Fabric data items
Fabric's data engineering, Data Factory and data warehouse workloads
hold the items below, and each has a reference here. Each reference
holds the depth, and this file picks the item and carries only what
crosses items. Take the steps in order: the wrong item sends every later
step to the wrong file, and Warehouse T-SQL rules applied to a SQL
Database, or the reverse, break code that would have run.
1. Pick the item
By the job asked for, or by the folder of the file in hand:
| The job |
Folder |
Read |
Orchestrating activities: the definition envelope, activity types, policy, deactivation, schedules |
<name>.DataPipeline/ |
data-pipeline.md |
Power Query transforms: mashup.pq, its execute job and publish, CI/CD patterns, a Gen1 upgrade |
<name>.Dataflow/ |
dataflow.md |
| Spark: a notebook, plain Python ones too, a lakehouse and its tables, a Spark job definition, an environment, a slow or failed job, an MLV, AI Functions |
<name>.Notebook/, .Lakehouse/, .SparkJobDefinition/, .Environment/ |
spark.md |
| T-SQL against a Warehouse or a SQL analytics endpoint, a warehouse snapshot, or tracing a query |
<name>.Warehouse/ |
warehouse.md |
| T-SQL against a Fabric SQL Database, its SQL project, its OneLake replica |
<name>.SQLDatabase/ |
sql-database.md |
Near these items, and not them:
- The REST conventions every item call shares, such as auth,
long-running operations and the error catalogue:
rest-api.
- Deployment pipelines, fabric-cicd and a workspace's Git integration:
deploy.
- The variable library a pipeline or dataflow reads:
variable-library.
- SQL
GRANT, row- and column-level security: security.
- KQL, an Eventhouse and the other Real-Time Intelligence items:
rti-items.
- A semantic model over a lakehouse or warehouse, Direct Lake among
them:
pbi-items.
2. What crosses items
- A Warehouse and a SQL Database are two engines. A Warehouse, and
the SQL analytics endpoint every lakehouse carries, run the
distributed engine with
warehouse.md's unsupported types
and features; a SQL Database is Azure SQL Database, where those work
(sql-database.md § Key
Differences from Fabric Warehouse). Name the engine before writing
T-SQL.
- A pipeline runs the other items. The activity's JSON is
data-pipeline.md's
(§ 4:
TridentNotebook, RefreshDataFlow,
SqlServerStoredProcedure); what it runs is the item's own file, and
a Warehouse procedure's call, by the Script or Stored procedure
activity, is warehouse.md
§ Pipeline Integration's.
- A lakehouse's tables are read by two engines. Spark writes and
maintains them (spark.md § Lakehouse
Table Maintenance, with the endpoint's metadata sync); T-SQL reads
them through the endpoint, whose metadata refresh is
sql-endpoint-api.md's.
- One MCP server serves pipelines and dataflows, the DataFactory
server, with one home:
data-pipeline-mcp.md.
The Warehouse's remote server is
warehouse-mcp.md.
3. Read the item's reference first
Read the file step 1 names before answering or editing: limits,
definition shapes and preview status live there, dated, and each names
its own Read first table for its definition, REST calls and MCP
server, then the REST spec's folder, then the live docs, Learn by
microsoft-learn-mcp, or with no server,
curl -sL -H 'Accept: text/markdown' <page URL> (undocumented, read
2026-11-07). A task that spans items, a pipeline that runs a notebook
and a warehouse procedure for one, reads each item's file.
A live page that contradicts a line here wins: say so, act on the page,
and route the fix by capture-learning, bar a dated measured line that
names the page it overrules and the call that measured it, which a page
changed since its date sends back to re-measure. An observed claim is
cited or marked pending a doc check, and stands, marked observed, only
where no page covers it.
