Skills overview

A skillis a portable folder of expertise the AI loads on demand. Instead of repeating the same instructions every time — your citation style, a lab’s reporting template, the steps to turn a dataset into a figure — you package them once, and the AI reaches for the skill when a task calls for it.

What a skill is

At its core a skill is a folder with a SKILL.md file: a name, a description of when to use it, and the instructions themselves. A skill can also bundle runnable helper scripts and reference files, so it carries not just whatto do but the tools to do it. The AI reads a skill’s description to decide whether it’s relevant to the task in front of it, and only then loads the full instructions.

Skills, agents & extensions

These three ways of extending Multilo stack rather than compete:

• Skills teach the AI howto do something — reusable expertise loaded into a task. No code host required.
• Agents are configured workers for a job (a draft writer, a citation checker). An agent can call on skills. See custom agents.
• Extensionsadd new software surfaces — editors, viewers, commands — running in the sandboxed extension host. See extensions overview.

Trust & permissions

A skill can list the tools it’s allowed to use, and that list is honored when the AI runs it — a skill can’t quietly reach for capabilities it didn’t declare. Skills you install from the marketplace are reviewed and signed, the same as extensions, so you can see their source and trust tier before enabling them.