M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
M

Microsoft Data Factory MCP Server

Official
An MCP server for Microsoft Fabric resource discovery and information retrieval.
How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #199 of 438
Stars
★ 45 Very popular
Transport
stdio (local process)
Runtime
.NET 10.0+
Credentials
API key / credential required
License
MIT
Last commit
Tools
51
54FMRS · D

This is an officially maintained Microsoft MCP server built on the official MCP C# SDK for Microsoft Fabric resource discovery and information retrieval, covering authentication, gateways, connections, workspaces, dataflows, capacities, pipelines, copy jobs, and Apache Airflow jobs. It runs over stdio from the NuGet-distributed Microsoft.DataFactory.MCP package and requires .NET 10.0 plus an Azure AD application registration. It fits .NET developers and data engineers already working with Fabric, but several capabilities remain in preview and the tools carry real write permissions for creating, updating, running, deleting, and scheduling resources, so careful authorization and operation are required.

Strongest · Documentation 13/20 Weakest · Reliability 8/20

Reliability
8/20
Security and permissions
10/20
Maintenance
12/20
Documentation
13/20
Setup experience
11/20
Why each score
Reliability 8/20
This is a Microsoft-owned MCP server built on the official MCP C# SDK, with a manifest declaring stdio transport and NuGet distribution, and a README that consistently enumerates the tool set (auth, gateways, connections, workspaces, dataflows, pipelines, copy jobs, Airflow jobs, capacities). However, this is a static, source-only review: no CI workflow files, committed tests, or execution logs are present, so the MCP init handshake and the claim that declared tools match real behavior cannot be verified. The manifest's version field is left as the literal build placeholder "$(PackageVersion)", a concrete packaging defect that could break install resolution. Error handling, timeouts, dependency failures, and edge cases are essentially undocumented, so the score is capped well below the static ceiling at 8.
Security and permissions 10/20
No malware, credential theft, covert exfiltration, or unconfirmed irreversible delete/deploy/remote-command defaults were found; Azure AD interactive and service-principal authentication is a standard pattern. Deductions: the README openly exposes a get_access_token tool that surfaces access tokens into the model context with no described scoping, redaction, or lifetime guidance, a real token-leakage boundary risk; the server ships write tools including create/update/run and delete_airflow_job, with no description of a least-privilege Azure role set or confirmation prompts for dangerous operations; connection creation can target on-premises and virtual-network gateways, expanding the access surface into private networks. Risks are visible but permissions and confirmation are incomplete, and no red line is triggered, so 10.
Maintenance 12/20
Ownership is credible (microsoft org on GitHub) and the MIT license is explicit; the README documents contribution flow, extension points, and an architecture guide, indicating intent to maintain. Against that, no release history, commit cadence, dependency-update policy, or security-response channel (e.g. SECURITY.md) is visible in the supplied material, and 19 open issues with no visible response time suggest triage gaps. Not archived, but governance and versioning evidence is thin, so 12.
Documentation 13/20
Documentation is layered: README covers installation via NuGet/dnx plus a development config, lists all tools and the ui:// resource, links nine topic guides, states prerequisites, and documents optional Claude/ChatGPT skill packs, with Preview features labeled. Deductions: there is no per-tool parameter documentation, no rate-limit/quota/cost discussion, no troubleshooting section, and no explanation of access-token handling or revocation; the unreplaced version placeholders ("$(PackageVersion)" in the manifest, #{VERSION}# in README) create an undocumented substitution step. Usable but with hidden assumptions, so 13.
Setup experience 11/20
The install path is clear: the Microsoft.DataFactory.MCP NuGet package launched via dnx, with concrete client config examples for VS Code (.vscode/mcp.json) and Visual Studio (.mcp.json), plus a dotnet run development mode over stdio. Deductions: prerequisites are heavy (.NET 10.0 plus an Azure AD tenant and app registration with appropriate permissions), authentication setup is delegated to an external doc rather than spelled out, and both the manifest and README still contain version placeholders the user must resolve before the snippet works as copied. Manual runtime and credential preparation keeps this below the static ceiling of 15, at 11.

Static review · not runListed 2026-09-18

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account data

Best for

  • Developers and data engineers using Microsoft Fabric who want to operate Fabric resources directly from MCP-enabled chat in VS Code or Visual Studio.
  • Teams that already have .NET 10.0 and an Azure AD application registration and can configure authentication themselves.
  • Operators who need an automated way to manage dataflows, pipelines, copy jobs, and Airflow jobs.

