microsoft/skills-for-fabric
20 skills2k installs17.7k starsGitHub
Install
npx skills add https://github.com/microsoft/skills-for-fabricSkills in this repo
1E2e Medallion Architecture>. > **Update Check — ONCE PER SESSION (mandatory)** > The first time this skill is used in a session, run the **check-updates** skill before proceeding. > - **GitHub Copilot CLI / VS Code**: invoke the `check-updates` skill. > - **Claude Code / Cowork / Cursor / Windsurf / Codex**: compare local vs remote package.json version. > - Skip if the check was already performed earlier in this session.141installs2Spark Authoring Cli>. > **Update Check — ONCE PER SESSION (mandatory)** > The first time this skill is used in a session, run the **check-updates** skill before proceeding. > - **GitHub Copilot CLI / VS Code**: invoke the `check-updates` skill. > - **Claude Code / Cowork / Cursor / Windsurf / Codex**: compare local vs remote package.json version. > - Skip if the check was already performed earlier in this session.136installs3Check Updates>. This skill checks for updates to the skills-for-fabric marketplace at the start of each session. The check-updates skill documents workflows, constraints, and examples from SKILL.md for agent-assisted execution.128installs4Spark Consumption Cli>. > **Update Check — ONCE PER SESSION (mandatory)** > The first time this skill is used in a session, run the **check-updates** skill before proceeding. > - **GitHub Copilot CLI / VS Code**: invoke the `check-updates` skill. > - **Claude Code / Cowork / Cursor / Windsurf / Codex**: compare local vs remote package.json version. > - Skip if the check was already performed earlier in this session.126installs5Sqldw Authoring Cli>. > **Update Check — ONCE PER SESSION (mandatory)** > The first time this skill is used in a session, run the **check-updates** skill before proceeding. > - **GitHub Copilot CLI / VS Code**: invoke the `check-updates` skill. > - **Claude Code / Cowork / Cursor / Windsurf / Codex**: compare local vs remote package.json version. > - Skip if the check was already performed earlier in this session.122installs6Sqldw Consumption Clisqldw-consumption-cli executes read-only T-SQL queries against Microsoft Fabric data warehouses, lakehouse SQL endpoints, and mirrored databases using the Go-based sqlcmd CLI tool. Supports schema discovery, row counting, SELECT queries, filtering, aggregation, performance monitoring, and CSV/JSON export. Includes agentic data exploration workflows with step-by-step schema discovery. Uses Entra ID authentication via DefaultAzureCredential. Preferred default for lakehouse queries unless PySpark or Spark DataFrames explicitly requested. Covers connection fundamentals, supported T-SQL surface area, cross-database queries, temporary tables, security, and monitoring via DMVs and Query Insights.118installs7Eventhouse Authoring CliThe eventhouse-authoring-cli skill enables developers to programmatically manage Fabric Eventhouse and KQL Database schemas, data ingestion, and policies through CLI-based KQL management commands. It supports table creation and alteration, bulk and streaming ingestion from storage and OneLake, retention/caching/partitioning policies, materialized views, stored functions, and update policies for automated ETL. Uses az rest with Kusto REST API and requires Admin or Ingestor role access. Ideal for schema deployment scripts, data pipeline automation, and infrastructure-as-code workflows in data engineering contexts.111installs8Eventhouse Consumption Clieventhouse-consumption-cli enables developers to run read-only KQL queries against Microsoft Fabric Eventhouse using `az rest` against the Kusto REST API. It covers schema discovery (`.show tables`), KQL operators (where, summarize, join, render), time-series patterns with `bin()`, and ingestion monitoring. Supports agentic data exploration workflows and JSON export. Requires Azure CLI and jq. Always use temp-file pattern for KQL bodies to avoid pipe escaping issues in bash/PowerShell. Must include time filters on time-series tables and discover schema before querying. Prefer `az rest` for CLI sessions and `has` for indexed term search. Avoid hardcoded URIs, missing `bin()` in aggregations, and inline KQL in shell commands.109installs9Dataflows Authoring CliThe dataflows-authoring-cli skill authors Microsoft Fabric Dataflows Gen2 through write-side CLI calls against Fabric Items and Connections APIs. It creates, updates, deletes, and refreshes dataflows, builds mashup.pq plus queryMetadata definitions, binds connections, configures output destinations to