Power BI CLI for Claude Code

Give Claude Code the Power BI skills it needs.
Install once, then just ask Claude to work with your semantic models and reports.

Why pbi-cli • Get Started • Modeling • Reporting • Skills • All Commands • Contributing

Why pbi-cli?

_{Cumulative downloads, refreshed daily from pypistats.org via GitHub Actions.}

Get Started

pipx install pbi-cli-tool    # 1. Install (handles PATH automatically)
pbi-cli skills install       # 2. Register Claude Code skills (one-time setup)
pbi connect                  # 3. Connect to Power BI Desktop

Open Power BI Desktop with a .pbix file, run the three commands above, and start asking Claude.

Requires: Windows with Python 3.10+ and Power BI Desktop running.

Install and set up pbi-cli from https://github.com/MinaSaad1/pbi-cli.git

pip install pbi-cli-tool

python -c "import site; print(site.getusersitepackages().replace('site-packages','Scripts'))"

Semantic Model Layer

Ask Claude to work with your Power BI semantic model. Requires pbi connect.

Create measures in bulk

Debug broken DAX

Snapshot and restore your model

Audit your model for issues

Test row-level security

Report Layer

Ask Claude to build and manage your Power BI reports. No connection needed -- works directly on PBIR files.

Build a report in 6 steps

Visuals, pages, themes, filters

32 visual types

Architecture

Two layers, one CLI:

• Semantic Model layer -- Direct in-process .NET interop from Python to Power BI Desktop via TOM/ADOMD. No MCP server, no external binaries, sub-second execution.
• Report layer -- Reads and writes PBIR (Enhanced Report Format) JSON files directly. No connection needed. Works with .pbip projects.

Desktop Auto-Sync

~/.pbi-cli/
  config.json          # Default connection preference
  connections.json     # Named connections
  repl_history         # REPL command history

Skills

After running pbi-cli skills install, Claude Code discovers 13 Power BI skills. Each skill teaches Claude a different area. You don't need to memorize commands.

Semantic Model Skills (require pbi connect)

Report Layer Skills (no connection needed)

Custom Visual Authoring

The Custom Visuals skill turns vibe-coding into the Power BI Visuals SDK loop. Claude scaffolds a sibling TypeScript project with npx pbiviz new, iterates against the Power BI Visuals API with tsc --noEmit between every change, packages a .pbiviz, and embeds it into your report:

# After installing the skill, just describe what you want in Claude Code:
#   "Build me a radial gauge that highlights values above target"
#
# Under the hood, the skill drives:
npx --yes powerbi-visuals-tools@^5.6.0 new mygaugevisual
# (Claude edits src/visual.ts + capabilities.json, runs tsc --noEmit until clean)
npx --yes powerbi-visuals-tools@^5.6.0 package
pbi visual import-custom dist/mygaugevisual.1.0.5.pbiviz --replace

The skill auto-installs Node and pbiviz on first run (with your consent), pins the SDK to a known-good version, keeps the TypeScript project as a sibling to your .pbip (no PBIR contamination), and operates under a curated npm allowlist for common deps (D3, Lodash, date-fns) -- anything off-list requires explicit approval.

Three new pbi-cli commands support the loop:

All Commands

27 command groups covering both the semantic model and the report layer. Every command below is a pbi subcommand, except where the pbi-cli prefix is shown.

Use --json for machine-readable output (for scripts and AI agents):

pbi --json measure list
pbi --json dax execute "EVALUATE Sales"
pbi --json visual list --page overview

Run pbi <command> --help for full options.

REPL Mode

For interactive work, the REPL keeps a persistent connection:

$ pbi repl

pbi> connect --data-source localhost:54321
Connected: localhost-54321

pbi(localhost-54321)> measure list
pbi(localhost-54321)> dax execute "EVALUATE TOPN(5, Sales)"
pbi(localhost-54321)> exit

Tab completion, command history, and a dynamic prompt showing your active connection.

Development

git clone https://github.com/MinaSaad1/pbi-cli.git
cd pbi-cli
pip install -e ".[dev]"

ruff check src/ tests/         # Lint
mypy src/                      # Type check
pytest -m "not e2e"            # Run tests (488 tests)

Bundled third-party software

pbi-cli-tool ships with Microsoft Analysis Services client library assemblies (Microsoft.AnalysisServices.*.dll) under src/pbi_cli/dlls/. These binaries are not covered by pbi-cli's MIT license. They are redistributed unmodified under the Microsoft Software License Terms for Microsoft Analysis Management Objects (AMO) and Microsoft Analysis Services - ADOMD.NET. Full terms are in THIRD_PARTY_LICENSES.md and the companion NOTICE file. By installing pbi-cli-tool you agree to those terms in addition to the MIT License that applies to the rest of the package.

Contributing

Contributions are welcome! Please open an issue first to discuss what you'd like to change.

1. Fork the repository
2. Create a feature branch
3. Make your changes with tests
4. Open a pull request

_{MIT License — bundled Microsoft DLLs are licensed separately, see THIRD_PARTY_LICENSES.md}

About

pbi-cli is built and maintained by Mina Saad. It started as a fix for the slowest part of my own job: authoring a measure meant a dialog, a refresh, and a visual check, every time.

If your team has the same problem one layer up, that is the work I do.

I work out what is actually costing a business, then build whatever fixes it: an AI agent, an automation, or a full application. Handed over so your team owns it.

Book a call · mina-saad.com