bro the claude code skill caveman just added 文言文 (wenyan) mode we joked about using chinese for token savings and this madlad actually built the compression layer lmao 😂

Apr 16, 2026 · 10:11 AM UTC

9
3
66
4,611
github.com/JuliusBrussee/cav…

GitHub - JuliusBrussee/caveman: 🪨 why use many token when few token do trick. Viral skill + proxy...

🪨 why use many token when few token do trick. Viral skill + proxy for coding agents that cuts 65% of tokens by talking like a caveman. - JuliusBrussee/caveman

github.com
5
691
Sort replies: Relevant Recent Liked
Replying to @leftcurvedev_
wait is this actually saving tokens or just cursed lol
1
1
168
the repo claims it saves up to 22-75% tokens following your prompts
1
176
more replies
Replying to @leftcurvedev_
is it fr?
1
167
Replying to @leftcurvedev_
He is Mad! @julius_brussee
1
80
Replying to @leftcurvedev_
He is Mad!!! @julius_brussee
72
Replying to @leftcurvedev_
confucius didn't die for this...but my token bill absolutely might.
2
79
Replying to @leftcurvedev_
But chinese actually saves tokens
3
157
Replying to @leftcurvedev_
gotta go learn chinese then
1
209
Replying to @leftcurvedev_
Tokenizer bias finally working in our favor for once. The semantic density of classical Chinese is unironically the ultimate context window hack.
21