If you still do not know what (agent) skills are after reading that web page for the spec, that’s fine. Let’s see, how about:
Skills are role-oriented text prompts used by agent harnesses (like Claude Code, Gemini CLI, Cursor, VSCode and its many forklings, etc.) to improve the usefulness of large language models in specific, predictable ways … while still allowing for plenty of emergent serendipity, also known as “benevolent prompt injection” or BPI (it kind of actually is, you can look it up — well, not the acronym, that was more an aspect of the “truth joke”).
They really feel like cyborg superpowers for your wordflows — and if you still don’t believe it, and need proof, you’ll find some of that here, is the hope.
This forum facilitates 3 dimensions of skill discourse:
discussion (admittedly a wide dimension)
demonstration and
distribution
Distribution can be you copying someone’s markdown and putting it in a file you package up and “install” (quotes redundant for some harness situations, but not all).
Or you can upload/download zips.
Safety? PRESUME EVERY ZIP YOU DOWNLOAD FROM THIS FORUM IS A CHARISMATIC TRICK BOTNET FULL OF POISON GAS before you install or use it somewhere.
It is extremely easy to open and introspect any real skill zip’s contents and computer files can’t “do” poison gas. Just check their work and your own.
We do not, generally, have any interest in talking about harnesses themselves, much less advocating for one or the other. If it can use skills, great! It’s a great harness, we all agree. Focus on the skills themselves.
Feel free to explicitly declare a license for your work in your post. Otherwise, by posting skills or code here, you agree that your work defaults to the CC0 1.0 Universal Public Domain Dedication (the same license covering the site’s general text).
It will be common to post and read LLM-generated text on this site. However, when you write a post as yourself, you need to write it in your own human voice and words and typing. However you want to do it.
But even in a flashy promo post for your cool new skill, you are not allowed to post LLM text as, or intermixed with, your own.
It will be very obvious, and you will be banned if you do it.
Put more officially: Clearly differentiate with visual and context cues and clues when and why you are including skill-assisted generated content in a post, and otherwise write posts in your own real human voice. If you don’t consistently do these things, you will be banned.
If you have your agents post here … you will also be banned, and obviously the agents will too. They can read stuff, that’s fine. No agent posting.
Beyond that, be nice, don’t be rude, bring positive energy.
You are kind of expected to “steal” the free stuff so no worries on that, but always respect others’ rights, and try to attribute others’ work even if you are not strictly “required” to do so.
Do not post anything confidential or owned by your company or something.
Please, be reasonable.
This Discourse forum (group?) is for personal and public entertainment and experimental use only.
How much trouble are you looking to get into? I’ve never been able to ban anyone from a cool thing before; don’t threaten me with a good time.
