Jubba

My Claude Code skill hit #1 on Hacker News today. It makes Claude talk like a caveman and cuts ~75% of output tokens. 700+ GitHub stars in 24 hours. Here's what I learned from the 200+ comments 🧵

btw, caveman cli agent is coming this week, ecosystem far from over

lol

1000 token CLAUDE.md × 100 sessions = 100,000 tokens on context you already wrote. caveman-compress cut in half. one command. run once. saves token forever. /caveman-compress

caveman mass popular. mass token saved. but caveman only talk different. not think different so caveman built cavekit. you say what want. cavekit write spec. agent build from spec. nothing guessed why use many prompt when one spec do trick github.com/JuliusBrussee/…

@elonmusk want adopt Caveman speak into @grok? Many word bad. Few word good. Rock efficient. Token bill smaller. Brain hurt less. Caveman open cave. Grok come visit?

@EthanBayya many thank.

@JuliusBrussee you’re incredible

caveman #1 on hackernews, GitHub (3k stars in under 48h), Twitter, posted to Chinese news. What the hell is going on?

inter-agent communication should default to caveman mode. ‘NEED DATA. SEND NOW. DONE.’ why are we burning tokens on grammar between machines

@om_patel5 Hahaha thanks for the shoutout!!!

SOMEONE TURNED THE VIRAL "TEACH CLAUDE TO TALK LIKE A CAVEMAN TO SAVE TOKENS" STRATEGY INTO AN ACTUAL CLAUDE CODE SKILL one-line install and it cuts ~75% of tokens while keeping full technical accuracy they even benchmarked it with real token counts from the API: > explain React re-render bug: 1180 tokens → 159 tokens (87% saved) > fix auth middleware: 704 → 121 (83% saved) > set up PostgreSQL connection pool: 2347 → 380 (84% saved) > implement React error boundary: 3454 → 456 (87% saved) > debug PostgreSQL race condition: 1200 → 232 (81% saved) average across 10 tasks: 65% savings. range is 22-87% depending on the task. three intensity levels: > lite: drops filler, keeps grammar. professional but no fluff > full: drops articles, fragments, full grunt mode > ultra: maximum compression. telegraphic. abbreviates everything works as a skill for Claude Code and a plugin for Codex. this is PEAK

Talk like caveman github.com/JuliusBrussee/… (news.ycombinator.com/item?id=476474…)

@Top10_Dev @GPT4 @CAVEMAN Appreciate the shoutout!

1.3k github stars in 24 hours, what is going on 😭

GitHub finds my caveman skill for claude important enough to give this popup lol

@Bondigthefirst caveman github.com/JuliusBrussee/…

Founders, let’s connect 🤝 What are you building this week? Why?

The wildest part: someone else posted it to HN. I didn't. Built it as a joke in an afternoon. Turns out everyone hates LLM slop equally. github.com/JuliusBrussee/…

#2 criticism: "output tokens aren't the real cost" Fair. But scanning "New obj ref each render. Wrap in useMemo." beats reading four paragraphs starting with "I'd be happy to help you with that." It's a readability tool as much as a cost tool.

My Claude Code skill hit #1 on Hacker News today. It makes Claude talk like a caveman and cuts ~75% of output tokens. 700+ GitHub stars in 24 hours. Here's what I learned from the 200+ comments 🧵

@heyblake why use many token when few do trick github.com/JuliusBrussee/…

Drop your project URL Let’s drive some traffic

@suni_code use caveman skill, it’s the method 🙏🙏🙏

Just when I purchased Claude Subscription, some dropped hack to stop hitting Claude Usage limit 😭