# Skills and Skillsets

The **Skills** tab assigns capabilities to the Coworker. Skills are organised into sub-tabs by type. On each sub-tab, click **+** (**Add skills**) to pick skills to add, and remove a skill from its card to detach it.

<details class="watch">
<summary><span class="watch-thumb"><img src="https://i.ytimg.com/vi/6Pl4VbR_o20/mqdefault.jpg" alt=""><span class="video-length">4:05</span></span><span class="watch-text"><span class="watch-label">Watch: Skills and Skillsets</span><span class="watch-meta">4 min · Episode 3 of 9</span></span></summary>
<div class="video"><iframe src="https://www.youtube-nocookie.com/embed/6Pl4VbR_o20?rel=0&autoplay=1" title="Skills and Skillsets · Neo: Coworker in depth, Episode 3" allow="accelerometer; clipboard-write; encrypted-media; gyroscope; picture-in-picture" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen loading="lazy"></iframe></div>
<p class="video-caption">Part of <a href="https://www.youtube.com/playlist?list=PLdwdGIspgPs0">Neo: Coworker in depth</a>. <a href="../videos/coworkers.html#ep-3">All episodes</a>.</p>
</details>

[![Skills tab — skill sub-tabs, the + (Add skills) button, and the Skillsets panel](../images/coworker-skills-v2.png)](../images/coworker-skills-v2.png)

## Skill Types

| Sub-tab | What it holds |
| --- | --- |
| **Reusable (MCP/API)** | Reusable skills wrapped over MCP endpoints, in-house tools, and API tools. They appear automatically when the underlying tool is enabled in your workspace. |
| **Custom Skills** | Agentic Flow and Voice agents exposed as skills. The **+** button creates a new Voice or Flow agent. |
| **Coworker-as-Skill** | Other Coworkers in the workspace that this Coworker can call as skills. |
| **Agent Skills (.md)** | Markdown (`SKILL.md`) skills from the workspace Skills area. See [Skills](../12_foundry/integrations/skills.html) for creating and managing these. |

> **Note**
>
> Which sub-tabs you see depends on the capabilities enabled for your deployment. **Coworker-as-Skill** and **Agent Skills (.md)** are off unless your platform administrator turns them on. Creating a **Chat** agent from the Custom Skills **+** menu is being rolled out and may not yet be available.

## How Skill Selection Works

You do not have to decide in advance which skill handles what. The Coworker reads each enabled skill’s **description** and picks the right one for each request. Clear, specific descriptions are what make it choose well — vague descriptions lead to the wrong skill being used.

When the Coworker loads an **Agent Skills (.md)** skill during a chat, NeoGPT shows **Using skill** followed by the skill’s name and a **Loaded** badge. Other skill types do not show this indicator — to see when they run, open the [Traces](observability.html) tab.

## Adding a Skill

- **Reusable skills** appear automatically once their underlying MCP, in-house, or API tool is enabled in the workspace — see [Foundry → Integrations](../12_foundry/integrations/index.html).
- To make an **Agentic Flow** or **Voice agent** available as a Custom skill, use the **Expose as Skill** action on that agent’s own Settings page. It then appears in the Custom Skills sub-tab of every Coworker in the workspace. (See [Agentic Flow](../9_agentic_flow_builder/index.html) and [Voice Agent](../10_voice_agent_builder/index.html).)

Each attached skill has an approval gate — **Allow** or **Ask** — set on the [Governance](governance.html) tab. New skills default to **Allow**.

## Skillsets

A **Skillset** is a named, reusable bundle of skills, carrying a name and description. Skillsets let you assign a whole group of skills in one action.

- **Manage Skillsets** from **Integrations → Skillsets**: create, rename, edit the description, add or remove skills, and delete.
- On a Coworker’s Skills tab, the **Skillsets** panel lists your skillsets. Click **+** (**Assign skillset**) on a set to copy all its skills onto the Coworker. The skills are then attached individually — remove any you do not need from their sub-tab.
- A skillset can hold **Custom Skills** and **Agent Skills**.

## Sharing a `SKILL.md` Skill

Markdown skills are managed in the workspace **Skills** area, and each carries its own access policy — set from a **Share** dialog — that controls who can use and change the skill definition:

| Role | Permissions granted |
| --- | --- |
| **Viewer** | Read. |
| **Executor** | Read, Execute. |
| **Editor** | Read, Execute, Edit. |
| **Owner** | Read, Execute, Edit, Delete, Share. |

> **Note**
>
> These five permissions govern the *skill definition* in the Skills area — who can view, run, edit, delete, or share it. They are separate from the Coworker’s **Allow / Ask** gate: when a `SKILL.md` skill is attached to a Coworker, the Allow/Ask gate controls whether its use needs approval, and the skill’s own sharing policy is unchanged. Platform-wide skills are managed by a Super Admin under **Control Center → Integrations → Skills**.