Lakehouse, Warehouse, ADX, or Azure SQL, and runs preview-driven loops with executeQuery and customMashupDocument before save. Tooling centers on az rest with jq, base64 encoding, curl, and uuidgen for per-query GUIDs during new dataflow creation. Agentic workflows cover end-to-end create, modify existing definitions, and iterative M preview with connector references for supportedConnectionTypes and credentialType discovery. Sister skill dataflows-consumption-cli handles executing saved queries and refresh status reads. MUST DO rules require workspace and item ID discovery via COMMON-CLI patterns, base64 definition envelopes, and bounded preview queries. Triggers include create dataflow, preview Power Query M, trigger refresh, bind connection, and configure dataflow output destination annotations.90installs10Search Consumption CliThe search-consumption-cli skill discovers Microsoft Fabric items across workspaces using the Catalog Search API when the containing workspace is unknown. It searches by display name, description, or workspace name text with optional OData-style type filters such as Lakehouse, Report, or SemanticModel, returning item and workspace IDs for downstream API calls. Authentication requires Catalog.Read.All scope via az rest with JSON bodies written to temp files to avoid shell quoting issues. Agentic workflow asks for item keywords, runs search, disambiguates multiple matches by presenting name type and workspace, then returns ids for follow-on skills. Pagination uses continuationToken until null with pageSize up to one thousand. MUST rules include disambiguation before assuming a match and avoiding workspace-only searches because the API returns items not workspaces. AVOID guidance covers inventing filter syntax beyond eq, ne, or, and parentheses, and assuming Dataflow Gen1 or Gen2 appear in catalog index yet. Triggers include which workspace has, find item across workspaces, discover items, and search for semantic models or lakehouses tenant-wide.90installs11Sqldw Operations CliThe sqldw-operations-cli skill analyzes Microsoft Fabric Data Warehouse performance through read-only sqlcmd queries against queryinsights views. It diagnoses slow queries, CPU and storage resource consumers, user activity patterns, baseline regressions, SQL pool pressure windows, cache warmth, and clustering key recommendations for warehouse tables. Prerequisites include Azure CLI authentication, Go sqlcmd with Entra ID, jq for REST parsing, and workspace Admin or Member role because queryinsights retains thirty days of history with up to fifteen minute delay. Connection flows discover warehouse FQDN via Fabric REST, then run templated analyses from query-reference.md covering long-running queries, pressure correlation, cold cache detection, and unsupported feature guardrails. Agentic workflows guide investigators from symptom to SQL evidence to actionable recommendations such as OPTIMIZE, column pruning, capacity scaling, or staggering workloads. Triggers include DW slow query analysis, warehouse CPU consumers, pressure events, cache warmth analysis, and cluster key recommendations.88installs12Dataflows Consumption CliThe dataflows-consumption-cli skill monitors, inspects, and queries saved Microsoft Fabric Dataflows Gen2 through read-only CLI using az rest against Fabric control-plane APIs. Capabilities include listing dataflows, decoding base64 definitions such as mashup.pq and queryMetadata.json, discovering parameters, retrieving refresh status and job history, classifying queries by staging, and executing read-only executeQuery requests that return Apache Arrow IPC responses. Agents must run check-updates once per session, list workspaces then filter by displayName client-side, and remember getDefinition is a POST not GET. Tool stack centers on az CLI for auth and REST, with jq, base64, curl, and bash or pwsh for parsing. The six-step agentic exploration pattern supports chat-with-my-dataflows discovery. Security model documents permission matrix per operation. Boundaries route preview and credential discovery to dataflows-authoring-cli. Use when listing dataflows, inspecting definitions, checking refresh history, executing saved queries, or parsing Arrow responses from Fabric dataflows.86installs13Spark Operations CliThe spark-operations-cli skill diagnoses failed Microsoft Fabric Spark jobs, unhealthy Livy sessions, and performance bottlenecks using Fabric REST APIs and az rest CLI tools. All diagnostic operations are read-only while session cleanup such as stopping zombie sessions requires explicit user confirmation. Coverage