mcp-fabric-api
Provides tools for integrating with Git repositories, enabling version control operations such as connecting, committing, updating, and managing credentials for Fabric items.
Provides tools for managing GraphQL APIs in Microsoft Fabric, including creating, updating, and deleting GraphQL API endpoints.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-fabric-api list my workspaces"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-fabric-api
MCP (Model Context Protocol) server for the Microsoft Fabric REST APIs. Built for data engineers and data analysts who want to use AI assistants beyond Copilot — such as Claude, Claude Code, or any MCP-compatible client — to build and manage their Fabric components. Covers workspaces, lakehouses, warehouses, notebooks, pipelines, semantic models, reports, dataflows, eventhouses, eventstreams, reflexes, GraphQL APIs, SQL endpoints, variable libraries, git integration, deployment pipelines, mirrored databases, KQL databases, ML models, ML experiments, copy jobs, external data shares, and environments (Spark compute + libraries, with notebook attach/detach).
Safe by default: This server blocks all destructive operations (create, update, delete) until you explicitly configure the
WRITABLE_WORKSPACESenvironment variable. Read operations always work. SetWRITABLE_WORKSPACES="*"to allow writes to all workspaces, or use patterns to limit access. See Workspace Safety Guard for details.
Prerequisites
Node.js 18+
Access to a Microsoft Fabric workspace
One of:
Azure CLI (
az login) — easiest on WindowsAzure app registration with device code flow enabled — best for Mac / Claude Desktop
Service principal credentials — best for headless / automated scenarios
Related MCP server: Fabric-Analytics-MCP
Quick Start
Windows (Azure CLI):
az login
npx @einlogic/mcp-fabric-apiMac / Claude Desktop (Device Code):
AUTH_METHOD=device-code AZURE_CLIENT_ID=your-app-id AZURE_TENANT_ID=your-tenant-id npx @einlogic/mcp-fabric-apiOn first API call, a sign-in URL and code will appear in the logs. Open the URL in your browser, enter the code, and authenticate.
Setup
Claude Desktop
Add to your Claude Desktop config file:
macOS:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
Windows (uses Azure CLI credentials):
{
"mcpServers": {
"fabric": {
"command": "npx",
"args": ["-y", "@einlogic/mcp-fabric-api"]
}
}
}macOS (uses device code flow):
{
"mcpServers": {
"fabric": {
"command": "npx",
"args": ["-y", "@einlogic/mcp-fabric-api"],
"env": {
"AUTH_METHOD": "device-code",
"AZURE_CLIENT_ID": "your-app-client-id",
"AZURE_TENANT_ID": "your-tenant-id"
}
}
}
}When the server starts, check the Claude Desktop logs for a sign-in prompt:
macOS:
~/Library/Logs/Claude/mcp-server-fabric.logWindows:
%APPDATA%\Claude\logs\mcp-server-fabric.log
The prompt will say: "To sign in, use a web browser to open https://microsoft.com/devicelogin and enter the code XXXXXXX". Complete the sign-in once and the token is cached for the session.
Claude Code CLI
Windows:
claude mcp add fabric -- npx -y @einlogic/mcp-fabric-apimacOS:
claude mcp add fabric -e AUTH_METHOD=device-code -e AZURE_CLIENT_ID=your-app-id -e AZURE_TENANT_ID=your-tenant-id -- npx -y @einlogic/mcp-fabric-apiTo verify it was added:
claude mcp listHTTP Mode (Remote)
For remote deployments, set environment variables:
export TRANSPORT=http
export PORT=3000
export AZURE_CLIENT_ID=your-client-id
export AZURE_CLIENT_SECRET=your-client-secret
export AZURE_TENANT_ID=your-tenant-id
npx @einlogic/mcp-fabric-apiThe server exposes:
POST /mcp— MCP endpoint (StreamableHTTP)GET /mcp— SSE stream for server notificationsDELETE /mcp— Session cleanupGET /.well-known/oauth-protected-resource— OAuth metadata
Authentication Methods
The server supports multiple authentication methods via the AUTH_METHOD environment variable. Choose the method that fits your platform and scenario:
Method |
| Required env vars | Best for |
Azure CLI (default) |
| None | Windows with |
Device Code |
|
| Mac / Claude Desktop |
Client Secret |
|
| Headless / automated |
Interactive Browser |
| None (optional: | Systems with browser access |
Default (Azure CLI): Uses the DefaultAzureCredential chain from the Azure Identity SDK. On a developer machine this picks up credentials from az login. No extra configuration needed. This is the original behavior and works best on Windows where Claude Desktop can access the Azure CLI token cache.
Device Code: On first API call, prints a URL and one-time code to stderr. You open the URL in any browser, enter the code, and sign in with your Azure account. The token is cached in memory for the session. This is the recommended method for Mac users with Claude Desktop, because the Claude Desktop process on macOS cannot access the Azure CLI token cache.
To use device code flow, you need an Azure app registration with "Allow public client flows" enabled:
Go to Azure Portal > App registrations > New registration
Name it (e.g., "Fabric MCP") and register
Under Authentication > Advanced settings, set "Allow public client flows" to Yes
Under API permissions, add
https://api.fabric.microsoft.com/Workspace.ReadWrite.All(or the scopes your tools need)Copy the Application (client) ID and your Directory (tenant) ID
Client Secret: Uses a service principal with client credentials. Requires an Azure app registration with a client secret. Suitable for CI/CD pipelines, automated scripts, or any headless environment where interactive sign-in is not possible.
Interactive Browser: Opens a browser window for OAuth sign-in. Works on systems where the server process can launch a browser. Optional AZURE_CLIENT_ID and AZURE_TENANT_ID can be provided to target a specific app and tenant.
Workspace Safety Guard
Control which workspaces allow write operations (create, update, delete) via the WRITABLE_WORKSPACES environment variable. Only workspaces matching the configured name patterns will permit CUD (Create, Update, Delete) operations. Read operations are never restricted.
Default behavior: When
WRITABLE_WORKSPACESis not set or empty, all destructive operations are blocked. You must explicitly configure this variable to enable writes.
| Behavior |
Not set / empty | All writes blocked (safe default) |
| All workspaces writable |
| Only matching workspaces writable |
Set comma-separated glob patterns:
WRITABLE_WORKSPACES=*-Dev,*-Test,Sandbox*Wildcard examples:
*matches all workspaces (allow everything)*-Devmatches "Sales-Dev", "Finance-Dev"Sandbox*matches "Sandbox-123", "Sandbox-Mike"Exact-Namematches only "Exact-Name" (case-insensitive)
Guarded tools (131 total) — every tool that creates, updates, or deletes workspace items:
Domain | Guarded tools |
Workspace |
|
Folder |
|
Tags |
|
OneLake Data Access Security |
|
SQL Database |
|
KQL Queryset |
|
KQL Dashboard |
|
Lakehouse |
|
Warehouse |
|
Notebook |
|
Pipeline |
|
Semantic Model |
|
Report |
|
Dataflow |
|
Eventhouse |
|
Eventstream |
|
Reflex |
|
GraphQL API |
|
Variable Library |
|
Git Integration |
|
Deployment Pipeline |
|
Mirrored Database |
|
KQL Database |
|
ML Model |
|
ML Experiment |
|
Copy Job |
|
External Data Share |
|
Environment |
|
Not guarded: Read operations (list, get, get_definition, get_bim, get_tmdl), query execution (DAX, KQL, SQL, GraphQL), run/refresh/cancel operations, export operations, and deployment pipeline CRUD (tenant-level, not workspace-scoped).
