mbrummerstedt/powerbi-analyst-mcp

MCP Developer Tools

A local server that lets LLMs like Claude query Power BI semantic models directly, turning natural‑language requests into live DAX results while keeping all data on the user's machine.

Track releases GitHub

Python Latest v0.4.1 · 1mo ago Security brief →

Features

Run natural‑language analytics against live Power BI models via LLMs (e.g., Claude).
Automatically save large query results to local CSV and page them on demand, preserving context window limits.
Log every DAX query and result for audit trails and reusable patterns across sessions.
Authenticates with Power BI using the OAuth device‑code flow; all data stays on the user's computer.

Recent releases

View all 4 releases →

v0.4.1 Mixed 1mo

⚠ Upgrade required

Build-time validation in `build.sh` will fail if `manifest.json` tools do not match registered code tools
`CLAUDE.md` added documenting bundle rules for future development

Notable features

`list_apps` becomes the default discovery path with updated tool descriptions
403 error handling in `execute_dax` now gives an actionable message about missing Build permission

Full changelog

What's changed

list_apps is now the default discovery path — tool descriptions updated to guide the LLM toward apps first, list_workspaces demoted to fallback
403 error handling in execute_dax — actionable message when Build permission is missing, pointing users to check their app access
Build-time tool validation — build.sh now fails if manifest.json tools don't match registered tools in code
CLAUDE.md added — documents bundle rules for future development

Installation

The .mcpb bundle is distributed internally. Contact your admin for the latest bundle file.

View release on GitHub

v0.1.2 New feature 2mo

Notable features

Add `glama.json` for Glama MCP directory listing
Add Glama badge to README
Add Dockerfile for containerized deployment

Full changelog

Changes

Defer MSAL authentication to first tool call — server now boots with placeholder credentials (enables Glama server inspection)
Add glama.json for Glama MCP directory listing
Add Glama badge to README
Add Dockerfile for containerized deployment

View release on GitHub

v0.1.1 Feature 2mo

Notable features