Rocky-Project Hail Mary agent skill that cut output tokens ~47%
Persona-based prompting cuts tokens 47% without breaking code like Caveman styles do.
Agentic Coding for Builders who Ship
Caveman mode cuts tokens 70% while Claude Code alternatives charge monthly.
Developers using AI coding assistants in terminal
Claude Code · Aider · Continue
Inspired by this HN post [1] I saw a few hours ago about caveman speak for token-efficient output, I added /Caveman and /Rocky (from project Hail Mary) to Claurst.
Caveman grammar drops ~70% of output tokens keeping technical substance and Rocky brings a close compression with more personality. Both are first-class slash commands with three levels: lite, full, ultra.
It's fun, makes you happy and saves tokens :D
Would love for you to try it out and get feedback!
Persona-based prompting cuts tokens 47% without breaking code like Caveman styles do.
4x token savings on screenshots with readable text at 800px grey.
Rotates Claude OAuth tokens with auth precedence handling before API key fallback.
Hash-verified edits cut Claude editing tokens by 44%—measurable, specific, real win.
Transparent benchmarks show 39% cost cuts — rare to see real numbers in AI tooling.
Claude Code hooks that cut token costs, but only matters if you're burning through CLI agent budgets.