Claude Desktop config with guard (Windows):
{
"mcpServers": {
"fabric": {
"command": "npx",
"args": ["-y", "@einlogic/mcp-fabric-api"],
"env": {
"WRITABLE_WORKSPACES": "*-Dev,*-Test,Sandbox*"
}
}
}
}Claude Desktop config with guard (macOS):
{
"mcpServers": {
"fabric": {
"command": "npx",
"args": ["-y", "@einlogic/mcp-fabric-api"],
"env": {
"AUTH_METHOD": "device-code",
"AZURE_CLIENT_ID": "your-app-client-id",
"AZURE_TENANT_ID": "your-tenant-id",
"WRITABLE_WORKSPACES": "*-Dev,*-Test,Sandbox*"
}
}
}
}Claude Code CLI with guard:
WRITABLE_WORKSPACES="*-Dev,*-Test" claude mcp add fabric -- npx -y @einlogic/mcp-fabric-apiError when not configured:
WRITABLE_WORKSPACES is not configured. Destructive actions are blocked by default. Set WRITABLE_WORKSPACES to a comma-separated list of workspace name patterns, or "*" to allow all.Error when workspace not in allow list:
Workspace "Production-Analytics" is not in the writable workspaces list. Allowed patterns: *-Dev, *-Test, Sandbox*Debug Logging
Enable verbose debug logging to diagnose API errors, inspect request/response details, and trace long-running operations. All log output goes to stderr (visible in Claude Desktop's log files, never interferes with JSON-RPC on stdout).
Set the LOG_LEVEL environment variable to debug:
Claude Desktop config:
{
"mcpServers": {
"fabric": {
"command": "npx",
"args": ["-y", "@einlogic/mcp-fabric-api"],
"env": {
"LOG_LEVEL": "debug"
}
}
}
}Claude Code CLI:
LOG_LEVEL=debug claude mcp add fabric -- npx -y @einlogic/mcp-fabric-apiWhat gets logged at debug level:
Category | Details logged |
HTTP requests | Method, full URL, request body size in bytes |
HTTP responses | Status code, duration (ms), |
Definition uploads | Part paths, payload types, payload sizes — never payload content |
API errors | Full error body including |
LRO polling | Operation ID, poll count, elapsed time, final status |
Pagination | Page count, items per page, total items |
SQL/KQL queries | Server, database, duration, column/row counts — never query text or result data |
Rate limiting | Retry-after duration, affected endpoint |
Compliance: Debug logging never captures actual data content — no query text, no query results, no definition payloads, no bearer tokens. Only structural metadata (URLs, sizes, counts, timing, error details) is logged.
Viewing logs in Claude Desktop:
macOS:
~/Library/Logs/Claude/mcp-server-fabric.logWindows:
%APPDATA%\Claude\logs\mcp-server-fabric.log
You can also tail the log in real time:
# macOS
tail -f ~/Library/Logs/Claude/mcp-server-fabric.log
# Windows (PowerShell)
Get-Content "$env:APPDATA\Claude\logs\mcp-server-fabric.log" -WaitThe x-ms-request-id value logged with every API error is the key identifier needed when opening a support case with Microsoft for Fabric API issues.
Read vs Write Tool Annotations
Every tool declares MCP ToolAnnotations so MCP clients can distinguish read-only tools from mutations when presenting them in a configuration UI. This lets users enable all read tools in a single toggle while requiring explicit approval for writes.
Annotation | Meaning | Typical tool names |
| Pure read, no state change |
|
| Creates or updates state |
|
| Deletes, cancels, or removes |
|
All 273 tools carry these hints. The workspace safety guard (WRITABLE_WORKSPACES) still applies independently on top of whatever the client permits.
File-Based I/O
To avoid large payloads overwhelming MCP clients, definition tools use file paths instead of inline content. The server reads files from disk when sending definitions to Fabric, and writes files to disk when retrieving definitions from Fabric.
Input tools — the server reads definition files from the specified path and uploads to Fabric:
Tool | Parameter | Description |
|
| Path to model.bim JSON file |
|
| Path to model.bim JSON file |
|
| Directory of |
|
| Directory of |
|
| Directory containing notebook definition files |
|
| Directory containing eventstream definition files |
|
| Directory of PBIR report definition files |
|
| Directory of PBIR report definition files |
|
| Directory of |
|
| Directory of |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of |
|
| Directory of definition files (or inline |
|
| Directory of definition files (or inline |
|
| Directory of environment definition files to seed on create |
|
| Directory of environment definition files |
|
| Path to an |
|
| Path to |
|
| Any local file (binary-safe), uploaded to the lakehouse's Files section via the OneLake Data Access API |
Output tools — the server retrieves definitions from Fabric and writes them to disk:
Tool | Parameter | What gets written |
|
| Single |
|
| TMDL files preserving folder structure |
|
| Notebook definition files |
|
| Lakehouse definition files |
|
| Warehouse definition files |
|
| Pipeline definition files |
|
| Report definition files (report.json, pages, visuals) |
|
| Dataflow definition files |
|
| Eventstream definition files |
|
| GraphQL schema definition files |
|
| Reflex definition files |
|
| Variable library files (variables.json, valueSets/) |
|
| Mirrored database definition files |
|
| KQL database definition files |
|
| Any file (binary-safe) from the lakehouse's Files section, via the OneLake Data Access API |
|
| Raw Livy/driver/executor log text for a notebook's Spark session |
|
| Copy job definition files |
|
| Environment definition files |
|
| SQL database definition files ( |
|
| KQL queryset definition files |
|
| KQL dashboard definition files |
|
| Staging |
|
| Published |
TMDL directory structure example:
/tmp/my-model/
model.tmdl
definition.pbism
definition/
tables/
Sales.tmdl
Product.tmdl
relationships.tmdlDevelopment
git clone https://github.com/your-org/mcp-fabric-api.git
cd mcp-fabric-api
npm install
npm run build
npm start
npm run dev # Watch mode
npm run inspect # Launch MCP InspectorTools (273 total)
Auth (4 tools)
Tool | Description |
| Show current Azure identity, tenant, and token expiry |
| List subscriptions/tenants from local |
| Switch to a different Azure tenant (with rollback on failure) |
| Clear cached tokens to force re-acquisition |
Workspace (11 tools)
Tool | Description |
| List all accessible Fabric workspaces |
| Get details of a specific workspace |
| Create a new workspace |
| Update a workspace's name or description |
| Delete a workspace |
| List all items in a workspace (with optional type filter) |
| List who has Admin/Member/Contributor/Viewer access to a workspace |
| Get a specific role assignment by ID |
| Grant a user, group, service principal, or the entire tenant a workspace role |
| Change a principal's workspace role |
| Remove a principal's workspace role assignment |
Lakehouse (21 tools)
Tool | Description |
| List all lakehouses in a workspace |
| Get lakehouse details (SQL endpoint, OneLake paths) |
| Create a new lakehouse (LRO, schemas enabled by default) |
| Update lakehouse name or description |
| Delete a lakehouse |
| List all tables in a lakehouse (falls back to SQL endpoint for schema-enabled lakehouses) |
| Load data into a table from OneLake (LRO). Not supported for schema-enabled lakehouses |
| Create a OneLake shortcut (file, folder, table, or schema level) with support for multiple target types |
| Get SQL endpoint details |
| Get lakehouse definition (LRO). Writes files to |
| Update lakehouse definition (LRO). Reads from |
| List all OneLake shortcuts in a lakehouse |
| Get details of a specific OneLake shortcut |
| Delete a OneLake shortcut |
| List files/folders under a path in the Files section (OneLake Data Access API) |
| Get metadata (size, last modified, etag) for a file or folder in Files |
| Create a folder in the Files section |
| Upload a local file into the Files section |
| Download a file from the Files section to local disk |
| Move or rename a file or folder within the Files section |
| Delete a file or folder from the Files section |
Warehouse (9 tools)
Tool | Description |
| List all warehouses in a workspace |
| Get warehouse details including connection string and provisioning status |
| Create a new warehouse (LRO) |
| Update warehouse name or description |
| Delete a warehouse |
| Get SQL connection details for a warehouse |
| List all tables in a warehouse |
| Get warehouse definition (LRO). Writes files to |
| Update warehouse definition (LRO). Reads from |
Notebook (15 tools)
Tool | Description |
| List all notebooks in a workspace |
