> ## Documentation Index
> Fetch the complete documentation index at: https://docs.automate.it.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Skills

> Teach the worker your voice once; reuse it on every task

A **skill** is a reusable instruction bundle: a name, written instructions, and optionally attached files and folders. Attach a skill to a task or automation and the worker reads it before drafting.

<Frame>
  <img src="https://mintcdn.com/workingdevshero/pGwpOR4S5W_Ps2je/images/screens/skills.png?fit=max&auto=format&n=pGwpOR4S5W_Ps2je&q=85&s=f7c12eff4387e64131a7130199d02f19" alt="The Skills screen: reusable instruction bundles" width="2880" height="1800" data-path="images/screens/skills.png" />
</Frame>

Typical skills:

* **Brand voice** — "Confident, playful, never corporate. Short sentences. No emoji in headlines."
* **Formatting rules** — how you structure threads, where links go, hashtag policy.
* **Product background** — a features file and pricing sheet the worker should draw facts from, attached to the skill so they travel with it.

## Creating and using skills

Manage skills on the **Skills** screen — create, edit, and attach files or whole folders from your [workspace files](/workspace/media). When creating a task or automation, attach any number of skills; the content records which skills it was written against.

Skills are shape-compatible with the `SKILL.md` convention used by agent runtimes, so the same bundle that guides the built-in worker also works for a [bring-your-own agent](/developers/overview) reading it over MCP or the CLI.

One skill per concern beats one mega-skill: "brand voice" and "launch-week messaging" as separate skills lets each task pull exactly the context it needs.