Day 6 | Where do I find my skills?

Last time I walked through my AI workflow. This post is about the skills that workflow uses, and where I find them.

I start with open-source projects. I look at Chinese and American ones, and at material in other languages too. Sometimes the Chinese open-source community takes a project from overseas and adds a lot more Chinese documentation and detail. Following those projects back to the original and across to the forks, I often find things I wouldn't have found otherwise.

So I try not to stick to one language. The Chinese community documents some things in more detail, and other languages can have hidden gems too.

Besides Superpowers, which I use in Claude, I keep an eye on skills other people share on X.

One post I came across listed the author's go-to skills:

• /grill-me: questions you about the request before any work starts, so the vague parts get pinned down.
• /humanizer: cleans up stiff, formulaic writing so the text reads more naturally.
• /fact-checker: helps check the claims in a piece before it goes out.
• /prompt-master: turns scattered ideas into a clearer task prompt.
• /linkedin-hook: helps write the opening of a LinkedIn post so people keep reading.

The list breaks everyday work into specific pieces: clarifying the request, writing the prompt, rewriting the text, checking the facts. Each one has its own tool.

For this series, for example, I have the AI polish each draft with humanizer-zh-tw, then I check the content and layout myself.

That's how I collect skills: read open-source projects, look for versions in other languages, and watch what people share on X. A lot of the useful stuff has already been put together by someone, so that's a good place to start.