| Get notebook details |
| Create a new notebook (LRO) |
| Update notebook name or description |
| Delete a notebook |
| Get notebook definition (LRO). Writes files to |
| Update notebook definition (LRO). Reads files from |
| Run a notebook on demand |
| Get notebook run status |
| Cancel a running notebook |
| List Spark Livy sessions for a notebook (state, compute sizing, durations) |
| Get details of a specific Livy session |
| Download the Livy/driver/executor log for a Livy session to local disk — the driver log is the closest thing to notebook "results" available via API (unstructured print/display output, stack traces) |
| Attach a Fabric Environment to a notebook by mutating its definition metadata (handles both |
| Remove the attached environment from a notebook |
Pipeline (16 tools)
Tool | Description |
| List all data pipelines |
| Get pipeline details |
| Create a new pipeline |
| Update pipeline name or description |
| Delete a pipeline |
| Run a pipeline on demand |
| Get pipeline run status |
| Cancel a running pipeline |
| List all run instances |
| Get per-activity run details for a pipeline run (status, input, output, error per activity) |
| List pipeline schedules |
| Create a pipeline schedule |
| Update a pipeline schedule |
| Delete a pipeline schedule |
| Get pipeline definition (LRO). Writes files to |
| Update pipeline definition (LRO). Reads from |
Semantic Model (15 tools)
Tool | Description |
| List all semantic models |
| Get semantic model metadata (name, ID, description) — does not return the definition |
| Create a semantic model from a BIM/JSON file (LRO). Reads |
| Create a semantic model from TMDL files (LRO). Reads |
| Update semantic model name or description — does not modify the definition |
| Delete a semantic model |
| Trigger a model refresh (Power BI API) |
| Execute a DAX query (Power BI API) |
| Get definition in BIM/JSON format (LRO). Writes |
| Get definition in TMDL format (LRO). Writes TMDL files to |
| Update definition from BIM/JSON file (LRO). Reads |
| Update definition from TMDL files (LRO). Reads |
| Get refresh history (Power BI API) |
| Take over ownership of a semantic model (Power BI API) |
| Get data sources of a semantic model (Power BI API) |
Report (13 tools)
Tool | Description |
| List all reports |
| Get report details |
| Create a new report from PBIR definition files (LRO). Reads from |
| Update report name or description |
| Delete a report |
| Clone a report (Power BI API) |
| Export report to file format (PDF, PPTX, PNG, etc.) via Power BI API |
| Check report export status |
| Get report definition (LRO). Writes files to |
| Update report definition from PBIR directory (LRO). Reads from |
| Rebind a report to a different semantic model/dataset (Power BI API) |
| Get the list of pages in a report (Power BI API) |
| Get data sources used by a report (Power BI API) |
Dataflow Gen2 (8 tools)
Tool | Description |
| List all Dataflow Gen2 items |
| Get dataflow details |
| Create a new dataflow |
| Update dataflow name or description |
| Delete a dataflow |
| Trigger a dataflow refresh |
| Get refresh job status |
| Get dataflow definition (LRO). Writes files to |
Eventhouse (7 tools)
Tool | Description |
| List all eventhouses |
| Get eventhouse details |
| Create a new eventhouse (LRO) |
| Update eventhouse name or description |
| Delete an eventhouse |
| Get query service URI and connection details |
| Execute a KQL query against a KQL database |
Eventstream (7 tools)
Tool | Description |
| List all eventstreams |
| Get eventstream details |
| Create a new eventstream (LRO) |
| Update eventstream name or description |
| Delete an eventstream |
| Get eventstream definition (LRO). Writes files to |
| Update eventstream definition (LRO). Reads from |
Reflex / Activator (7 tools)
Tool | Description |
| List all Reflex (Activator) items |
| Get reflex details |
| Create a new reflex |
| Update reflex name or description |
| Delete a reflex |
| Get reflex definition (LRO). Writes files to |
| Update reflex definition (LRO). Reads from |
GraphQL API (7 tools)
Tool | Description |
| List all GraphQL API items |
| Get GraphQL API details |
| Create a new GraphQL API |
| Update GraphQL API name or description |
| Delete a GraphQL API |
| Get GraphQL schema definition (LRO). Writes files to |
| Execute a GraphQL query |
SQL Endpoint (4 tools)
Tool | Description |
| List all SQL endpoints |
| Get SQL endpoint details |
| Get TDS connection string |
| Execute a T-SQL query against a lakehouse or warehouse SQL endpoint |
Variable Library (7 tools)
Tool | Description |
| List all variable libraries in a workspace |
| Get variable library details including active value set name |
| Create a variable library, optionally with definition files from |
| Update name, description, or active value set |
| Delete a variable library |
| Get definition (LRO). Writes files (variables.json, valueSets/) to |
| Update definition from directory of |
Git Integration (9 tools)
Tool | Description |
| Get Git connection details for a workspace |
| Get Git status of items (sync state between workspace and remote) |
| Connect a workspace to a Git repository (Azure DevOps or GitHub) |
| Disconnect a workspace from its Git repository |
| Initialize a Git connection after connecting (LRO) |
| Commit workspace changes to the connected Git repository (LRO) |
| Update workspace from the connected Git repository (LRO) |
| Get Git credentials configuration for the current user |
| Update Git credentials configuration for the current user |
Deployment Pipeline (12 tools)
Tool | Description |
| List all deployment pipelines accessible to the user |
| Get details of a specific deployment pipeline |
| Create a new deployment pipeline |
| Update deployment pipeline name or description |
| Delete a deployment pipeline |
| List all stages in a deployment pipeline |
| List all items in a specific stage |
| Assign a workspace to a pipeline stage |
| Unassign a workspace from a pipeline stage |
| Deploy items from one stage to another (LRO) |
| List operations (deployment history) |
| Get details of a specific deployment operation |
Mirrored Database (11 tools)
Tool | Description |
| List all mirrored databases in a workspace |
| Get details of a specific mirrored database |
| Create a new mirrored database (LRO) |
| Update mirrored database name or description |
| Delete a mirrored database |
| Get mirrored database definition (LRO). Writes files to |
| Update definition (LRO). Reads from |
| Start mirroring for a mirrored database |
| Stop mirroring for a mirrored database |
| Get the mirroring status |
| Get mirroring status of individual tables |
KQL Database (7 tools)
Tool | Description |
| List all KQL databases in a workspace |
| Get details of a specific KQL database |
| Create a new KQL database (LRO). Requires a parent eventhouse |
| Update KQL database name or description |
| Delete a KQL database |
| Get KQL database definition (LRO). Writes files to |
| Update definition (LRO). Reads from |
ML Model (5 tools)
Tool | Description |
| List all ML models in a workspace |
| Get details of a specific ML model |
| Create a new ML model (LRO) |
| Update ML model name or description |
| Delete an ML model |
ML Experiment (5 tools)
Tool | Description |
| List all ML experiments in a workspace |
| Get details of a specific ML experiment |
| Create a new ML experiment (LRO) |
| Update ML experiment name or description |
| Delete an ML experiment |
Copy Job (11 tools)
Tool | Description |
| List all copy jobs in a workspace |
| Get details of a specific copy job |
| Create a new copy job |
| Update copy job name or description |
| Delete a copy job |
| Get copy job definition (LRO). Writes files to |
| Update definition (LRO). Reads from |
| Run a copy job on demand |
| Get copy job run status |
| Cancel a running copy job |
| List all run instances for a copy job |
External Data Share (4 tools)
Tool | Description |
| List all external data shares for an item |
| Get details of a specific external data share |
| Create a new external data share for an item |
| Revoke an external data share |
Environment (20 tools)
Fabric Environments are the attachable runtime unit for notebooks and Spark job definitions — they bundle a runtime version, a Spark compute configuration, and a set of libraries. Tools cover the full CRUD surface plus the staging/publish lifecycle.
