No Code Dataflow Gen 2 in Microsoft Fabric Using GitHub Copilot and Fabric Skills

Posted on by Reza Rad

What if you could create a complete Dataflow Gen 2 in Microsoft Fabric — with data connections, merge transformations, and a fully built date table — without writing a single line of M code?

Not as a theoretical future capability. Right now. Today.

In this video, I show you exactly how to do that using the combination of GitHub Copilot and Microsoft Fabric Skills. You describe what you want in plain English. The AI connects to your Fabric workspace, reads the data sources, figures out the right transformations, and builds the entire Dataflow Gen 2 for you — including generating the M code underneath.

Let me walk you through what this looks like in practice, and then how to set it up yourself.

Published by Reza Rad | RADACAD

📺 Watch the full video on YouTube: No Code Dataflow in Microsoft Fabric using Github Copilot and Skills

🌐 RADACAD Blog: radacad.com

What You Are Going to See First (Video: 0:00)

Before explaining the setup, let me show you the end result. Seeing it is more convincing than any explanation.

I have a Fabric workspace with a lakehouse. That lakehouse contains an Excel file with three product-related tables: Product, ProductSubcategory, and ProductCategory.

Here is the prompt I sent to GitHub Copilot:

“I want to create a Dataflow Gen 2 in this workspace. I want the name of it to be [name]. I want it to connect to that lakehouse, to that Excel file, and I want the product tables to be combined together to create one product table.”

Notice what I did NOT say: I did not mention which column to use for merging. I did not say anything about a merge transformation. I did not write any M code. I did not open the Power Query editor. I did not navigate to the Fabric portal.

And here is what happened: GitHub Copilot — powered by Fabric Skills — connected to the lakehouse, identified the three tables, figured out the right join columns (SubcategoryID and CategoryID — the fields available in both tables), wrote the M code for the merge steps, and created the Dataflow Gen 2 in my workspace.

Then I added one more sentence to the same prompt:

“I want the date table to be added to this, with the 1st of July to be the start of the fiscal year.”

And Copilot added a complete, fully structured date table to the Dataflow — including all standard date columns (Year, Month, Day, Quarter) and all fiscal columns (Fiscal Year, Fiscal Month, Fiscal Quarter), configured with 1 July as the fiscal year start.

When I went to the workspace and opened the Dataflow, here is what I found:

  • A product table — with the full diagram view showing the three source tables (ProductCategory, ProductSubcategory, Product), the merge steps, and the final combined table
  • The M code — visible in the Advanced Editor, showing all the steps written entirely by the AI from my prompt
  • A date table — fully built with calendar and fiscal columns, exactly as I described

All created entirely from my prompts. Nothing else.

💡 This is the key point of this video: I created a complete Dataflow Gen 2 — with multi-table merges and a fiscal date table — without opening the Power Query editor, without writing a single line of M code, and without navigating to a single portal screen.

What Is Happening Under the Hood? (Video: 2:45)

To understand why this works, you need to understand three components working together.

Component 1: The LLM — GitHub Copilot (or Any AI)

GitHub Copilot is the AI brain. It understands your natural language prompt, reasons about what needs to happen, and writes the commands to make it happen.

But here is the crucial thing: on its own, the LLM cannot connect to your Fabric workspace. It does not know your lakehouses exist. It has no access to your data or APIs. All it can do is talk.

That changes when you connect it to Fabric Skills.

And this is not GitHub Copilot specific. You can use Claude, ChatGPT, or any other LLM that supports agentic experiences. In my video I use GitHub Copilot because it integrates neatly with VS Code — but the Fabric Skills work with any compatible AI.

Component 2: VS Code — The Development Environment

Visual Studio Code is the free, open-source editor where GitHub Copilot runs. It is not Microsoft-specific and not AI-specific — it is simply the most widely used code editor in the world. For this use case, it is where your prompts are typed and where the agent executes the work.

Download VS Code for free at code.visualstudio.com.

