fabric-lakehouse

name: fabric-lakehouse description: "Create and manage Microsoft Fabric Lakehouses with Delta tables, OneLake shortcuts, and schema definitions. Covers lakehouse.metadata.json, shortcuts.metadata.json, .platform config, and table schemas. USE FOR: create Fabric lakehouse, Delta Lake tables, OneLake shortcuts, lakehouse schema, lakehouse metadata, table columns, Fabric storage, lakehouse CI/CD, dbo schema, shortcut config. DO NOT USE FOR: deploying lakehouses (use fabric-deployment), querying data (use fabric-notebook), semantic models on lakehouse (use fabric-semantic-model)."

Fabric Lakehouse Skill

Authoritative guidance for authoring Microsoft Fabric Lakehouse definitions stored in Git-integrated workspaces.

Triggers

• User asks to create or modify a Fabric Lakehouse
• User asks about Delta table schemas in Fabric
• User asks about OneLake shortcuts
• User asks about lakehouse.metadata.json or shortcuts.metadata.json

Rules

1. File structure — Each lakehouse lives in a folder named <DisplayName>.Lakehouse/ containing:

   • .platform — JSON metadata with type: "Lakehouse", displayName, and config.logicalId
   • lakehouse.metadata.json — Table schema definitions
   • shortcuts.metadata.json — OneLake shortcut configurations

2. .platform schema — Must reference the Fabric Git integration schema:

   {
     "$schema": "https://developer.microsoft.com/json-schemas/fabric/gitIntegration/platformProperties/2.0.0/schema.json",
     "metadata": {
       "type": "Lakehouse",
       "displayName": "<name>"
     },
     "config": {
       "version": "2.0",
       "logicalId": "<unique-guid>"
     }
   }
   

3. Table schema format — lakehouse.metadata.json defines tables with columns, data types, and nullability:

   {
     "defaultSchema": "dbo",
     "tables": [
       {
         "name": "Customer",
         "schema": "dbo",
         "columns": [
           { "name": "CustomerID", "dataType": "int", "isNullable": false },
           { "name": "FirstName",  "dataType": "string", "isNullable": false },
           { "name": "Email",      "dataType": "string", "isNullable": true }
         ]
       }
     ]
   }
   

4. Supported data types — Use these data types in column definitions:

   • int — Integer values
   • long — 64-bit integers
   • string — Text values
   • decimal(p,s) — Decimal with precision and scale (e.g., decimal(10,4))
   • date — Date values
   • datetime — Date and time values
   • boolean — True/false values

5. Shortcuts format — shortcuts.metadata.json is an array of shortcut definitions:

   [
     {
       "name": "publicholidays",
       "path": "/Tables/dbo",
       "target": {
         "type": "OneLake",
         "oneLake": {
           "path": "Tables/dbo/publicholidays",
           "itemId": "<lakehouse-guid>",
           "workspaceId": "<workspace-guid>",
           "artifactType": "Lakehouse"
         }
       }
     }
   ]
   

6. Shortcut paths — Shortcuts can target:

   • /Tables/dbo — For table-level shortcuts (Delta tables)
   • /Files — For file-level shortcuts (unstructured data)

7. Cross-workspace references — Shortcuts reference items by workspaceId and itemId GUIDs. These values are parameterized via config/parameter.yml for multi-environment deployments.

8. Deployment order — Lakehouses must be deployed before Notebooks, SemanticModels, and Reports that depend on them. The fabric-cicd library handles this ordering via the item_type_in_scope list: ["Lakehouse", "Notebook", "SemanticModel", "Report"].

Table Design Guidelines

Output

• <Name>.Lakehouse/.platform — Platform metadata JSON
• <Name>.Lakehouse/lakehouse.metadata.json — Table schemas
• <Name>.Lakehouse/shortcuts.metadata.json — OneLake shortcuts (optional)