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

# Skills overview

> Reusable instruction packages that make agents do a task the same expert way every time.

A skill is a reusable package of instructions that you give to Erstan's [Agents](/agents/overview) so they handle a task a consistent, expert way. Instead of re-explaining how to age receivables or format a vendor-bill summary in every chat, you capture the know-how once as a skill and any agent can load it.

<Frame caption="The Skills library (Workspace tab)">
  <img src="https://mintcdn.com/erstan/VvzUuagDdyVkLVu5/images/skills-workspace-list.png?fit=max&auto=format&n=VvzUuagDdyVkLVu5&q=85&s=5b990b210dc61c3841695d84209a7c9e" alt="The Skills page Workspace tab with skill rows, search, and filter pills" width="2880" height="1800" data-path="images/skills-workspace-list.png" />
</Frame>

## What a skill contains

Every skill is a small file package authored in the [skill editor](/skills/create-edit):

* **SKILL.md** — the required instructions file. Its YAML `name` is the immutable lowercase package identity, and `description` tells agents when to use it. The friendly Display name is stored separately in Erstan; Agent Skills packages do not define a `slug`. Step-by-step guidance and optional **Examples** and **Gotchas** follow the frontmatter.
* **Supporting files** — optional reference material the agent can pull in (templates, lookups, notes).
* **Actions** — optional executable code attached to the skill. A skill with no actions is labeled **Prompt-only**; one with actions shows a count like **2 actions**.

<Note>
  Prompt-only skills are pure guidance — the most common kind and the easiest to author. Skills with actions can run code, and their entry file can be test-run in a console inside the editor.
</Note>

## Workspace vs Built-in skills

The Skills page (open **Skills** from the sidebar) has two tabs, each with a count badge.

<Tabs>
  <Tab title="Workspace">
    Skills created by you or your teammates (**Custom**) or installed from Erstan and partners. These are the skills you can edit, publish, share, and organize. Each has a status, a folder, and a visibility setting.
  </Tab>

  <Tab title="Built-in">
    Skills Erstan ships out of the box, labeled **Built-in**. They are read-only — you can inspect their files, About, and usage, but only owners and admins can prioritize, disable, or replace them. See [Built-in skills](/skills/built-in).
  </Tab>
</Tabs>

<Info>
  The Skills surface exists both workspace-wide and inside a [team](/teams/overview). The team version scopes to that team and hides the Built-in tab and the Location filter.
</Info>

## How a skill is classified

When you browse, three properties tell you what a skill is and whether agents can use it.

| Property       | Values                                 | What it means                                                                                                     |
| -------------- | -------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| **Source**     | By Erstan, Partner, Custom             | Where the skill came from: an official Erstan template, a partner/community skill, or one made in your workspace. |
| **Status**     | Draft, Published, Deprecated, Archived | The lifecycle state of a workspace skill. **Published** is what makes a skill usable by agents.                   |
| **Capability** | Prompt-only, Has actions               | Whether the skill includes executable actions or is instructions only.                                            |

<Tip>
  On skill tiles and badges, **Published** status appears as **"Agent ready."** The editor and filters call it **Published** — it is the same state, two labels.
</Tip>

## How agents use a skill

<Warning>
  Only **Published** skills are usable by agents. A skill stays a **Draft** until you publish it, and publishing is blocked until any package validation errors are resolved. Add a skill to an agent on the agent's setup — see [Giving agents tools, skills & connections](/agents/tools-skills-connections).
</Warning>

When an agent runs a task that matches a skill it has been given, it loads the skill's SKILL.md and follows those instructions — so the same job is done the same way every time, whether the agent runs from chat, a form, or a schedule. Skills carry guidance only; the ability to read or write NetSuite still comes from the agent's [connectors and write policy](/connectors/netsuite-manage), and writes still pause for [approval](/chat/approvals).

## Browsing, searching, and filtering

<Steps>
  <Step title="Open Skills">
    Choose **Skills** in the sidebar, then pick the **Workspace** or **Built-in** tab.
  </Step>

  <Step title="Search">
    Type in the search box to find skills by name or description.
  </Step>

  <Step title="Refine with filter pills">
    Narrow by **Location** (Company, Personal, Shared with me, a specific team), **Source**, **Category**, **Status**, **Creator**, and **Capability** (Prompt-only / Has actions). The Built-in tab offers a **Category** filter.
  </Step>

  <Step title="Sort and switch views">
    Change **Sort** (Recently updated, Name, Usage 30d) and toggle **List** or **Card** view. Use **Clear filters** to reset.
  </Step>
</Steps>

<Note>
  Your filter, sort, and view choices are saved in your browser per workspace and team, so they persist between visits on the same device.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="Create and edit skills" icon="pen-line" href="/skills/create-edit">
    Author a skill by hand or with Skill Chat in the package editor.
  </Card>

  <Card title="Publish, share & lifecycle" icon="share-2" href="/skills/publish-share">
    Publish so agents can use it, set visibility, and manage versions.
  </Card>

  <Card title="Built-in skills" icon="package" href="/skills/built-in">
    Review the skills Erstan ships and prioritize, disable, or replace them.
  </Card>

  <Card title="Starter prompts" icon="book-open" href="/skills/starter-prompts">
    Ready-made prompts you can copy, run, or send to the best-fit agent.
  </Card>
</CardGroup>
