3.1 What are skills

When to use a skill

Use a skill when you keep asking AI to follow the same steps.

For example:

• Before every launch, you check PDP benefits and conversion risks.
• Every week, you summarize store data and create next actions.
• Before writing ads, you check for exaggerated claims.
• For competitor research, you want the same table format every time.

A skill turns repeated instructions into a reusable workflow, so the AI misses fewer steps and produces more consistent results.

Screenshot placeholder: Skill detail page. Capture the skill name, summary, use cases, and install button.

Use chat or install a skill

Use this quick rule:

Your situation	Recommended action
You only have a temporary question	Use normal chat
You only need to rewrite one sentence	Use normal chat
You do the same task many times a week	Use a skill
Your team needs consistent output format	Use a skill
The task has fixed review standards	Use a skill

If the workflow is still unclear, run it in normal chat a few times. Once the request becomes repeatable, install or create a skill.

What to check in skill details

You do not need to read every word. Check these four things first:

1. What task the skill solves.
2. What information you need before using it.
3. What the output will look like.
4. Which role should receive it.

If all four match your task, install and test it.

Screenshot placeholder: Skill instruction area. Mark use cases, input requirements, output format, and suitable role.

A simple example

Suppose you are preparing a product launch.

Normal chat:

Help me write a launch plan.

Using a skill:

Please use the "Product launch plan" skill. Based on this product information, create a plan from 7 days before launch to 7 days after launch.
Include channels, content themes, owners, and checkpoints.

The second request is more likely to produce an actionable result because it names the skill, input, and output.

What to prepare before using a skill

You do not need a large brief, but give the AI enough business context:

• Goal: what you want to complete.
• Object: brand, product, project, or channel.
• Materials: PDP, assets, data, links, or files.
• Requirements: output format, voice, platform limits, or deadline.

The more specific the input, the easier it is for the skill to produce a usable result.