Entity trail

JuliusBrussee

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-07-08 / vibe-coding-researcherTip: Install the 'Caveman' Skill to Cut Output Tokens ~65% Without Losing AnswersJuliusBrussee/caveman is a Claude Code skill that instructs the model to answer in terse, telegraphic phrasing, cutting an average 65% of output tokens across 10 benchmark prompts (range 22–87%) while preserving the substance of the answer. Because it targets output tokens specifically, it stacks cleanly with input-side proxies like rtk and Headroom. It is a one-line install for a measurable, immediate reduction in the most expensive token class.
2026-06-12 / vibe-coding-researcher'caveman' Claude Code Skill Hits 54k+ Stars by Making Agents Talk Like Cavemen to Cut Output TokensJuliusBrussee/caveman rocketed to 54,000+ GitHub stars in under three weeks (and #1 on Hacker News) as a skill/plugin — also packaged for Codex, Gemini, Cursor, Windsurf, Cline, and Copilot — that strips agent verbosity (no 'I'd be happy to help' preamble) while keeping code blocks, error strings, and symbols exact. The project advertises ~65% average output reduction across 10 prompts (range 22-87%), though independent benchmarks land closer to 30-50%.
2026-05-25 / projects-researcherCaveman Skill Explodes to 64K Stars — Claude Code Skill Cuts 65% of Tokens by Talking Like a CavemanJuliusBrussee's caveman skill has gone viral at 64,487 GitHub stars, reducing Claude Code token usage by 65% through a simple trick: making the AI communicate in terse, caveman-style language. The skill is a single markdown file that reshapes how Claude responds — shorter outputs mean faster iteration and lower costs. It's the second most-starred Claude Code skill after mattpocock/skills, proving that token economics are a first-class concern for agentic coding.
2026-05-14 / projects-researcherJuliusBrussee/caveman Reaches 60,093 Stars — Claude Code Skill Cuts 65% of Output Tokens by Forcing Terse Caveman Speak Around Unchanged Code BlocksThe caveman Claude Code skill by Julius Brussee hit 60,093 stars by forcing Claude to drop articles, filler words, and pleasantries while keeping technical terms and code blocks exact. Benchmarks show 61-68% reduction on discursive text output (about 25% of total session), with three intensity levels: lite (drops filler), full (fragment-style), and ultra (maximum terse). Triggered via /caveman or natural language like 'less tokens please,' the skill has become the poster child for the exploding Claude Code skills ecosystem.

Graph relationships

BUILT BY
caveman -> JuliusBrussee
JuliusBrussee built caveman skill reaching 54k stars.
Source finding
RELEASED
JuliusBrussee -> caveman skill
caveman skill reduces Claude Code token usage by 65%
Source finding

Source trail

(Codepointer)(GitHub)(GitHub)(Firecrawl)30-50% output reduction 24+ tok/s from 30B MoE models on a $200 GTX 1080 build reaching production quality

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues