What is a Skill?
A Skill is a set of instructions you save once and reuse whenever you ask Analytics Studio's reporting assistant a question. Instead of typing out the same request every time — which assignments for a specific group are overdue, what last month's completion rates looked like, how a cohort is progressing — you save the logic once as a Skill, then just ask for it by name going forward.
Think of a Skill as a repeatable recipe for a report, rather than the report itself. It doesn't just store what your data looked like on a given day — it stores how to build the answer, so it can be run again with fresh data or different inputs any time you need it.
What Skills are good for
Skills work well anywhere you find yourself asking for the same type of analysis over and over, especially when:
You want a consistent report that always uses the same metrics and formatting, no matter who runs it or when.
You want a guided process that asks for a bit of input first — for example, "which team or cohort do you want to check?" — before pulling the data.
You want to check for the same kind of issue repeatedly, such as scanning for data anomalies or flagging incomplete records.
You're building a repeatable onboarding or walkthrough that steps someone through key metrics in a set order.
How to create a Skill
On the Skills page, you'll see three options:
Create with AI — the fastest option, describe what you want ("create a Skill that checks completion rates by cohort broken down by week"). The assistant fills in the Name, Identifier, Description, and Instructions for you automatically.
Pro Tip! You can turn an existing Omni AI conversation into a Skill, once you have your desired output just type "turn this into a Skill"
Write your own — for full manual control. You fill in four fields yourself:
Name (up to 48 characters) — what it's called
Identifier (up to 48 characters) — a slash-command handle, e.g.
/quarterly-report, using lowercase letters, numbers, and dashes onlyDescription (up to 1,024 characters) — a summary of what it does
Instructions (up to 20,000 characters) — the step-by-step logic, written in Markdown
All four fields are required.
Upload a skill — for bringing in a Skill you've already written elsewhere as a Markdown file. There are two accepted formats:
With frontmatter (recommended). Frontmatter is a small block of labels at the very top of a file, wrapped between two lines of three dashes (---), that tells Analytics Studio the file's identifier, name, and description before the actual instructions start:
---
identifier: /skill-name
name: Skill Name
description: Brief summary
---
Instructions in Markdown format here.
With frontmatter, Analytics Studio reads those three fields automatically and nothing more is needed.
Without frontmatter. Analytics Studio treats the whole file as the instructions, then prompts you to fill in Name, Identifier, and Description by hand afterward.
For most people, Create with AI is the easiest way to get started.
Skills vs. a saved report
It's a fair question: if you've already saved a report, why do you need a Skill as well?
The difference is that a saved report is fixed. It reflects a specific set of filters and data as of the moment you saved it — if you or a teammate wants a different date range or a different segment, someone has to go in and edit it by hand.
A Skill isn't a result — it's the instructions for producing one. That means:
Anyone using it can get a fresh, correct version for their own situation just by asking, without editing anything.
It can do more than a single report can, such as asking a follow-up question first, then running several queries and combining the results.
It captures the judgment behind the report — which numbers matter and how they should be presented — so the same standard is applied every time it's used, rather than everyone building their own slightly different version.
What Skills can't do yet
Skills are still being developed, and one feature not available yet is Sharing a Skill.
Sharing isn't live yet. You can't share a Skill with specific teammates, a group, or your whole organization the way you can share a dashboard. This is planned but not yet released.