dax-optimizer-sdk

Programmatic DAX Optimizer analysis for Power BI, Fabric, and Analysis Services models. Exports a model to VPAX format, uploads it to the DAX Optimizer service, and returns structured performance analysis results — all from a single function call.

Prerequisites

| Tool | Purpose | Install | |------|---------|---------| | .NET 8.0+ | Runtime for vpax-bridge and DAX Optimizer CLI | Download | | DAX Optimizer CLI | Uploads VPAX and runs analysis | dotnet tool install --global Dax.Optimizer.CLI |

vpax-bridge (bundled) is built automatically on npm install via dotnet build.

You also need a DAX Optimizer account with a workspace ID.

Authentication

DAX Optimizer auth (for analysis upload)

Three modes — pass an auth object:

// PAT (CI/CD, automation — requires Enterprise license)
auth: { method: 'pat', username: 'group:automation', token: process.env.TOKEN! }

// Interactive (opens browser, credentials cached by CLI)
auth: { method: 'interactive' }

// Group (opens browser, switches to group account)
auth: { method: 'group', username: 'my-team' }

Fabric auth (for VPAX export from cloud models)

For Fabric / Power BI Service models, add interactive: true to open a browser for Azure AD sign-in:

interactive: true

Or pass a pre-acquired token:

accessToken: '<bearer-token>'

Not needed for local Power BI Desktop models.

Quick start

Local Power BI Desktop

import { analyzeDaxModel } from 'dax-optimizer-sdk'

const result = await analyzeDaxModel({
  auth: { method: 'pat', username: 'group:automation', token: process.env.TOKEN! },
  server: 'localhost:65072',
  databaseId: 'f97515be-bbb7-434b-83fc-b2599bf8a754',
  workspaceId: process.env.WORKSPACE_ID!,
  modelName: 'My Model',
  modelCreate: true,
  contractId: '...',
})

Fabric / Power BI Service

const result = await analyzeDaxModel({
  auth: { method: 'pat', username: 'group:automation', token: process.env.TOKEN! },
  server: 'powerbi://api.powerbi.com/v1.0/myorg/My Workspace',
  databaseId: 'My Dataset',
  interactive: true,  // opens browser for Azure AD sign-in
  workspaceId: process.env.WORKSPACE_ID!,
  modelName: 'My Dataset',
  modelCreate: true,
  contractId: '...',
})

Where do server and databaseId come from?

This library does not include Power BI model discovery. The caller provides the connection details:

• Local PBI Desktop — use Power BI MCP ListLocalInstances + Connect + database_operations List
• Fabric — use the XMLA endpoint URL: powerbi://api.powerbi.com/v1.0/myorg/<workspace> and the dataset name as databaseId

API

analyzeDaxModel(options): Promise<DaxOptimizerResult>

End-to-end: preflight check → VPAX export → analysis → results → cleanup.

checkPrerequisites(): PreflightResult

Returns { dotnet, daxoptimizer, vpaxBridge } booleans.

exportVpax(options): Promise<string>

Exports a model to VPAX. Supports local, Fabric (interactive), and Fabric (token).

runAnalysis(options): Promise<DaxOptimizerResult>

Analyzes an existing VPAX file.

logout(): Promise<void>

Clears cached DAX Optimizer CLI credentials.

isInteractive(auth): boolean

Returns true if the DAX Optimizer auth mode may open a browser.

Error handling

| Error class | When | |-------------|------| | AuthError | Invalid auth config or login/logout failure | | PrerequisiteError | Missing tool (includes install command) | | VpaxExportError | VPAX export failed | | AnalysisError | DAX Optimizer CLI failed or results couldn't be parsed |

Known limitations

• Direct Lake models (compat level 1604) are not yet supported by DAX Optimizer
• TMDL/PBIP offline export produces a VPAX without data statistics, which DAX Optimizer rejects

License

MIT