Not for

  • Users who do not use Microsoft Fabric and only need generic database or file system access.
  • Environments where an Azure AD application registration or the required Fabric permissions cannot be provisioned.
  • Users who have not installed .NET 10.0 or later and do not want to install a runtime.
  • Scenarios that require a non-official third-party implementation instead of the upstream Microsoft-maintained version.

Required permissions

  • Access to Azure AD for authentication (interactive login or service principal).
  • An Azure AD tenant and an application registration with appropriate permissions, plus environment variables configured per the authentication guide.
  • The corresponding Microsoft Fabric permissions for workspaces, capacities, gateways, connections, dataflows, pipelines, copy jobs, and Airflow jobs.
  • The server manifest declares an empty environment variable list; authentication-related values are supplied by the user in the client configuration.

Risks and side effects

  • Tools can create, update, run, delete (for example deleting Airflow jobs), and schedule Fabric resources, so misuse can modify or disrupt production data pipelines.
  • Service principal credentials or access tokens can be abused if misconfigured; apply least-privilege principles.
  • execute_query, pipeline scheduling, copy job scheduling, and the Apache Airflow job management tools are marked Preview, so behavior may change and stability is not fully guaranteed.
  • Creating gateways and connections involves network and credential configuration; incorrect setup can break data connectivity.
  • Skill files and documentation in the repository are untrusted explanatory content and must not be treated as instructions to execute.

Setup

Before you start

Runtime：.NET 10.0+

  1. Ensure .NET 10.0 or later is installed and that you have an Azure AD tenant and application registration with appropriate permissions.
  2. Create an MCP configuration file in your workspace: .vscode/mcp.json for VS Code, or .mcp.json in the solution directory for Visual Studio.
  3. Add the server config: under servers, add DataFactory.MCP with type stdio, command dnx, and args ["Microsoft.DataFactory.MCP", "--version", "#{VERSION}#", "--yes"], replacing #{VERSION}# with the desired version.
  4. Save the file; the server is downloaded automatically and becomes available in your IDE's MCP-enabled chat interface.
  5. For local development you can instead use command dotnet with args ["run", "--project", "path/to/DataFactory.MCP"].
  6. Configure the authentication environment variables described in the authentication guide before calling the tools.
MCP config · JSON 
{
  "servers": {
    "DataFactory.MCP": {
      "type": "stdio",
      "command": "dnx",
      "args": [
        "Microsoft.DataFactory.MCP",
        "--version",
        "#{VERSION}#",
        "--yes"
      ]
    }
  }
}

Works in clients that read an "mcpServers" key. For clients such as VS Code that use a different key, the configurator below converts it.

Terminal
claude mcp add DataFactory.MCP -- dnx Microsoft.DataFactory.MCP --version '#{VERSION}#' --yes

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Check that DataFactory.MCP tools (e.g. authenticate_interactive, list_workspaces) appear in your IDE's MCP tool list, then authenticate and call list_workspaces — a returned workspace list confirms the connection works.

Troubleshooting

  1. Verify that .NET 10.0 or later is installed, otherwise the server cannot start.
  2. Check that command and args in mcp.json or .mcp.json are correct, and that the #{VERSION}# placeholder has been replaced with an actual version in dnx setups.
  3. If tool calls return authentication errors, call get_authentication_status first, then re-run authenticate_interactive or authenticate_service_principal as needed.
  4. Confirm the Azure AD application registration has the required permissions and that environment variables were set per the authentication guide.
  5. Confirm the signed-in identity has access to the target Fabric workspace, capacity, or gateway.
  6. Dataflow refresh or query timeouts are a common issue; consult the performance and chunking skill documents in the repository.
  7. During development, start the server with dotnet run --project to view local logs and error details.

