Caveman
ProductivityWhy use many token when few token do trick. Cuts ~65% of output tokens.
Install
$ /plugin marketplace add JuliusBrussee/caveman
Update
$ /plugin marketplace update caveman
Then run /reload-plugins. Official marketplaces auto-update at startup.
When to use it
- Long working sessions where output token cost adds up
- You read answers fast and do not need pleasantries
- Pair-programming flow where terse status updates beat prose
Commands & flags
- /caveman
- Activate caveman mode at full (default) intensity.
- /caveman full
- Set intensity to full; other levels are lite and ultra.
- "caveman mode"
- Activate by saying caveman mode in plain prose.
- /caveman-stats
- Show real token usage and savings report for the session.
- "stop caveman"
- Revert to normal prose output.
How to use it
Activate with /caveman or by saying "caveman mode". Intensity levels: lite, full (default), ultra. The model drops articles, filler, pleasantries, and hedging while keeping all technical substance: code blocks stay intact, errors are quoted exactly, security warnings revert to normal prose automatically.
/caveman full
/caveman-stats real token usage report for the session
Say "stop caveman" to revert.
Why it earns its place
Most assistant output is packaging: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..." becomes "Bug in auth middleware. Token expiry check use < not <=. Fix:". Same information, fraction of the tokens. The skill has auto-clarity rules so destructive operations and multi-step sequences stay unambiguous.
Try it
> /caveman full