Find out which dashboards are actually earning their keep
Audits your BI platform's usage data to identify unused dashboards, duplicate reports, conflicting metrics, and orphaned data sources. Generates a cleanup plan with recommendations for archival, consolidation, and maintenance.
Create a skill called "Dashboard Auditor". Connect to my BI platform's API (I'll tell you if it's Tableau, Power BI, Looker, or Metabase) and audit all dashboards/reports. For each, collect: title, creator, creation date, last modified date, last viewed date, total views in last 30/60/90 days, unique viewers, data sources used, and metrics displayed. Then analyze: (1) Unused dashboards — not viewed in 90+ days. (2) Duplicates — dashboards using the same data sources with similar metrics. (3) Orphaned data sources — connections not used by any active dashboard. (4) Stale dashboards — data sources that haven't refreshed recently. Generate a cleanup report with recommended actions (archive, consolidate, refresh, keep) and estimated maintenance savings.
Every BI deployment ends up with hundreds of dashboards, most of which nobody
uses. They clutter search results, confuse new analysts, and sometimes show
wrong numbers that nobody notices because nobody looks at them. This skill
audits the graveyard and helps you clean it up.
Catch conflicting metric definitions before they reach a meeting
Scans your SQL queries, dbt models, and dashboard definitions to find every place a key metric is calculated. Flags inconsistencies — like marketing counting revenue with refunds and finance counting without — before they cause a trust-destroying meeting.
Write DAX by describing what you want, not how
Describe a measure in plain English — "percentage of total revenue by product category" — and get correct DAX with an explanation of evaluation context, filter propagation, and common pitfalls. Also debugs existing DAX that's returning wrong results.
Keep Final Cut libraries from ballooning via render, proxy, and optimized files
Final Cut libraries can balloon to hundreds of gigabytes due to generated media — render files, proxy media, optimized media, and analysis files. This recipe monitors library sizes, tracks growth, and provides safe cleanup steps so you never run out of space mid-project.
Verify dashboards before you trust them
A sanity-check workflow for broker performance charts, retirement projections, and portfolio analytics that may be misleading, incomplete, or just buggy.