Tool | Description |
| List all environments in a workspace |
| Get environment metadata including publish state |
| Create a new environment (optionally seeded with a definition from |
| Update environment display name or description |
| Delete an environment |
| Get the environment public definition (LRO). Writes files to |
| Override the environment public definition (LRO). Reads from |
| Trigger a publish to make staging changes effective (LRO) |
| Cancel an ongoing publish operation |
| List staging libraries (published + pending) |
| Export staging external libraries as |
| Upload |
| Upload a |
| Delete a custom library from staging by filename |
| Remove a single external library from staging (one at a time) |
| Get the staging Spark compute configuration |
| Update staging Spark compute (pool, cores, memory, runtime, sparkProperties) |
| List published (currently effective) libraries |
| Get published Spark compute configuration |
| Export published external libraries as |
Folder (6 tools, preview API)
In-workspace folder hierarchy for organizing items.
Tool | Description |
| List folders in a workspace, optionally scoped to a subtree |
| Get details of a specific folder |
| Create a folder (optionally nested under a parent folder) |
| Rename a folder |
| Move a folder to a different parent (or to the workspace root) |
| Delete a folder |
Tags (3 tools)
Tenant-wide tags applied to items for categorization and discovery.
Tool | Description |
| List all tags defined in the tenant |
| Apply one or more existing tags to a workspace item |
| Remove one or more tags from a workspace item |
Catalog (1 tool, preview API)
Tool | Description |
| Search for Fabric items across every workspace the caller can access, without needing to know which workspace they're in |
OneLake Data Access Security (4 tools, preview API)
Fine-grained, table/folder-level read access roles inside a Lakehouse (or other OneLake-backed item) — finer-grained than workspace roles.
Tool | Description |
| List data access roles defined on an item |
| Get details of a specific role |
| Create or update (upsert) a single role without touching other roles on the item |
| Delete a role by name |
SQL Database (10 tools)
Fabric's native transactional SQL database item (OLTP), distinct from Warehouse (OLAP).
Tool | Description |
| List all SQL databases in a workspace |
| Get database details including connection string |
| Create a new database (fresh, or restored from a source/deleted database via |
| Update database name or description |
| Delete a database |
| Get the public definition (dacpac/sqlproj, LRO). Writes files to |
| Update the public definition (LRO). Reads from |
| List deleted databases still eligible for point-in-time restore |
| Start streaming database changes into OneLake as Delta tables |
| Stop mirroring |
KQL Queryset (7 tools)
Saved KQL query tabs associated with a KQL/Eventhouse database.
Tool | Description |
| List all KQL querysets in a workspace |
| Get queryset details |
| Create a new queryset (optionally seeded from |
| Update name or description |
| Delete a queryset |
| Get the definition (LRO). Writes files to |
| Update the definition (LRO). Reads from |
KQL Dashboard (7 tools)
Real-time visualization dashboards built on KQL/Eventhouse data — the Real-Time Intelligence equivalent of a Power BI Report.
Tool | Description |
| List all KQL dashboards in a workspace |
| Get dashboard details |
| Create a new dashboard (optionally seeded from |
| Update name or description |
| Delete a dashboard |
| Get the definition (LRO). Writes files to |
| Update the definition (LRO). Reads from |
License
AGPL-3.0
Available Tools
219 toolsauth_clear_token_cacheADestructive
Clear cached authentication tokens to force re-acquisition on next API call
Inputs: no named parameters.
auth_get_current_accountARead-only
Show the current Azure identity, tenant, and token expiry by decoding the Fabric JWT
Inputs: no named parameters.
auth_list_available_accountsARead-only
List Azure subscriptions and tenants the local user has logged into via 'az login'. Reads local CLI state only — does not query Entra ID or list other users.
Inputs: no named parameters.
auth_switch_tenantA
Switch to a different Azure tenant. Acquires a new token to verify access; rolls back on failure.
Inputs: tenantId (required).
copy_job_cancel_runBDestructive
Cancel a running copy job
Inputs: copyJobId (required), workspaceId (required), jobInstanceId (required).
copy_job_createB
Create a new copy job in a workspace
Inputs: displayName (required), workspaceId (required), description.
copy_job_deleteBDestructive
Delete a copy job
Inputs: copyJobId (required), workspaceId (required).
copy_job_getBRead-only
Get details of a specific copy job
Inputs: copyJobId (required), workspaceId (required).
copy_job_get_definitionARead-only
Get the definition of a copy job (long-running). Writes definition files to the specified output directory.
Inputs: copyJobId (required), workspaceId (required), outputDirectoryPath (required).
copy_job_get_run_statusARead-only
Get the status of a copy job run
Inputs: copyJobId (required), workspaceId (required), jobInstanceId (required).
copy_job_listARead-only
List all copy jobs in a workspace
Inputs: workspaceId (required).
copy_job_list_runsARead-only
List all run instances for a copy job
Inputs: copyJobId (required), workspaceId (required).
copy_job_runC
Run a copy job on demand
Inputs: copyJobId (required), workspaceId (required).
copy_job_updateA
Update a copy job's name or description
Inputs: copyJobId (required), workspaceId (required), description, displayName.
copy_job_update_definitionB
Update a copy job's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: copyJobId (required), workspaceId (required), parts, partsDirectoryPath.
dataflow_createA
Create a new Dataflow Gen2 item in a workspace
Inputs: displayName (required), workspaceId (required), description.
dataflow_deleteADestructive
Delete a Dataflow Gen2 item
Inputs: dataflowId (required), workspaceId (required).
dataflow_getARead-only
Get details of a specific Dataflow Gen2 item
Inputs: dataflowId (required), workspaceId (required).
dataflow_get_definitionARead-only
Get the definition of a Dataflow Gen2 item (long-running). Writes definition files to the specified output directory.
