | English + German |
  33. | Plugin/marketplace files | Yes | No (just the skill) |
  34. | Effect | Same | Same |
  35. The three intensity levels are unnecessary complexity — "full" is what you want 95% of the time. If you need verbose output, just say "normal mode."
  36. ## Philosophy
  37. The best compression is compressing the compressor itself. A 450-token prompt that tells you to be brief is ironic. This is the distillate — what remains after boiling away everything unnecessary.
  38. ## License
  39. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.