Anandhuvimalan

Powerbi Analyst Skill — Data skill for Claude Code

Data community

A study-first Claude/Agent Skill that turns CSV, Excel, or other source data into production-ready Power BI artifacts: Power Query (M), TMDL relationships, DAX measures, and a visualization blueprint.

How to install Powerbi Analyst Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Anandhuvimalan/powerbi-analyst-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Powerbi Analyst Skill does

Alternatives in Data

Dummy Dataset — Generate realistic dummy datasets in CSV, JSON, or SQL 7.8k ★
MCP Server Chart — 🤖 A visualization mcp & skills contains 25+ visual charts using @antvis 4k ★
Portaljs Add Dataset — Add a dataset (CSV, TSV, JSON, or GeoJSON) to an existing PortalJS portal 2.3k ★

README

powerbi-analyst — a Claude / Agent Skill

A reusable [Agent Skill](https://docs.claude.com/en/docs/claude-code/skills) that makes an AI study source data **like a senior data analyst** and produce production-ready Power BI artifacts — instead of blindly running a fixed transformation template.

Give it a **CSV, Excel, JSON, a folder of files, or another tabular source**, and it will study the data first, decide what each column actually needs, and output:

Power Query (M) — one paste-ready .md per table, with the exact source path
relationships.tmdl.md — only the relationships real keys support
measures.tmdl.md — only measures that answer a real business question
Visualization Blueprint — step-by-step, every visual tied to a question
Data Quality & Governance Notes + Assumptions / Justifications

Philosophy

It behaves like a **Senior Enterprise Data Analyst and BI Consultant**, not a rule-based engine. The reference files are a *toolkit, not a checklist*:

Studies and classifies every column before any transformation.
Never auto-imputes missing values — decides per column and justifies it.
Treats outliers only when warranted; flags rather than deletes by default.
Never forces a star schema or invents dimensions.
Creates only meaningful measures and only valuable visuals.
Explains why each technique was applied and why others were skipped.
Avoids over-engineering and generic enterprise boilerplate.

All M and TMDL output is 100% copy-paste ready into Power BI Desktop (Advanced Editor / TMDL files) with zero comments.

Structure

powerbi-analyst/
├── SKILL.md                         # philosophy + workflow + hard output rules
└── references/
    ├── data-study.md                # column profiling & classification (phase 1)
    ├── power-query.md               # M connectors (CSV/Excel/JSON/folder), cleaning, imputation, outliers
    ├── tmdl-relationships.md        # relationship modeling
    ├