Inputs: dataflowId (required), workspaceId (required), outputDirectoryPath (required).
dataflow_get_refresh_statusBRead-only
Get the status of a dataflow refresh job
Inputs: dataflowId (required), workspaceId (required), jobInstanceId (required).
dataflow_listARead-only
List all Dataflow Gen2 items in a workspace
Inputs: workspaceId (required).
dataflow_refreshA
Trigger a refresh of a Dataflow Gen2 item
Inputs: dataflowId (required), workspaceId (required).
dataflow_updateA
Update a Dataflow Gen2 item's name or description
Inputs: dataflowId (required), workspaceId (required), description, displayName.
deployment_pipeline_assign_workspaceB
Assign a workspace to a deployment pipeline stage
Inputs: stageId (required), pipelineId (required), workspaceId (required).
deployment_pipeline_createC
Create a new deployment pipeline
Inputs: displayName (required), description.
deployment_pipeline_deleteCDestructive
Delete a deployment pipeline
Inputs: pipelineId (required).
deployment_pipeline_deployA
Deploy items from one stage to another in a deployment pipeline (long-running)
Inputs: pipelineId (required), sourceStageId (required), note, items, +2 more.
deployment_pipeline_getBRead-only
Get details of a specific deployment pipeline
Inputs: pipelineId (required).
deployment_pipeline_get_operationBRead-only
Get details of a specific deployment pipeline operation
Inputs: pipelineId (required), operationId (required).
deployment_pipeline_listARead-only
List all deployment pipelines accessible to the user
Inputs: no named parameters.
deployment_pipeline_list_operationsARead-only
List operations (deployment history) for a deployment pipeline
Inputs: pipelineId (required).
deployment_pipeline_list_stage_itemsARead-only
List all items in a specific deployment pipeline stage
Inputs: stageId (required), pipelineId (required).
deployment_pipeline_list_stagesCRead-only
List all stages in a deployment pipeline
Inputs: pipelineId (required).
deployment_pipeline_unassign_workspaceADestructive
Unassign a workspace from a deployment pipeline stage
Inputs: stageId (required), pipelineId (required), workspaceId (required).
deployment_pipeline_updateB
Update a deployment pipeline's name or description
Inputs: pipelineId (required), description, displayName.
environment_cancel_publishADestructive
Cancel an ongoing environment publish operation
Inputs: workspaceId (required), environmentId (required).
environment_createA
Create a new environment in a workspace. Optionally seed with a definition by passing definitionDirectoryPath.
Inputs: displayName (required), workspaceId (required), description, definitionDirectoryPath.
environment_deleteCDestructive
Delete an environment
Inputs: workspaceId (required), environmentId (required).
environment_delete_staging_custom_libraryADestructive
Delete a custom library from environment staging by full filename (e.g., samplelibrary.jar)
Inputs: libraryName (required), workspaceId (required), environmentId (required).
environment_export_published_external_librariesARead-only
Export the published external libraries as an environment.yml file. Writes the yaml to disk.
Inputs: workspaceId (required), environmentId (required), outputFilePath (required).
environment_export_staging_external_librariesARead-only
Export the staging external libraries as an environment.yml file. Writes the yaml to disk.
Inputs: workspaceId (required), environmentId (required), outputFilePath (required).
environment_getARead-only
Get metadata of a specific environment including its publish state
Inputs: workspaceId (required), environmentId (required).
environment_get_definitionARead-only
Get the environment public definition (long-running). Writes definition files to the specified output directory.
Inputs: workspaceId (required), environmentId (required), outputDirectoryPath (required).
environment_get_published_spark_computeARead-only
Get the published (currently effective) Spark compute configuration
Inputs: workspaceId (required), environmentId (required).
environment_get_staging_spark_computeARead-only
Get the staging Spark compute configuration for an environment
Inputs: workspaceId (required), environmentId (required).
environment_import_staging_external_librariesA
Upload an environment.yml file to replace the environment's external libraries list. Reads yaml from a file.
Inputs: workspaceId (required), yamlFilePath (required), environmentId (required).
environment_listARead-only
List all environments in a workspace
Inputs: workspaceId (required).
environment_list_published_librariesARead-only
List published (currently effective) libraries for an environment
Inputs: workspaceId (required), environmentId (required).
environment_list_staging_librariesARead-only
List staging libraries (published + pending) for an environment
Inputs: workspaceId (required), environmentId (required).
environment_publishA
Trigger an environment publish to make staging changes effective (long-running)
Inputs: workspaceId (required), environmentId (required).
environment_remove_staging_external_libraryADestructive
Remove a single external library from environment staging (one at a time)
Inputs: name (required), version (required), workspaceId (required), environmentId (required).
environment_updateA
Update an environment's display name or description
Inputs: workspaceId (required), environmentId (required), description, displayName.
environment_update_definitionB
Override the environment public definition (long-running). Reads definition files from the specified directory.
Inputs: workspaceId (required), environmentId (required), definitionDirectoryPath (required).
environment_update_staging_spark_computeA
Update the staging Spark compute configuration (pool, cores, memory, runtimeVersion, sparkProperties). Set a sparkProperty value to null to remove it.
Inputs: workspaceId (required), environmentId (required), driverCores, driverMemory, +6 more.
environment_upload_staging_custom_libraryA
Upload a custom library file (.jar, .py, .whl, .tar.gz, max 100MB) into environment staging. Reads the file from disk.
Inputs: workspaceId (required), environmentId (required), libraryFilePath (required), libraryName.
eventhouse_createA
Create a new eventhouse in a workspace (long-running operation)
Inputs: displayName (required), workspaceId (required), description.
eventhouse_deleteADestructive
Delete an eventhouse
Inputs: workspaceId (required), eventhouseId (required).
eventhouse_execute_kqlC
Execute a KQL query against an eventhouse's KQL database
Inputs: query (required), database (required), workspaceId (required), eventhouseId (required), +1 more.
eventhouse_getBRead-only
Get details of a specific eventhouse
Inputs: workspaceId (required), eventhouseId (required).
eventhouse_get_sql_endpointARead-only
Get the query service URI and connection details for an eventhouse
Inputs: workspaceId (required), eventhouseId (required).
eventhouse_listARead-only
List all eventhouses in a workspace
Inputs: workspaceId (required).
eventhouse_updateA
Update an eventhouse's name or description
Inputs: workspaceId (required), eventhouseId (required), description, displayName.
eventstream_createA
Create a new eventstream in a workspace (long-running operation)
Inputs: displayName (required), workspaceId (required), description.
eventstream_deleteBDestructive
Delete an eventstream
Inputs: workspaceId (required), eventstreamId (required).
eventstream_getBRead-only
Get details of a specific eventstream
Inputs: workspaceId (required), eventstreamId (required).
eventstream_get_definitionARead-only
Get the definition of an eventstream (long-running). Writes definition files to the specified output directory and returns the list of files written.
Inputs: workspaceId (required), eventstreamId (required), outputDirectoryPath (required).
eventstream_listARead-only
List all eventstreams in a workspace
Inputs: workspaceId (required).
eventstream_updateB
Update an eventstream's name or description
Inputs: workspaceId (required), eventstreamId (required), description, displayName.
eventstream_update_definitionA
Update the definition of an eventstream (long-running). Reads definition files from the specified directory.
Inputs: workspaceId (required), eventstreamId (required), definitionDirectoryPath (required).
git_commit_to_gitA
Commit workspace changes to the connected Git repository (long-running)
Inputs: mode (required), workspaceId (required), items, comment, +1 more.
git_connectB
Connect a workspace to a Git repository (Azure DevOps or GitHub)
Inputs: workspaceId (required), gitProviderDetails (required).
git_disconnectADestructive
Disconnect a workspace from its Git repository
Inputs: workspaceId (required).
git_get_connectionBRead-only
Get the Git connection details for a workspace
Inputs: workspaceId (required).
git_get_credentialsARead-only
Get the Git credentials configuration for the current user in a workspace
Inputs: workspaceId (required).
git_get_statusARead-only
Get the Git status of items in a workspace (shows sync state between workspace and remote)
Inputs: workspaceId (required).
git_initialize_connectionA
Initialize a Git connection after connecting (long-running). Chooses whether workspace or remote content takes precedence.
