oto.cx
powerbi

microsoft

powerbi.

READ Power BI with the person's rights: workspaces, reports, semantic models,

add to otoauth : —prefix : powerbi_*

setup & usage

⚠ connect with your Microsoft 365 account

on the "Power BI" card, click Authorize Power BI and choose your work account. Nothing else to install or register. The account itself is the "Microsoft 365 account" connector: Power BI borrows it and only adds its own permissions (read workspaces, reports and semantic models).

  • the agent acts with your rights: the workspaces you belong to, your "My workspace", and the models you can open; each person in the org connects their own account
  • ⚠️ many organizations require a Microsoft 365 administrator to authorize oto a first time, and an earlier approval only covers the permissions it named: if Microsoft shows "admin approval required", an administrator approves oto once for the whole organization (microsoft_admin_consent, see the "Microsoft 365 account" connector)
  • an account already linked for another Microsoft service only needs to authorize this one; an account that has not authorized Power BI is refused by the tools, naming this card
  • in Power BI itself: querying a model in DAX needs Read and Build on it, and the tenant setting « Semantic model Execute Queries REST API » turned on (on by default; it may be limited to a security group)
    • ▸ find, query, read the measures

    powerbi reads only (op=workspaces|reports|report_pages|datasets|query|measures), under the account the call names (_account=), otherwise the default one. Without workspace_id, the person's "My workspace".

  • "which Power BI reports do we have?" → powerbi() (the workspaces), then powerbi(op="reports", workspace_id=…); a report's datasetId is the model behind it
  • "what is in that model?" → powerbi(op="measures", workspace_id=…, dataset_id=…): every measure with its table, folder, description and, when allowed, its DAX formula
  • "what does it show for October?" → powerbi(op="query", workspace_id=…, dataset_id=…, dax="EVALUATE SUMMARIZECOLUMNS('Date'[Month], \"CA\", [CA])"): aggregate in DAX, at most 2,000 rows come back (max_rows)
    • · what is misleading
  • ⚠️ a measure's DAX formula is only given to a person with WRITE permission on the model (Contributor, Member or Admin of its workspace): with Read only, op="measures" returns names, tables, folders and descriptions, expression_visible=false and a note — no formula. Nothing in oto can go around it
  • a query is refused (400) on a DAX error, and cut (truncated) beyond Power BI's limits (100,000 rows, 15 MB) or max_rows
  • models hosted in Azure Analysis Services, or live-connected to one, cannot be queried
  • what the agent reads goes back to the conversation and nowhere else: oto stores no row
    • · scope

    read only: workspaces, reports and their pages, semantic models, one DAX query at a time, measure definitions. No refresh, no export, no edit, no report rendering.

    ⚙ multiple comptes

    this connector accepts multiple comptes: connecting with another compte adds it alongside the others, named by its address; reconnecting with the same one replaces its own.

  • the first linked compte is the default
  • without specification, the agent takes the only linked compte, otherwise the one marked as default; otherwise the call is refused, naming the available comptes — never a random choice
  • target a compte for a call: _account="<name>" on the tool; list them: oto_identity(op='list', connector='powerbi')
  • set the default: oto_identity(op='set', connector='powerbi', identity_id='<name>')
  • ⚠️ no tool walks through the comptes on its own: a total across several comptes is obtained by calling each one (_account) and adding up

    • outils

    chargement…

    usage

    claude code — install the oto plugin — claude plugin marketplace add otomata-tech/oto-plugin — mcp + skill configured.
    claude.ai — add mcp.oto.cx/mcp as a custom connector (guide)
    cli — pipx install oto-cli then oto powerbi …
    ← tous les connecteurscatalogue mcp →