About
Built by Power BI developers, for Power BI developers
What the library covers
The current catalog spans six focused domains, each chosen because it represents work that's repetitive, error-prone, or simply too tedious to do well by hand:
Analyze — Best Practice Analysis parsing, bookmark leak detection, color audits, interaction matrices, measure dependency analysis, VertiPaq compression review, and full PBIP project audits covering star schema, relationships, DAX patterns, Power Query, and security.
Build — Page cloning with full bookmark and drillthrough preservation, dynamic titles, KPI card packs, paginated visuals, time granularity switchers, search slicers, expand/collapse matrices, Top N filters, configuration panels, and over a dozen other report-construction patterns.
Format — Row-by-row conditional formatting, synced Y axes, exact pixel column widths, and other layout details that Power BI makes harder than they should be.
DAX — ABC classification, Pareto analysis, time intelligence kits that generate 23 calculation items per measure, and conversion of implicit measures to explicit ones for full control.
Document — Visual inventory extraction, field-well mapping, and selection-pane hierarchy reporting into Excel.
Optimize — Duplicate measure detection, consolidation, and reference cleanup.
The catalog grows roughly every two to three weeks. New skills are sourced from real consulting work and tested against multiple models before they ship.
How each skill is built
A skill is not a prompt. It's a structured package containing a Claude markdown instruction file, supporting markdown documentation, and Python scripts that do the deterministic work — file parsing, TMDL manipulation, visual.json edits, validation logic.
This structure matters for three reasons.
It's transparent. You can read every skill before you run it. There's no hidden API call, no proprietary binary, no black box. If you want to know what a skill is going to do to your project, you open the folder and read it.
It's version-controllable. Skills live alongside your PBIP project in source control. Diffs are reviewable. Rollbacks are trivial.
It's customizable. If a skill is 90% right for your scenario, you can fork it and adjust the remaining 10%. The methodology for writing skills is documented, and we actively encourage developers to write their own using the same structure — or to contribute back patterns that other people would benefit from.
Each skill takes serious time to build properly. We're not shipping prompt-engineering experiments. We're shipping engineering work, tested against the messy realities of production semantic models.
Honest about limitations
No automation works in 100% of scenarios. Power BI projects vary enormously — different schemas, different naming conventions, custom themes, non-standard folder layouts, edge cases in the PBIR format that surface only on certain models.
What we commit to is this: every skill is tested against a wide range of edge cases before release, and when something breaks in the wild, it gets diagnosed, fixed, and re-tested. You'll find issues. We expect you to. When you do, we want to hear about them — that feedback is how the library improves.
This is a more useful promise than "it always works," because that promise isn't true of any developer tool, and you know it.
Why this approach is safer for your data
The PBIP-first architecture isn't just a design choice — it's a privacy boundary.
Skills operate on text files on disk: TMDL model metadata and PBIR visual definitions. They don't connect to your live model. They don't run DAX queries against your data. They don't read row values, customer records, or any of the sensitive content sitting inside your dataset.
That's the difference between this approach and an MCP-server-based workflow. With MCP, query results flow back to the language model — meaning row data leaves your machine. With PowerBI Skills, the only thing Claude sees is your model structure: table names, column names, measure expressions, visual configurations. Your data stays local.
For developers working under GDPR, HIPAA, or internal data-handling policies, that distinction is meaningful.
What you actually get out of it
Two things, and the second matters more than people expect.
First, time. Tasks that take two to four hours by hand run in under a minute. Bookmark debugging across a complex report. Color auditing across forty visuals. Converting every implicit measure in a model. Building a time period slicer with proper bridge-table performance. These aren't hypothetical wins; they're the unglamorous work that fills your week.
Second, knowledge. Because the skills are readable, working with them teaches you things. You learn how PBIR stores visual configurations. You see what the Tabular Object Model expects. You understand why certain DAX patterns perform better. The library isn't designed to hide complexity from you — it's designed to handle it consistently while leaving the reasoning visible.
Try before you commit
There's a free skill — Clone Page — available after sign-up. It's a real, working skill, not a marketing demo. It clones any report page with all visuals, bookmark navigators, interactions, and drillthrough bindings as a fully independent copy. Run it on a real project. See how it handles your edge cases. Read the code. Decide from there whether the rest of the catalog is worth your time.
Support and community
Reach out directly through the contact page, or join the Discord channel where users discuss skills, share custom adaptations, and report issues. Responses are direct and come from the people who actually wrote the skills, not a support queue.
What's coming
The library is actively maintained. New skills ship regularly. Existing skills are refined as the PBIP/PBIR format evolves and as community feedback surfaces edge cases. Recent additions include configuration panels, ABC classification, and Pareto analysis — and the next batch is already in development.
This is a long-term project. The work being done now is laying foundations for a much larger library, and every purchase supports that trajectory.
If you're a Power BI developer who values your time and trusts code you can read, you're in the right place.
Usage terms
Skills are licensed for personal use. One purchase, all your projects — no limits. Redistribution or sharing is not permitted.