Inputs: workspaceId (required), initializationStrategy (required).
git_update_credentialsB
Update the Git credentials configuration for the current user in a workspace
Inputs: source (required), workspaceId (required), connectionId.
git_update_from_gitA
Update workspace from the connected Git repository (long-running)
Inputs: workspaceId (required), remoteCommitHash (required), options, workspaceHead, +1 more.
graphql_api_createB
Create a new GraphQL API item in a workspace
Inputs: displayName (required), workspaceId (required), description.
graphql_api_deleteBDestructive
Delete a GraphQL API item
Inputs: workspaceId (required), graphqlApiId (required).
graphql_api_execute_queryB
Execute a GraphQL query against a GraphQL API endpoint. Uses the Power BI scope token.
Inputs: query (required), workspaceId (required), graphqlApiId (required), variables.
graphql_api_getBRead-only
Get details of a specific GraphQL API item
Inputs: workspaceId (required), graphqlApiId (required).
graphql_api_get_definitionARead-only
Get the GraphQL schema definition of a GraphQL API item (long-running). Writes definition files (.graphql, etc.) to the specified output directory and returns the list of files written.
Inputs: workspaceId (required), graphqlApiId (required), outputDirectoryPath (required).
graphql_api_listARead-only
List all GraphQL API items in a workspace
Inputs: workspaceId (required).
graphql_api_updateA
Update a GraphQL API item's name or description
Inputs: workspaceId (required), graphqlApiId (required), description, displayName.
kql_database_createA
Create a new KQL database in a workspace (long-running). Requires a parent eventhouse.
Inputs: displayName (required), workspaceId (required), databaseType (required), parentEventhouseItemId (required), +1 more.
kql_database_deleteBDestructive
Delete a KQL database
Inputs: workspaceId (required), kqlDatabaseId (required).
kql_database_getCRead-only
Get details of a specific KQL database
Inputs: workspaceId (required), kqlDatabaseId (required).
kql_database_get_definitionBRead-only
Get the definition of a KQL database (long-running). Writes definition files to the specified output directory.
Inputs: workspaceId (required), kqlDatabaseId (required), outputDirectoryPath (required).
kql_database_listARead-only
List all KQL databases in a workspace
Inputs: workspaceId (required).
kql_database_updateA
Update a KQL database's name or description
Inputs: workspaceId (required), kqlDatabaseId (required), description, displayName.
kql_database_update_definitionA
Update a KQL database's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: workspaceId (required), kqlDatabaseId (required), parts, partsDirectoryPath.
lakehouse_createA
Create a new lakehouse in a workspace (long-running operation). Schemas are enabled by default (preview).
Inputs: displayName (required), workspaceId (required), description, enableSchemas.
lakehouse_create_shortcutA
Create a OneLake shortcut in a lakehouse at any level (file, folder, table, or schema). For schema-enabled lakehouses (preview): set path='Tables' and name to the desired schema name to create a schema shortcut that imports all Delta tables from the target as a new schema. For table-level shortcuts within a schema: set path='Tables/'. Supports targets: OneLake, ADLS Gen2, Amazon S3, Google Cloud Storage, S3 Compatible, Dataverse, Azure Blob Storage, OneDrive/SharePoint.
Inputs: name (required), path (required), target (required), lakehouseId (required), +2 more.
lakehouse_deleteBDestructive
Delete a lakehouse
Inputs: lakehouseId (required), workspaceId (required).
lakehouse_delete_shortcutADestructive
Delete a OneLake shortcut from a lakehouse
Inputs: lakehouseId (required), workspaceId (required), shortcutName (required), shortcutPath (required).
lakehouse_getBRead-only
Get details of a specific lakehouse including SQL endpoint and OneLake paths
Inputs: lakehouseId (required), workspaceId (required).
lakehouse_get_definitionARead-only
Get the definition of a lakehouse (long-running). Writes definition files to the specified output directory.
Inputs: lakehouseId (required), workspaceId (required), outputDirectoryPath (required).
lakehouse_get_shortcutBRead-only
Get details of a specific OneLake shortcut in a lakehouse
Inputs: lakehouseId (required), workspaceId (required), shortcutName (required), shortcutPath (required).
lakehouse_get_sql_endpointCRead-only
Get the SQL endpoint details for a lakehouse
Inputs: lakehouseId (required), workspaceId (required).
lakehouse_listARead-only
List all lakehouses in a workspace
Inputs: workspaceId (required).
lakehouse_list_shortcutsARead-only
List all OneLake shortcuts in a lakehouse
Inputs: lakehouseId (required), workspaceId (required).
lakehouse_list_tablesARead-only
List all tables in a lakehouse. For schema-enabled lakehouses, automatically falls back to querying INFORMATION_SCHEMA via the SQL endpoint.
Inputs: lakehouseId (required), workspaceId (required).
lakehouse_load_tableA
Load data into a lakehouse table from a file path (long-running operation). Not supported for schema-enabled lakehouses — use sql_endpoint_execute_query with COPY INTO or notebooks instead.
Inputs: pathType (required), tableName (required), lakehouseId (required), workspaceId (required), +3 more.
lakehouse_updateB
Update a lakehouse's name or description
Inputs: lakehouseId (required), workspaceId (required), description, displayName.
lakehouse_update_definitionA
Update a lakehouse's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: lakehouseId (required), workspaceId (required), parts, partsDirectoryPath.
mirrored_database_createA
Create a new mirrored database in a workspace (long-running)
Inputs: displayName (required), workspaceId (required), description.
mirrored_database_deleteADestructive
Delete a mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_getCRead-only
Get details of a specific mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_get_definitionARead-only
Get the definition of a mirrored database (long-running). Writes definition files to the specified output directory.
Inputs: workspaceId (required), mirroredDatabaseId (required), outputDirectoryPath (required).
mirrored_database_get_mirroring_statusBRead-only
Get the mirroring status of a mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_get_tables_mirroring_statusARead-only
Get the mirroring status of individual tables in a mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_listARead-only
List all mirrored databases in a workspace
Inputs: workspaceId (required).
mirrored_database_start_mirroringC
Start mirroring for a mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_stop_mirroringB
Stop mirroring for a mirrored database
Inputs: workspaceId (required), mirroredDatabaseId (required).
mirrored_database_updateB
Update a mirrored database's name or description
Inputs: workspaceId (required), mirroredDatabaseId (required), description, displayName.
mirrored_database_update_definitionA
Update a mirrored database's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: workspaceId (required), mirroredDatabaseId (required), parts, partsDirectoryPath.
ml_experiment_createB
Create a new ML experiment in a workspace (long-running)
Inputs: displayName (required), workspaceId (required), description.
ml_experiment_deleteADestructive
Delete an ML experiment
Inputs: workspaceId (required), mlExperimentId (required).
ml_experiment_getARead-only
Get details of a specific ML experiment
Inputs: workspaceId (required), mlExperimentId (required).
ml_experiment_listARead-only
List all ML experiments in a workspace
Inputs: workspaceId (required).
ml_experiment_updateB
Update an ML experiment's name or description
Inputs: workspaceId (required), mlExperimentId (required), description, displayName.
ml_model_createA
Create a new ML model in a workspace (long-running)
Inputs: displayName (required), workspaceId (required), description.
ml_model_deleteBDestructive
Delete an ML model
Inputs: mlModelId (required), workspaceId (required).
ml_model_getBRead-only
Get details of a specific ML model
Inputs: mlModelId (required), workspaceId (required).
ml_model_listBRead-only
List all ML models in a workspace
Inputs: workspaceId (required).
ml_model_updateA
Update an ML model's name or description
Inputs: mlModelId (required), workspaceId (required), description, displayName.
notebook_attach_environmentA
Attach a Fabric Environment to a notebook so it uses that environment's compute/libraries. Mutates the notebook definition metadata (no local files involved).