includes notebook run failures, pipeline Spark activity triage, OOM and shuffle spill diagnosis, data skew detection, driver and executor log retrieval, Spark Advisor findings, event log copy for local Spark History Server, and stuck Livy session investigation. Skill disambiguation routes new notebook creation to spark-authoring-cli and interactive PySpark to spark-consumption-cli. Prerequisites require listing workspaces and items with JMESPath filtering and always passing resource https://api.fabric.microsoft.com to az rest. Diagnostic workflows progress from job identification through log retrieval, advisor analysis, and root cause classification. Use when a Spark job, notebook run, Lakehouse job, or pipeline activity fails and needs CLI-based triage.86installs14Eventstream Authoring CliThe eventstream-authoring-cli skill creates, wires, and publishes Microsoft Fabric Eventstream real-time ingestion topologies through the Fabric Items REST API. It models graph-based definitions with twenty-five source types including Event Hubs, IoT Hub, CDC connectors, Kafka, and SampleData, eight transformation operators such as Filter, Aggregate, GroupBy, Join, ManageFields, Union, Expand, and SQL, plus four destination types including Lakehouse Delta, Eventhouse, Activator, and Custom Endpoint with DefaultStream and DerivedStream routing. Workflows create empty items, build eventstream.json topologies programmatically, base64-encode definitions, and submit via definition or create-with-definition endpoints with LRO polling patterns from COMMON-CLI. Key rules require every source to have a DefaultStream, operator inputNodes references, DerivedStream inputSerialization, and distinction from Eventhouse KQL skills. Gotchas cover max eleven custom endpoints, naming constraints, and deterministic regeneration of modules. Triggers include create eventstream, deploy topology, CDC source flattening, multi-table fan-out routing, and update eventstream definitions.85installs15Synapse MigrationThe synapse-migration skill ports Azure Synapse Analytics Spark workloads to Microsoft Fabric through API-driven migration workflows without requiring the UI Migration Assistant. It translates mssparkutils calls to notebookutils including the env to runtime namespace change, replaces Linked Services with Fabric Data Connections and OneLake Shortcuts, and migrates Spark Pools, Lake Databases, Notebooks, and Spark Job Definitions across phased orchestration. Authentication uses distinct token audiences for Synapse ARM management plane, Synapse data plane, and Fabric api.fabric.microsoft.com control plane with az rest patterns documented per phase. Resource files cover spark pool to environment mapping, lake database to lakehouse conversion, external HMS migration, connector refactoring for Kusto and Cosmos, library compatibility against Fabric runtime 1.3, validation testing, security governance, and migration report generation with portal links. Context loading guidance instructs agents to read only the phase-specific resource file instead of loading all references upfront. Triggers include migrate from synapse, synapse to fabric, mssparkutils to notebookutils, port synapse noteboo.85installs16Eventstream Consumption CliThe eventstream-consumption-cli skill lists, inspects, and monitors Microsoft Fabric Eventstream real-time ingestion pipelines via the Fabric Items REST API. It discovers Eventstreams in workspaces, fetches definitions, base64-decodes eventstream.json graph topologies, and traces flow from sources through operators to destination nodes. Validation covers source connection IDs, destination wiring, retention policies between one and ninety days, and throughput level settings. Monitoring workflows check health, explain topology to users, and correlate downstream analytics targets such as Eventhouse, Lakehouse, Activator, and Real-Time Hub integrations. Listing uses GET workspace eventstreams endpoints with JMESPath name filters, while definition inspection retrieves and decodes graph JSON for operator and stream analysis. The skill distinguishes Eventstream routing pipelines from Eventhouse KQL consumption skills. Gotchas reference COMMON-CLI workspace discovery, wrong token audience 401 errors, and base64 definition decoding requirements. Triggers include list eventstreams, inspect topology, eventstream health, describe configuration, and check retention or throughput settings.83installs17Databricks MigrationThe databricks-migration skill is a comprehensive guide for moving Databricks workloads to Microsoft Fabric. It provides an exhaustive dbutils