Capturing expert know-how: using Skills in the Agent
A Skill isn't another way to ask data questions. It turns your team's proven approach into a reusable SOP: the Agent runs the fixed steps by skillCode, instead of re-explaining the metric definitions every time.
Where to start
| Order | Doc | For whom |
|---|---|---|
| 1 | What is a Skill | You're new to skillCodes |
| 2 | Use Skills for common tasks | You want to invoke an existing Skill in a chat |
| 3 | Turn your experience into a Skill | Admins/developers maintaining Skills |
| 4 | Knowledge bases and team reuse | You need to add terms, rules, or metric context |
Skill vs. pages vs. Knowledge
| Capability | Best for | Typical examples |
|---|---|---|
| Skill | Reusing a method, judgment rules, or an execution flow | Weekly report steps, refund root-cause SOP |
| Pages | Reusing a visualization, list, dashboard, or action entry point | Customer list, sales dashboard |
| Knowledge | Providing background definitions and business terminology | VIP criteria, approval rules, glossary |
The three work together: Knowledge defines the terms, Skill defines the steps, and pages carry the stable results.
FAQ
What's the difference between a Skill and Knowledge?
Skill = how (steps, inputs, output format); Knowledge = what (terms, rules, case background). See Knowledge bases and team reuse.
Is this the same as lovrabet skill?
No. This refers to the business Skills under App Management → AI Asset Center → Skills in the Lovrabet workspace. For the external CLI's lovrabet skill, see Lovrabet CLI 2.0.
How is this different from just asking data questions?
Direct data questions suit one-off lookups. Skills suit the same set of steps being run repeatedly, where the team needs a unified approach.