Inputs: notebookId (required), workspaceId (required), environmentId (required), environmentWorkspaceId.
notebook_cancel_runADestructive
Cancel a running notebook execution
Inputs: notebookId (required), workspaceId (required), jobInstanceId (required).
notebook_createA
Create a new notebook in a workspace (long-running operation)
Inputs: displayName (required), workspaceId (required), description.
notebook_deleteCDestructive
Delete a notebook
Inputs: notebookId (required), workspaceId (required).
notebook_detach_environmentADestructive
Remove the attached environment from a notebook by deleting the environment binding from its definition metadata.
Inputs: notebookId (required), workspaceId (required).
notebook_getBRead-only
Get details of a specific notebook
Inputs: notebookId (required), workspaceId (required).
notebook_get_definitionARead-only
Get the content/definition of a notebook (long-running). Writes definition files to the specified output directory.
Inputs: notebookId (required), workspaceId (required), outputDirectoryPath (required).
notebook_get_run_statusARead-only
Get the status of a notebook run
Inputs: notebookId (required), workspaceId (required), jobInstanceId (required).
notebook_listARead-only
List all notebooks in a workspace
Inputs: workspaceId (required).
notebook_runC
Run a notebook on demand
Inputs: notebookId (required), workspaceId (required), parameters.
notebook_updateA
Update a notebook's name or description
Inputs: notebookId (required), workspaceId (required), description, displayName.
notebook_update_definitionA
Update the content/definition of a notebook (long-running). Reads definition files from the specified directory.
Inputs: notebookId (required), workspaceId (required), definitionDirectoryPath (required).
pipeline_cancel_runADestructive
Cancel a running pipeline execution
Inputs: pipelineId (required), workspaceId (required), jobInstanceId (required).
pipeline_createB
Create a new data pipeline in a workspace
Inputs: displayName (required), workspaceId (required), description.
pipeline_create_scheduleC
Create a schedule for a pipeline
Inputs: type (required), pipelineId (required), workspaceId (required), startDateTime (required), +6 more.
pipeline_deleteCDestructive
Delete a data pipeline
Inputs: pipelineId (required), workspaceId (required).
pipeline_delete_scheduleBDestructive
Delete a pipeline schedule
Inputs: pipelineId (required), scheduleId (required), workspaceId (required).
pipeline_getCRead-only
Get details of a specific data pipeline
Inputs: pipelineId (required), workspaceId (required).
pipeline_get_definitionARead-only
Get the definition of a data pipeline (long-running). Writes definition files to the specified output directory.
Inputs: pipelineId (required), workspaceId (required), outputDirectoryPath (required).
pipeline_get_run_statusBRead-only
Get the status of a pipeline run
Inputs: pipelineId (required), workspaceId (required), jobInstanceId (required).
pipeline_listARead-only
List all data pipelines in a workspace
Inputs: workspaceId (required).
pipeline_list_runsARead-only
List all run instances for a pipeline
Inputs: pipelineId (required), workspaceId (required).
pipeline_list_schedulesARead-only
List all schedules for a pipeline
Inputs: pipelineId (required), workspaceId (required).
pipeline_runC
Run a data pipeline on demand
Inputs: pipelineId (required), workspaceId (required), parameters.
pipeline_updateA
Update a pipeline's name or description
Inputs: pipelineId (required), workspaceId (required), description, displayName.
pipeline_update_definitionB
Update a data pipeline's definition (long-running). Reads definition files from the specified directory.
Inputs: pipelineId (required), workspaceId (required), definitionDirectoryPath (required).
pipeline_update_scheduleB
Update an existing pipeline schedule
Inputs: pipelineId (required), scheduleId (required), workspaceId (required), enabled, +1 more.
reflex_createB
Create a new Reflex (Activator) item in a workspace
Inputs: displayName (required), workspaceId (required), description.
reflex_deleteBDestructive
Delete a Reflex (Activator) item
Inputs: reflexId (required), workspaceId (required).
reflex_getBRead-only
Get details of a specific Reflex (Activator) item
Inputs: reflexId (required), workspaceId (required).
reflex_get_definitionARead-only
Get the definition of a Reflex (Activator) item (long-running). Writes definition files to the specified output directory and returns the list of files written.
Inputs: reflexId (required), workspaceId (required), outputDirectoryPath (required).
reflex_listARead-only
List all Reflex (Activator) items in a workspace
Inputs: workspaceId (required).
reflex_updateA
Update a Reflex (Activator) item's name or description
Inputs: reflexId (required), workspaceId (required), description, displayName.
reflex_update_definitionA
Update a Reflex (Activator) item's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: reflexId (required), workspaceId (required), parts, partsDirectoryPath.
report_cloneC
Clone a report via the Power BI API
Inputs: name (required), reportId (required), workspaceId (required), targetModelId, +1 more.
report_create_definitionA
Create a new report in a workspace (long-running operation). Reads PBIR definition files from the specified directory.
Inputs: displayName (required), workspaceId (required), definitionDirectoryPath (required), description.
report_deleteBDestructive
Delete a report
Inputs: reportId (required), workspaceId (required).
report_exportA
Export a report to a file format (PDF, PPTX, PNG, etc.) via the Power BI API. Returns an export ID to check status.
Inputs: format (required), reportId (required), workspaceId (required).
report_getCRead-only
Get details of a specific report
Inputs: reportId (required), workspaceId (required).
report_get_datasourcesARead-only
Get the data sources used by a report via the Power BI API
Inputs: reportId (required), workspaceId (required).
report_get_definitionARead-only
Get the full definition of a report (PBIR or PBIR-Legacy format). Writes all definition files (report.json, pages, visuals, etc.) to the specified output directory and returns the list of files written.
Inputs: reportId (required), workspaceId (required), outputDirectoryPath (required).
report_get_export_statusBRead-only
Get the status of a report export operation via the Power BI API
Inputs: exportId (required), reportId (required), workspaceId (required).
report_get_pagesARead-only
Get the list of pages in a report via the Power BI API
Inputs: reportId (required), workspaceId (required).
report_listARead-only
List all reports in a workspace
Inputs: workspaceId (required).
report_rebindA
Rebind a report to a different semantic model/dataset via the Power BI API
Inputs: reportId (required), datasetId (required), workspaceId (required).
report_updateB
Update a report's name or description
Inputs: reportId (required), workspaceId (required), description, displayName.
report_update_definitionA
Update a report's full definition (PBIR or PBIR-Legacy). Reads definition files from the specified directory.
Inputs: reportId (required), workspaceId (required), definitionDirectoryPath (required).
semantic_model_create_bimA
Create a new semantic model with a BIM/JSON definition (long-running). Reads the model.bim JSON from the specified file path.
Inputs: displayName (required), workspaceId (required), definitionFilePath (required), description.
semantic_model_create_tmdlA
Create a new semantic model with a TMDL definition (long-running). Reads .tmdl and .pbism files from the specified directory.
Inputs: displayName (required), workspaceId (required), filesDirectoryPath (required), description.
semantic_model_deleteBDestructive
Delete a semantic model
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_execute_daxARead-only
Execute a DAX query against a semantic model via the Power BI API
Inputs: query (required), workspaceId (required), semanticModelId (required).
semantic_model_get_bimARead-only
Get the full definition of a semantic model in TMSL/BIM JSON format (long-running). Writes the decoded model.bim to the specified output file path and returns the path.
