A reference for agent skills — not a rewrite of the docs.

A skill is a folder with a SKILL.md file — metadata plus instructions telling an agent how to perform a specific task — and, optionally, bundled scripts, references, and templates. Agents load skills through progressive disclosure: only a name and description at startup, full instructions when a task actually matches, and any bundled files only when the instructions call for them. That keeps a large skill library cheap to keep on hand.

The format was originally developed at Anthropic and released as an open standard, which is why it now shows up across a wide range of agents rather than staying locked to one product. What differs between agents is everything around the format: where skills live, how they’re installed, what trust and security model applies, and whether the agent can write its own skills. That’s the part worth a dedicated guide per ecosystem, and why this site is organized by agent rather than as one generic skills explainer. Every page here needs a clear reason to exist beyond paraphrasing what already ranks — sometimes that’s missing information, but just as often it’s better synthesis, more verification, or a more usable structure than what’s already out there.