Known limitations

This page is current as of 2026-09-23. Microsoft and GitHub product names, preview labels, and UI paths can change. The lab pages omit lastVerified because the full track has not been run end to end on Windows yet.

Preview and changing features

Preview feature

The lab uses or discusses several features that are still moving:

Power BI Project files, TMDL, and PBIR. Desktop options and file output can change by Desktop release.
Power BI Prep data for AI output. The lab writes reviewable drafts under fabric/ai-prep/; it does not write Desktop-managed Copilot folders.
Power BI Modeling MCP. The cloud-agent template pins @microsoft/powerbi-modeling-mcp@0.5.0-beta.13, but the lab does not depend on it in cloud sessions.
VS Code hooks and agent customizations. Commands and panels can move between VS Code releases.
GitHub Copilot CLI and Copilot cloud agent. Cloud agent issue assignment is public preview in GitHub docs.
APM 0.x. This repo pins APM CLI 0.31.0 for the tested flow.
Astro Starlight 0.x in this site.

Windows-only parts

Windows only

Power BI Desktop runs only on Windows. Labs that open the PBIP, refresh data in Desktop, connect Desktop MCP, or run live DAX tests are Windows-only. macOS can still use the file-mode path for TMDL, PBIR, validation, agent customization, hooks, cloud agent, and the site.

Plan requirements

Needs a paid Copilot plan

GitHub Copilot code review and Copilot cloud agent require paid Copilot plan support and may require org or repository policy enablement. Business and Enterprise accounts can be blocked by admin policy even when the UI appears locally.

Fabric Copilot, Power BI Copilot, and Fabric data agents are not used in this lab. Microsoft Learn currently states that Fabric data agents need paid Fabric capacity such as F2 or higher, or Power BI Premium capacity with Fabric enabled. Those products are roadmap topics, not prerequisites.

Agent nondeterminism

Agents can produce different wording, plans, DAX, and intermediate edits each run. The labs use scripts, checks, and WhatVaries callouts so you verify durable state instead of exact chat text. When an agent overreaches, reject the diff, narrow the prompt, or restore a checkpoint.

Not fully verified on Windows yet

The repo and site have validation scripts, but the full lab track has not yet been run end to end on a clean Windows 11 machine. That is why lab frontmatter does not include lastVerified. Treat screenshots, menu names, and preview UI labels as subject to change until a full Windows pass is recorded.

Tested versions and pins

The site pins Astro and Starlight here:

  "dependencies": {
    "@astrojs/markdown-remark": "7.3.1",
    "@astrojs/starlight": "0.42.1",
    "astro": "7.3.2",
    "gray-matter": "4.0.3",
    "mermaid": "11.17.2",
    "sharp": "0.35.4",
    "starlight-links-validator": "0.26.0",
    "unist-util-visit": "5.1.0",
    "yaml": "2.9.1"
  }

View site/package.json on GitHub

APM and Skills for Fabric are pinned here:

lockfile_version: '1'
apm_version: 0.31.0
dependencies:
- repo_url: microsoft/skills-for-fabric
  name: skills-for-fabric
  host: github.com
  resolved_commit: 65bfb5eb2c488cd03183df88f76ac3de65dcb910
  resolved_ref: v0.3.17
  version: 0.3.17
  package_type: skill_bundle
  deployed_files:
  - .agents/skills/semantic-model-authoring
  - .agents/skills/semantic-model-authoring/SKILL.md
  - .agents/skills/semantic-model-authoring/apm.yml
  - .agents/skills/semantic-model-authoring/references/connection-binding.md
  - .agents/skills/semantic-model-authoring/references/dax-guidelines.md
  - .agents/skills/semantic-model-authoring/references/dax-perf-decision-guide.md
  - .agents/skills/semantic-model-authoring/references/dax-perf-patterns.md

View apm.lock.yaml on GitHub

The Python toolchain uses hashed requirements. Key versions include DuckDB 1.5.5, jsonschema 4.26.0, PyYAML 6.0.3, pytest 9.1.1 in dev, and ruff 0.16.7 in dev.

The CI workflow pins action SHAs and currently uses Python 3.12 and Node 22:

        os: [ubuntu-latest, windows-latest]
    runs-on: ${{ matrix.os }}
    timeout-minutes: 20
    defaults:
      run:
        shell: pwsh
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
        with:
          fetch-depth: 0
          persist-credentials: false

View .github/workflows/ci.yml on GitHub

The cloud agent setup uses Ubuntu, Python 3.12, Node 22, and APM CLI 0.31.0:

jobs:
  copilot-setup-steps:
    runs-on: ubuntu-latest
    timeout-minutes: 20
    permissions:
      contents: read
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1

      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
        with:
          python-version: "3.12"

      - uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
        with:
          node-version: "22"

      - name: Python tools and validation
        shell: pwsh
        run: ./scripts/Initialize-DevEnvironment.ps1

      # Installs the pinned Skills for Fabric skill from apm.yml into .agents/skills, where the
      # cloud agent discovers project skills. --only apm skips the MCP servers that package declares.
      - name: Skills for Fabric (APM)
        run: |
          python -m venv "$RUNNER_TEMP/apm"
          "$RUNNER_TEMP/apm/bin/pip" install --disable-pip-version-check --quiet apm-cli==0.31.0
          "$RUNNER_TEMP/apm/bin/apm" install --only apm --target copilot

View .github/workflows/copilot-setup-steps.yml on GitHub

Product scope not covered

The labs do not cover:

Fabric workspaces, deployment pipelines, or production deployment.
Power BI Copilot or Fabric data agents against a published model.
Fabric ontology or Fabric IQ modeling.
Tenant governance, sensitivity labels, endorsement, lineage, and monitoring at scale.
Real data, PHI, customer data, or regulated deployments.

See Roadmap for phase two ideas.