Engineering productivityskill
Caveman skill — terse output style for token savings
Problem solved: Claude Code chat responses burn tokens on prose the operator doesn't read.
What
A community Claude Code skill that forces ultra-terse "caveman" output style, reducing token consumption "dramatically" in long sessions. Quality tradeoff is real — one operator flagged "the question is how it drops evals" — so test before adopting on production tasks.
← All casesC-082