@kirillk_web3 - Ponytail Skill Claims 90%
@kirillk_web3 on X A viral X post promotes "Ponytail," a free GitHub skill for Claude Code that supposedly makes Claude write less code and use far fewer tokens. The headline claims are hype-heavy and the post itself offers no independent measurements, so treat the 58,000-star figure and the savings numbers as unverified until you check the repo.
- The post credits Ponytail with 54% less code, 20% lower cost, and 27% faster output, but gives no benchmark, test setup, or methodology behind those numbers.
- The 90% token reduction in the headline is attributed to Anthropic's CEO in the framing, yet the quoted post itself never makes that claim or cites any source for it.
- Star counts and forks are the main evidence offered, and stars measure attention, not whether the skill performs as advertised.
The post is a promotional quote-tweet from a Web3 account, and it's thin on verifiable detail. Its core claim is that a free GitHub skill called "Ponytail" for Claude Code hit 58,000 stars in a week, with 2.9k forks, and that it makes Claude write less code: 54% less code, 20% cheaper, 27% faster. The headline also says it cut token usage by 90% and frames the whole thing as something Anthropic's CEO supposedly noticed, but nothing in the body backs that up. There's no link to the repo in the text, no methodology for the percentages, no baseline, and no named benchmark. The pitch is essentially "install this and your bill drops," wrapped in urgency language ("bookmark this before you forget") and a flow chart ending in "Money."
Worth separating two things here. Claude Code skills are plain instruction files that shape how the model behaves, so a prompt-style tool that nudges Claude toward terser, more minimal output is technically plausible, and fewer output tokens would mean lower cost. But "less code" and "20% cheaper" are different metrics, and the post never says how either was measured, on what tasks, or against what control. Star counts are also easy to inflate and say little about whether the thing works. A quick check on the actual repository, its issues, and any independent tests would tell you far more than this post does.
If you want to save it, file it as "possibly interesting, unverified." Before you install anything or repeat the numbers, look for the source repo yourself, read the README for how the claims were tested, and try it on a real task of your own to see whether the output is still correct and whether it's actually cheaper.