You: "Create a sales report from this CSV with revenue by region and monthly trends"
AI: [uploads data, creates semantic model with measures, builds report with charts]

Result: Working report in Fabric with:
  ✓ Data loaded into Lakehouse
  ✓ Semantic model with correct measures (TotalRevenue, TotalProfit, etc.)
  ✓ Visuals bound to the right fields
  ✓ Ready for you to apply your preferred colors/theme in Fabric portal

git clone https://github.com/yourorg/fabric-mcp-server-powerbi-creator.git
cd fabric-mcp-server-powerbi-creator

# Create virtual environment
python -m venv .venv
.venv\Scripts\Activate.ps1  # Windows PowerShell

# Install
pip install -e .
pip install azure-storage-file-datalake azure-identity

python -c "import fabric_mcp; print('OK')"

az login

{
  "mcpServers": {
    "fabric": {
      "command": "fabric-mcp-server-powerbi-creator"
    }
  }
}

{
  "mcpServers": {
    "fabric": {
      "command": "C:/path/to/project/.venv/Scripts/fabric-mcp-server-powerbi-creator.exe"
    }
  }
}

You: "Create a sales dashboard from sales-data.csv. Show revenue by region 
     as a bar chart and monthly trends as a line chart."

AI: "I'll need your Fabric workspace URL and Azure DevOps repo URL."

You: [paste URLs]

AI: [uploads CSV → creates Delta table → builds semantic model → 
     generates report with bar chart + line chart → deploys to Fabric]

"Done! Your report 'SalesReport' is live in Fabric."

You: "I want my report to look like this:" [attaches screenshot of existing report]

AI: "I can see card KPIs at top, donut chart on left, and trend line on right. 
     I'll create the data model and visuals with that layout."

[AI analyzes the screenshot and generates visuals that mimic the layout 
 and chart types - with your actual data wired up automatically]

"Deployed! Check Fabric to see the report. You can apply colors and 
 formatting in the Fabric portal."

1. list_workspaces() → Extract workspace ID from URL
2. list_lakehouses(workspace_id) → Find available Lakehouses
3. upload_csv_to_lakehouse(workspace_id, lakehouse_id, "sales-data.csv")
   → Writes to OneLake Files/
4. load_csv_to_lakehouse(workspace_id, lakehouse_id, "fact_Sales", "Files/sales-data.csv")
   → Creates Delta table
5. deploy_report_with_model(repo_url, branch, model_name, report_name, ...)
   → Pushes TMDL + PBIR to Git
6. sync_workspace(workspace_id)
   → Triggers "Update from Git" in Fabric (async polling)
7. refresh_semantic_model(workspace_id, model_name)
   → Loads data from Lakehouse into model

Result: Live Power BI report in Fabric!

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   Local CSV     │───▶│  OneLake Storage │───▶│  Delta Tables   │
│   (your data)   │    │  (ADLS Gen2)     │    │  (Lakehouse)    │
└─────────────────┘    └──────────────────┘    └────────┬────────┘
                                                        │
                                                        ▼
┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│  Azure DevOps   │◀───│   MCP Server     │───▶│  Fabric API     │
│  Git Repository │    │  (this project)  │    │  (sync/refresh) │
└────────┬────────┘    └──────────────────┘    └─────────────────┘
         │
         ▼
┌─────────────────┐    ┌──────────────────┐
│  Fabric Sync    │───▶│  Workspace Items │
│  (updateFromGit)│    │  (Model + Report)│
└─────────────────┘    └──────────────────┘

# Windows PowerShell
$env:FABRIC_TENANT_ID="your-tenant-id"
$env:FABRIC_CLIENT_ID="your-client-id"
$env:FABRIC_CLIENT_SECRET="your-client-secret"

fabric-mcp-server-powerbi-creator/
├── src/fabric_mcp/
│   ├── server.py           # MCP server + tools
│   ├── fabric_api.py       # Fabric REST API client
│   ├── lakehouse.py        # OneLake + Delta table operations
│   ├── semantic_model.py   # TMDL builder (Direct Lake)
│   ├── report_builder.py   # PBIR builder with visuals
│   └── theme_generator.py  # Power BI theme generator
├── examples/
│   └── deploy_report_e2e.py  # Complete workflow example
├── test-templates/
│   └── sales-dashboard/      # Sample data + template
├── templates/
│   └── *.json               # Power BI theme templates
├── pyproject.toml
└── README.md

# 1. First authenticate
az login

# 2. Edit the example file and fill in YOUR values:
#    - WORKSPACE_ID (from Fabric workspace URL)
#    - LAKEHOUSE_ID (from Lakehouse URL)
#    - REPO_URL (Azure DevOps repo)
#    - BRANCH

# 3. Run the example
cd fabric-mcp-server-powerbi-creator
$env:PYTHONPATH = "src"
python examples/deploy_report_e2e.py