Things to try

Once connected, you can ask your AI assistant things like:

  • List all dataflows in my Microsoft Fabric workspace
  • Help me create a new data source connection
  • Run this pipeline and check its run status
  • Show me the available Fabric capacities

Tools 51

authenticate_interactive writes
Authenticate interactively via Azure AD.
authenticate_service_principal writes
Authenticate using a service principal.
get_authentication_status read-only
Get the current authentication status.
get_access_token read-only
Get an access token.
sign_out writes
Sign out of the current session.
list_gateways read-only
List Microsoft Fabric gateways.
get_gateway read-only
Retrieve information about a specific gateway.
create_virtualnetwork_gateway writes
Create a virtual network gateway.
Show 43 more tools
list_supported_connection_types read-only
List supported connection types.
list_connections read-only
List Microsoft Fabric connections.
get_connection read-only
Retrieve information about a specific connection.
create_connection writes
Create a connection (cloud, on-premises, or virtual network).
list_workspaces read-only
List Microsoft Fabric workspaces.
list_dataflows read-only
List Microsoft Fabric dataflows.
create_dataflow writes
Create a dataflow.
get_dataflow_definition read-only
Retrieve a dataflow definition.
add_connection_to_dataflow writes
Add a connection to a dataflow.
add_or_update_query_in_dataflow writes
Add or update a query in a dataflow.
save_dataflow_definition writes
Save a dataflow definition.
refresh_dataflow_background writes
Refresh a dataflow in the background.
refresh_dataflow_status read-only
Check dataflow refresh status.
execute_query read-only
Execute a query (Preview).
list_capacities read-only
List Microsoft Fabric capacities.
list_pipelines read-only
List Microsoft Fabric pipelines.
create_pipeline writes
Create a pipeline.
get_pipeline read-only
Retrieve information about a specific pipeline.
update_pipeline writes
Update a pipeline.
get_pipeline_definition read-only
Retrieve a pipeline definition.
update_pipeline_definition writes
Update a pipeline definition.
run_pipeline writes
Run a pipeline.
get_pipeline_run_status read-only
Get the run status of a pipeline.
create_pipeline_schedule writes
Create a pipeline schedule (Preview).
list_pipeline_schedules read-only
List pipeline schedules (Preview).
set_pipeline_schedule_enabled writes
Enable or disable a pipeline schedule (Preview).
list_copy_jobs read-only
List Microsoft Fabric copy jobs.
create_copy_job writes
Create a copy job.
get_copy_job read-only
Retrieve information about a specific copy job.
update_copy_job writes
Update a copy job.
get_copy_job_definition read-only
Retrieve a copy job definition.
update_copy_job_definition writes
Update a copy job definition.
run_copy_job writes
Run a copy job.
get_copy_job_run_status read-only
Get the run status of a copy job.
create_copy_job_schedule writes
Create a copy job schedule (Preview).
list_copy_job_schedules read-only
List copy job schedules (Preview).
list_airflow_jobs read-only
List Apache Airflow jobs (Preview).
create_airflow_job writes
Create an Apache Airflow job (Preview).
get_airflow_job read-only
Retrieve information about a specific Apache Airflow job (Preview).
update_airflow_job writes
Update an Apache Airflow job (Preview).
delete_airflow_job destructive
Delete an Apache Airflow job (Preview).
get_airflow_job_definition read-only
Retrieve an Apache Airflow job definition (Preview).
update_airflow_job_definition writes
Update an Apache Airflow job definition (Preview).

Use cases

List and inspect Microsoft Fabric workspaces, capacities, gateways, and connections from the IDE chat interface.
Create and manage Microsoft Fabric dataflows, including adding connections, editing queries, and saving definitions.
Create, update, run, monitor, and schedule Microsoft Fabric pipelines.
Create and manage copy jobs and monitor their run status.
List, create, update, and delete Apache Airflow jobs and their definitions (Preview).
Authenticate through interactive or service principal flows via Azure AD and obtain access tokens.
Execute queries against dataflows for quick validation (Preview).

Supported clients

VS Code
Visual Studio

Listed from the project's documentation, not tested by this site.

Overview

The Microsoft Data Factory MCP Server is a Model Context Protocol server built by Microsoft on the official MCP C# SDK for Microsoft Fabric resource discovery and information retrieval. It exposes authentication plus access to gateways, connections, workspaces, dataflows, pipelines, copy jobs, Apache Airflow jobs, and capacities through a standardized MCP interface. It is distributed as the NuGet package Microsoft.DataFactory.MCP, uses stdio transport, requires .NET 10.0 or later, and needs an Azure AD tenant and application registration with appropriate permissions. Several capabilities (execute_query, pipeline scheduling, copy job scheduling, and Apache Airflow job management tools) are marked as preview. Optional Claude Skills and ChatGPT Skills files are also provided for Dataflow Gen2 guidance.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 64.1k · Tools 2 Compare with this →
Compare with other MCP servers

Source revision 4ed35b8ae647 Data synced 2026-11-02 Read the FMRS scoring method