I built a tool that measures whether a Claude Code skill actually improves output quality, and tested it on Caveman

I was tired of maintaining skill.md files, so I built an open-source CLI that creates, manages, and observes skills using a Github repo. You can monitor, track, and share skills between sessions of any agent, all while iteratively improving/versioning them.

Anthropic's valuation surges to $965 billion, surpassing OpenAI

Production infrastructure for vibe coders

why is my claude code using these stupid american dates

We're almost there guys. AGI Soon

"Hand of to claude code"... Failed to unzip = 10m tokens down the drain