Inputs: workspaceId (required), outputFilePath (required), semanticModelId (required).
semantic_model_get_datasourcesBRead-only
Get the data sources of a semantic model via the Power BI API
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_get_detailsARead-only
Get details (name, ID, description, etc.) of a specific semantic model — does not return the definition
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_get_refresh_historyARead-only
Get the refresh history of a semantic model via the Power BI API
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_get_tmdlARead-only
Get the definition of a semantic model in TMDL format (human-readable, LLM-friendly). Writes all TMDL files to the specified output directory preserving the folder structure and returns the list of files written.
Inputs: workspaceId (required), semanticModelId (required), outputDirectoryPath (required).
semantic_model_listARead-only
List all semantic models in a workspace
Inputs: workspaceId (required).
semantic_model_refreshC
Trigger a refresh of a semantic model via the Power BI API
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_take_overC
Take over ownership of a semantic model via the Power BI API
Inputs: workspaceId (required), semanticModelId (required).
semantic_model_update_bimA
Update a semantic model's full definition from TMSL/BIM JSON (long-running). Reads the model.bim JSON from the specified file path.
Inputs: workspaceId (required), semanticModelId (required), definitionFilePath (required).
semantic_model_update_detailsA
Update a semantic model's name or description — does not modify the definition
Inputs: workspaceId (required), semanticModelId (required), description, displayName.
semantic_model_update_tmdlA
Update a semantic model's definition from TMDL files (long-running). Reads .tmdl and .pbism files from the specified directory.
Inputs: workspaceId (required), semanticModelId (required), filesDirectoryPath (required).
sql_endpoint_execute_queryA
Execute a T-SQL query against a lakehouse or warehouse SQL endpoint
Inputs: query (required), itemId (required), itemType (required), workspaceId (required), +2 more.
sql_endpoint_getBRead-only
Get details of a specific SQL endpoint, including connection string
Inputs: workspaceId (required), sqlEndpointId (required).
sql_endpoint_get_connection_stringARead-only
Get the TDS connection string for a SQL endpoint
Inputs: workspaceId (required), sqlEndpointId (required).
sql_endpoint_listARead-only
List all SQL endpoints in a workspace
Inputs: workspaceId (required).
variable_library_createA
Create a new variable library in a workspace (long-running). Optionally provide a directory containing the definition files (variables.json, settings.json, valueSets/*.json).
Inputs: displayName (required), workspaceId (required), description, definitionDirectoryPath.
variable_library_deleteBDestructive
Delete a variable library
Inputs: workspaceId (required), variableLibraryId (required).
variable_library_getARead-only
Get details of a specific variable library, including its active value set name
Inputs: workspaceId (required), variableLibraryId (required).
variable_library_get_definitionARead-only
Get the definition of a variable library (long-running). Writes all definition files (variables.json, settings.json, valueSets/*.json, .platform) to the specified output directory and returns the list of files written.
Inputs: workspaceId (required), variableLibraryId (required), outputDirectoryPath (required).
variable_library_listARead-only
List all variable libraries in a workspace
Inputs: workspaceId (required).
variable_library_updateB
Update a variable library's name, description, or active value set
Inputs: workspaceId (required), variableLibraryId (required), description, displayName, +1 more.
variable_library_update_definitionB
Update a variable library's definition (long-running). Reads definition files (variables.json, settings.json, valueSets/*.json, .platform) from the specified directory.
Inputs: workspaceId (required), variableLibraryId (required), definitionDirectoryPath (required).
warehouse_createA
Create a new warehouse in a workspace (long-running operation)
Inputs: displayName (required), workspaceId (required), description.
warehouse_deleteBDestructive
Delete a warehouse
Inputs: warehouseId (required), workspaceId (required).
warehouse_getARead-only
Get details of a specific warehouse including connection string and provisioning status
Inputs: warehouseId (required), workspaceId (required).
warehouse_get_definitionARead-only
Get the definition of a warehouse (long-running). Writes definition files to the specified output directory.
Inputs: warehouseId (required), workspaceId (required), outputDirectoryPath (required).
warehouse_get_sql_endpointBRead-only
Get the SQL connection details for a warehouse
Inputs: warehouseId (required), workspaceId (required).
warehouse_listARead-only
List all warehouses in a workspace
Inputs: workspaceId (required).
warehouse_list_tablesARead-only
List all tables in a warehouse
Inputs: warehouseId (required), workspaceId (required).
warehouse_updateB
Update a warehouse's name or description
Inputs: warehouseId (required), workspaceId (required), description, displayName.
warehouse_update_definitionA
Update a warehouse's definition (long-running). Accepts definition parts inline or a directory path.
Inputs: warehouseId (required), workspaceId (required), parts, partsDirectoryPath.
workspace_createA
Create a new Fabric workspace
Inputs: displayName (required), capacityId, description.
workspace_deleteBDestructive
Delete a workspace
Inputs: workspaceId (required).
workspace_getBRead-only
Get details of a specific workspace
Inputs: workspaceId (required).
workspace_listARead-only
List all accessible Fabric workspaces
Inputs: no named parameters.
workspace_list_itemsARead-only
List all items in a workspace, optionally filtered by type
Inputs: workspaceId (required), type.
workspace_updateB
Update a workspace's name or description
Inputs: workspaceId (required), description, displayName.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 219 tools
Most tools are clearly distinguished by a resource prefix and action verb, making the majority of the 219 tools easy to tell apart. Some ambiguity exists around SQL endpoint access (sql_endpoint_get vs lakehouse_get_sql_endpoint vs warehouse_get_sql_endpoint) and semantic model definition format variants (get_bim vs get_tmdl vs get_details), but the descriptions largely resolve these cases.
The naming convention is highly consistent throughout: resource_action in snake_case, with predictable verbs like list, get, create, update, delete, run, and cancel. Even format-specific variants like semantic_model_create_bim vs semantic_model_create_tmdl follow the same structural pattern. This makes the tool surface very predictable despite its size.
219 tools is an extreme count that far exceeds the 3-15 tool range considered well-scoped and exceeds the 50+ threshold for extreme mismatch. Even for a broad Microsoft Fabric API, this single server creates severe context bloat, high selection overhead, and poor agent usability. The surface should be split into multiple focused MCP servers or dramatically consolidated.
The tool set covers a very broad Fabric domain: CRUD for nearly every item type, definition management, run/status/cancel operations, git integration, deployment pipelines, environment management, and auth. Minor gaps exist, such as missing dataflow_update_definition and graphql_api_update_definition, but these can be worked around with recreate or direct API usage. No critical dead ends block common workflows.
Maintenance
Related MCP Connectors
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
Official Microsoft MCP Server to query Microsoft Entra data using natural language
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server for OpenAI API (chat completions, image generation, embeddings) via AceDataCloud
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceA Python-based MCP server that enables interaction with Microsoft Fabric APIs for managing workspaces, lakehouses, warehouses, and tables through natural language.17-
- AlicenseNot gradedqualityDmaintenanceThe Fabric-Analytics-MCP server enables AI agents to interact directly with Microsoft Fabric using natural language. It transforms complex data engineering tasks—such as workspace management, data exploration, and job execution—into intuitive, conversational workflows for LLMs like Claude or GitHub17 npm110MIT
- FlicenseNot gradedqualityDmaintenanceMCP server for reading and analyzing Fabric semantic models. Supports getting model definitions and executing DAX queries against Power BI datasets.17-
- FlicenseCqualityDmaintenanceA Python MCP server that lets you manage Microsoft Fabric through natural language in Claude Code or Claude Desktop, with 77+ tools covering workspaces, lakehouses, warehouses, SQL, DAX, semantic models, notebooks, pipelines, OneLake, and Microsoft Graph.83-