1---2name: data-items3description: "Use for Microsoft Fabric data items, in Git or by REST, `fab` or a script: a Data Pipeline (`.DataPipeline`, `pipeline-content.json`, `.schedules`, activity JSON, `/dataPipelines`); a Dataflow Gen2 (`.Dataflow`, `mashup.pq`, `queryMetadata.json`, `/dataflows`, its execute job, Gen1 upgrade); Spark: a notebook (`.Notebook`), PySpark or Spark SQL, a plain Python notebook with no Spark, a lakehouse and its tables, a Spark job definition or environment, a Spark job slow, queued or failed, Livy, a materialized lake view, AI Functions; a Warehouse or SQL analytics endpoint (`.Warehouse`, `/warehouses`, `/sqlEndpoints`, its T-SQL surface and unsupported types, schema evolution, MERGE, Time Travel, snapshots, monitoring); and a Fabric SQL Database (`.SQLDatabase`, `/sqlDatabases`), the Azure SQL engine, which Warehouse's limits do not bind. Picks the item, then reads its reference."4when_to_use: "Use before a Read, edit or REST call on an item of type DataPipeline, Dataflow, Notebook, Lakehouse, SparkJobDefinition, Environment, Warehouse, SQLEndpoint, WarehouseSnapshot or SQLDatabase; before writing T-SQL for a Warehouse or SQL Database, or PySpark in a notebook; or when tracing a Warehouse query (labels, queryinsights, DMVs, KILL, a Capacity Metrics spike, result set caching, a stale SQL endpoint) or a Spark job."5# model: inherit # any model: value blocks Copilot slash invocation6# effort: medium # unset = inherit session effort; there is no 'effort: inherit'7disable-model-invocation: false8metadata:9 source: https://skillmd.com/skills/wardawgmalvicious/data-items10---1112# Fabric data items1314Fabric's data engineering, Data Factory and data warehouse workloads15hold the items below, and each has a reference here. Each reference16holds the depth, and this file picks the item and carries only what17crosses items. Take the steps in order: the wrong item sends every later18step to the wrong file, and Warehouse T-SQL rules applied to a SQL19Database, or the reverse, break code that would have run.2021## 1. Pick the item2223By the job asked for, or by the folder of the file in hand:2425| The job | Folder | Read |26| --- | --- | --- |27| Orchestrating activities: the definition envelope, activity types, `policy`, deactivation, schedules | `<name>.DataPipeline/` | [data-pipeline.md](references/data-pipeline/data-pipeline.md) |28| Power Query transforms: `mashup.pq`, its execute job and publish, CI/CD patterns, a Gen1 upgrade | `<name>.Dataflow/` | [dataflow.md](references/dataflow/dataflow.md) |29| Spark: a notebook, plain Python ones too, a lakehouse and its tables, a Spark job definition, an environment, a slow or failed job, an MLV, AI Functions | `<name>.Notebook/`, `.Lakehouse/`, `.SparkJobDefinition/`, `.Environment/` | [spark.md](references/spark/spark.md) |30| T-SQL against a Warehouse or a SQL analytics endpoint, a warehouse snapshot, or tracing a query | `<name>.Warehouse/` | [warehouse.md](references/warehouse/warehouse.md) |31| T-SQL against a Fabric SQL Database, its SQL project, its OneLake replica | `<name>.SQLDatabase/` | [sql-database.md](references/sql-database/sql-database.md) |3233Near these items, and not them:3435- The REST conventions every item call shares, such as auth,36 long-running operations and the error catalogue: `rest-api`.37- Deployment pipelines, fabric-cicd and a workspace's Git integration:38 `deploy`.39- The variable library a pipeline or dataflow reads:40 `variable-library`.41- SQL `GRANT`, row- and column-level security: `security`.42- KQL, an Eventhouse and the other Real-Time Intelligence items:43 `rti-items`.44- A semantic model over a lakehouse or warehouse, Direct Lake among45 them: `pbi-items`.4647## 2. What crosses items4849- **A Warehouse and a SQL Database are two engines.** A Warehouse, and50 the SQL analytics endpoint every lakehouse carries, run the51 distributed engine with52 [warehouse.md](references/warehouse/warehouse.md)'s unsupported types53 and features; a SQL Database is Azure SQL Database, where those work54 ([sql-database.md](references/sql-database/sql-database.md) § Key55 Differences from Fabric Warehouse). Name the engine before writing56 T-SQL.57- **A pipeline runs the other items.** The activity's JSON is58 [data-pipeline.md](references/data-pipeline/data-pipeline.md)'s59 (§ 4: `TridentNotebook`, `RefreshDataFlow`,60 `SqlServerStoredProcedure`); what it runs is the item's own file, and61 a Warehouse procedure's call, by the Script or Stored procedure62 activity, is [warehouse.md](references/warehouse/warehouse.md)63 § Pipeline Integration's.64- **A lakehouse's tables are read by two engines.** Spark writes and65 maintains them ([spark.md](references/spark/spark.md) § Lakehouse66 Table Maintenance, with the endpoint's metadata sync); T-SQL reads67 them through the endpoint, whose metadata refresh is68 [sql-endpoint-api.md](references/warehouse/references/sql-endpoint-api.md)'s.69- **One MCP server serves pipelines and dataflows**, the DataFactory70 server, with one home:71 [data-pipeline-mcp.md](references/data-pipeline/references/data-pipeline-mcp.md).72 The Warehouse's remote server is73 [warehouse-mcp.md](references/warehouse/references/warehouse-mcp.md).7475## 3. Read the item's reference first7677Read the file step 1 names before answering or editing: limits,78definition shapes and preview status live there, dated, and each names79its own *Read first* table for its definition, REST calls and MCP80server, then the REST spec's folder, then the live docs, Learn by81`microsoft-learn-mcp`, or with no server,82`curl -sL -H 'Accept: text/markdown' <page URL>` (undocumented, read832026-11-07). A task that spans items, a pipeline that runs a notebook84and a warehouse procedure for one, reads each item's file.8586A live page that contradicts a line here wins: say so, act on the page,87and route the fix by `capture-learning`, bar a dated measured line that88names the page it overrules and the call that measured it, which a page89changed since its date sends back to re-measure. An observed claim is90cited or marked pending a doc check, and stands, marked observed, only91where no page covers it.