Semanticus is a free, cross-platform workbench for Power BI & Fabric semantic models, built for VS Code. You get an A–F AI-readiness score, one-click fixes, deep lineage, a DAX lab, and a door for your AI assistant to work the same live model you do. Behind a referee.
The daily-driver tooling you know from the Windows-only classics, rebuilt cross-platform inside VS Code, working on Power BI Desktop, PBIP/TMDL files, and Fabric XMLA endpoints.
Tables, measures, columns, relationships, calc groups, perspectives, RLS/OLS. A full model tree with a real DAX editor (autocomplete, go-to-definition, formatting) and a properties grid. Every edit is undoable.
An A–F grade over descriptions, naming, synonyms, formats, Prep-for-AI and more, with deterministic one-click safe fixes and grounded prompts for the rest. Hard gates keep the grade honest.
Tabular-Editor-compatible rules (your ignore annotations are honoured), auto-fixes, custom rulesets, and honest waivers: accepted findings stay visible, never hidden.
Trace any field end to end, see everything that breaks before you change it, and get tri-state safe-to-remove verdicts, report-aware down to the page and visual.
Build a visual from field wells, hover any point for its exact filter context, benchmark cold vs warm, capture query plans, and prove a rewrite is equivalent before you apply it.
A real M editor with standard-library autocomplete and applied-steps, spreadsheet-style column operations that write M, and a guided incremental-refresh policy builder.
An interactive ER canvas (bus-matrix, layered and free layouts; drag column-to-column to relate) and a VertiPaq treemap that shows where the memory actually goes.
Brandable model documentation with an authored narrative layer, a declarative model spec you can build from, and a full two-model diff with cherry-pick merge.
Dry-run first, always. The deploy gate runs BPA + readiness before any live write; shipping past a red gate takes a written, recorded reason.
Semanticus has two doors onto one live model: the Studio UI for you, and an MCP server for your AI assistant. Every capability ships through both. You watch each other work: live, attributed, reversible.
Connect Claude Code (or any MCP client) in one click. Ask it to describe every measure, hunt unused columns, optimise a slow DAX pattern, or take the model from a C to an A, then watch the tree, the score and the timeline update in front of you.
AI that ships to production needs more than vibes. Semanticus workflows are playbooks with teeth: each step carries a gate the engine verifies against the real model: DAX probes, equivalence proofs, readiness and best-practice re-scans. Declines need reasons. Overrides are recorded. The audit trail is hash-chained.
The lineage graph walks dependencies like a knowledge graph, and the safe-to-remove sweep is honest: green only when the model and your reports agree nothing uses it. No more deleting the column a hidden page filter needed.
Free, forever, for the full single-edit workflow. Pro adds the one-click bulk engine and enforced, evidence-verified workflow runs. See what’s in Pro.
Windows, macOS and Linux. Open Power BI Desktop models live, PBIP/TMDL/BIM files offline, or connect read-write to Fabric / Premium XMLA endpoints. The engine is bundled, so there is no separate .NET install.
No. The engine runs entirely on your machine, holds no API keys, performs no inference and sends no telemetry. AI work happens through your own AI assistant on your account, connected over MCP.
We love TE. Semanticus vendors its battle-tested open-source core. The difference is the shell and the thesis: cross-platform VS Code instead of WinForms, one consolidated tool instead of six, and AI-native dual-drive with verification instead of a human-only editor.
Enforcement and bulk. Free does everything one step at a time; Pro adds atomic change plans, fix-all, one-click safe fixes, rule-level waivers, enforced workflow runs and audit export. Full comparison →