Skip to main content

Using AI Skills

An AI skill is a knowledge file your organization uploads to teach NxAgent something it cannot infer: your naming convention, your error-handling standard, how your internal systems fit together. Once a skill exists, it shapes what NxAgent produces.

Your organization administrator creates and manages skills. See Governing AI for creating, uploading, and reviewing them.

Two ways a skill gets used​

Automatically​

An active skill applies on its own whenever it is relevant. There is no button and nothing to remember.

With a skill describing your naming convention in place, asking NxAgent to build a workflow produces a name that follows that convention without your mentioning it.

Deliberately, with the skill picker​

Sometimes you want to be certain a skill is used rather than hope it is picked up.

  1. In the chat input, enter /. A list of the skills available to you appears.
  2. Choose a skill with the arrow keys or the mouse. It is attached to the message you are about to send.
  3. Attach further skills the same way if you need more than one.
  4. Describe what you want, then send.

The picker shows organization-level and application-level skills together. To change your mind, detach a skill before sending.

Attaching a skill makes the behavior explicit and repeatable, which is what you want when the skill matters to this particular request. This is also how you apply a skill while building a workflow in Build mode: attach it, then describe what you want built.

Manage skills conversationally​

You can also ask about skills in the chat panel:

What skills are available?

Create a skill that captures our workflow naming convention: department, then process, then version number.

Write a skill that gets used​

A skill's description does real work: it is what NxAgent matches against when deciding whether the skill applies to the question in front of it. A vague description means the skill sits unused, and a clear one means it activates when it should.

How you know it worked​

Attached skills appear as chips above the input before you send. In the response, an applied skill shows in the output following the convention or standard it describes.

Next steps​