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

# Context vs skills

> Two ways to teach the agent: context is what's true, skills are how to work. When to use each.

There are two ways to teach Superscale, and it's worth knowing which is which. **Context** is what's true about your brand. **Skills** are how you want a job done. Both scope to your workspace, a brand, or a product, so the real difference is knowledge versus method.

<Frame>
  <img src="https://mintcdn.com/superscale/zBwb-PMIlqBo80UG/images/Files-vs-Skills.png?fit=max&auto=format&n=zBwb-PMIlqBo80UG&q=85&s=b26c3af4f27d4ac6db5966aa48433c06" alt="Files Vs Skills" width="1440" height="457" data-path="images/Files-vs-Skills.png" />
</Frame>

## The difference

| | Context | Skills |
| - | - | - |
| **What it is** | What's true about your brand, product, and audience | A repeatable method for doing a job the same way each time |
| **Format** | Free-text files, notes, and CSVs | A named, step-by-step procedure |
| **The agent** | Reads it for facts and voice | Runs it as a workflow |
| **Examples** | Brand voice, ICPs, claims, reviews, winning ads | Weekly account audit, competitor analysis, a QA check, a concept formula |
| **Scope** | Workspace, brand, or product | Workspace, brand, or product |
| **Usage** | Applied automatically whenever it's assigned to the scope you're working in, like the brand or product | Used only when you call it: type `/` to pick a skill, or describe the job so the agent matches it (ask for an "ad audit" and it runs your ad-audit skill) |
| **Also** | Versioned, searchable, and editable | Uses tools, runs on a schedule, starts from an expert template |

## When to reach for which

<CardGroup cols={2}>
  <Card title="Use context for facts">
    Anything that should hold across many ads: your voice, claims, products, ICPs, and proven winners.
  </Card>

  <Card title="Use a skill for a workflow">
    Anything you'd run the same way each time: a weekly report, a research method, an approval check, a hook formula.
  </Card>
</CardGroup>

<Note>
  A quick test: if you'd write it as "this is true...", it's context. If you'd write it as "do these steps...", it's a skill.
</Note>

## Skills go a step further

A skill isn't just a saved prompt. Once written, it can call the agent's tools, run on a [schedule](/features/scheduled-runs) so a job happens on its own, and start from an **expert skill**, a prebuilt template you adopt and tweak rather than writing from scratch.

<CardGroup cols={2} />


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.