Metric Semantic Layer Skill

"Active users" means three different things in three dashboards — that's the problem a semantic layer solves: define each metric once, precisely, and every tool reads the same definition. This skill writes that definition — the exact formula, base measure, allowed dimensions, default filters, and the edge cases that usually cause drift — in a tool-ready form (dbt MetricFlow / Cube / LookML).

Required Inputs

Ask for these only if they aren't already provided:

• The metric — its name and the business question it answers.
• The base data — the model/table and the column(s) it's computed from.
• The aggregation — sum, count, count distinct, average, ratio.
• Dimensions & filters — how it can be sliced, and any default filters (exclude test accounts, internal users, refunds).
• Tool — dbt MetricFlow, Cube, LookML, or tool-agnostic.

Output Format