to notebookutils substitution table covering filesystem operations, secret scope to Key Vault URL conversion, notebook run and exit, widget replacement with parameter-tagged cells, and library install replacement via Fabric Environments. Unity Catalog three-level namespaces collapse to Lakehouse two-level schemas, DBFS paths convert to OneLake, and Databricks Jobs map to Spark Job Definitions with pipeline orchestration for multi-task DAGs. The workload map also covers Delta Live Tables rewrites as parameterized notebooks plus Data Pipelines, MLflow tracking to Fabric ML Experiments, Delta Sharing to OneLake Shortcuts, and Photon to Native Execution Engine substitution. Widget migration documents parameters cells, pipeline base parameters, and Variable Library patterns because dbutils.widgets has no direct Fabric equivalent. Triggers include migrate from databricks, dbutils to notebookutils, unity catalog migration, or delta live tables fabric. Use whenever teams need systematic Databricks-to-Fabric porting with API mappings and cluster con.82installs18Dataflows Save As Authoring CliThe dataflows-save-as-authoring-cli skill assesses, plans, and executes Power BI Gen1 to Fabric Gen2.1 dataflow save-as operations via az rest and curl against Power BI and Fabric REST APIs. It scans workspaces or entire tenants for Gen1 dataflows, evaluates save-as readiness with seven risk signals including incremental refresh, BYOSA storage, Power Automate triggers, pipeline dependencies, linked entities, DirectQuery, and caller-not-owner, produces a Save-As Readiness Snapshot in markdown and JSON, and invokes saveAsNativeArtifact to create upgraded Gen2.1 copies while preserving originals. Two API audiences require Fabric api.fabric.microsoft.com for Gen2 items and analysis.windows.net powerbi api for Gen1 discovery and save-as. Workflows cover tenant scan, per-dataflow risk assessment, guarded execution with user confirmation, post-save-as integrity validation, and residual Gen1 reference detection. Cannot perform in-place migration, only save-as copy upgrade. Use when discovering Gen1 dataflows, assessing migration readiness, upgrading to Gen2.1, or validating post-save-as integrity.82installs19Activator Authoring CliThe activator-authoring-cli skill is a Microsoft Fabric workflow for building and managing Activator reflex items and alert rules entirely through the Fabric REST API and az rest CLI. It covers workspace and item ID resolution, reflex CRUD on the reflexes endpoint, and rule management via getDefinition and updateDefinition against Base64-encoded ReflexEntities.json payloads. Supported data sources include Eventstream sinks, KQL Eventhouse queries, Digital Twin Builder ontology connections, and Real-time Hub subscriptions, each with dedicated reference schemas for entity assembly. Rule conditions support detection thresholds, aggregation windows, occurrence options, and enrichments, while action types include TeamsMessage, EmailMessage, and FabricItemInvocation. The skill mandates Python json.dumps for building nested JSON strings because PowerShell ConvertTo-Json corrupts payloads, and it requires fabric_lro polling for long-running operations. Triggers include create an alert, notify me when, send a teams message when, or update an activator rule. Use whenever operators need CLI-authored Fabric Activator alerts tied to real-time analytics sources.81installs20Hdinsight MigrationThe hdinsight-migration skill ports Azure HDInsight Spark clusters and Hive workloads to Microsoft Fabric. It replaces legacy HiveContext and standalone SparkContext with pre-instantiated SparkSession, converts WASB and ABFS paths to OneLake abfss URLs via Shortcuts, transforms Hive DDL with STORED AS ORC and external tables to Delta Lake schemas in Fabric Lakehouse, and maps Oozie workflow actions including spark, hive, shell, sqoop, and coordinator to Fabric Pipeline activities and schedule triggers. notebookutils replaces subprocess and HDFS client calls for file and credential operations because HDInsight has no mssparkutils or dbutils equivalent. Migration workload map targets Spark notebooks to Fabric Spark, Hive to Lakehouse SQL endpoint, Oozie to Data Pipelines, WASB to OneLake Shortcuts, and Ranger to workspace roles plus OneLake data access. Prerequisites reference COMMON-CORE, COMMON-CLI, and SPARK-AUTHORING-CORE companion docs. Handoffs route notebook creation to spark-authoring-cli and warehouse DDL to sqldw-authoring-cli. Use when retiring HDInsight clusters, converting WASB paths, migrating Hive DDL, or replacing Oozie coordinators.80installs