Component 3: Fabric Skills — The Intelligence Layer

This is the most important piece. Fabric Skills is a library published by the Microsoft Fabric team themselves — available at github.com/microsoft/skills-for-fabric.

Fabric Skills is a set of instructions for the AI. It tells the AI:

  • Which Fabric REST APIs exist for each workload
  • What each API does and when to call it
  • Best practices — what it must do and what it must avoid
  • Guard rails to prevent the AI from making expensive or irreversible mistakes

For this video, the specific skill being used is the Dataflow Authoring skill. This skill teaches the AI:

  • How to create a new Dataflow Gen 2 in a Fabric workspace
  • How to read data from a lakehouse or other sources
  • How to write merge transformations in M
  • How to generate a date table with calendar and fiscal columns
  • How to apply destinations for the output tables
  • How to edit an existing Dataflow Gen 2 if one already exists

At the time of recording, Fabric Skills contains approximately 23–25 skills covering different areas of Fabric: semantic models, Spark compute, dataflows (authoring, consumption, save-as), report authoring (PBIR), Activator, migration from Synapse, HDInsight, and Databricks, and more.

There is also one skill whose job is to check for updates — every time you ask a question, it verifies that you are running the most current version of the skills. When the Fabric team ships improvements, your skills update automatically.

How to Set This Up (Video: 2:56)

The setup is four steps. Here is exactly what you need:

Step 1: Install VS Code

Download VS Code from code.visualstudio.com. It is completely free. Works on Windows, Mac, and Linux. Install it on your device.

Step 2: Set Up GitHub Copilot

You need a GitHub account and a GitHub Copilot subscription. There is a free tier — you do not need to pay to get started. Install the GitHub Copilot extension inside VS Code. The chat window that appears in VS Code is where you will type your prompts.

Step 3: Install GitHub Copilot CLI

The GitHub Copilot CLI is a command-line tool that enables certain commands needed for installing Fabric Skills. It is included in your GitHub Copilot subscription — no extra cost.

Once installed, you can access it in two ways:

  • Inside VS Code — in the prompt window at the bottom, there is an embedded GitHub Copilot CLI where you can run commands directly
  • Command Prompt / Terminal — type copilot and it brings up the Copilot CLI experience

To check which skills you already have installed, type:

copilot /skills list

To get information about a specific skill — for example the Dataflow Authoring skill:

copilot /skills info dataflow-authoring-cli

Step 4: Install Fabric Skills

The installation command is on the Fabric Skills GitHub page. It is a single command you run in the GitHub Copilot CLI. Once run, Fabric Skills appears as installed and every subsequent prompt you send to GitHub Copilot in VS Code automatically has access to the full Fabric knowledge base.

💡 Tip: Stick to the official Fabric Skills from the Microsoft team. There are community-built skills available — and those can be useful — but the Microsoft-published ones are actively maintained by the Fabric product team. When there are bugs or updates, the team fixes them. That gives you the most reliable and current experience.

Running Your First Dataflow Prompt (Video: 7:00)

Once the setup is complete, everything else follows the same pattern I showed at the start of this video.

In the GitHub Copilot chat window in VS Code, type your prompt:

I want to create a Dataflow Gen 2 in [workspace name]. 
I want the name of it to be [dataflow name]. 
I want it to connect to [lakehouse name] to [Excel file name] 
and I want the product tables to be combined together 
to create one product table.

Copilot reads the Dataflow Authoring skill, identifies what needs to happen, selects the right APIs, authenticates with your Microsoft account (first time only), and executes the workflow step by step.

If the Dataflow already exists, Copilot will ask you: do you want to edit it, or create a new one? This prevents accidental overwrites.

If you want to add more tables or transformations, just continue the conversation:

Also add a date table with 1st of July as the start of the fiscal year.

Copilot extends the Dataflow with the new content — without you having to rebuild from scratch.

What can you do beyond what I showed in this demo?

  • Set data destinations — tell Copilot which Lakehouse or Warehouse the output tables should load into
  • Edit an existing Dataflow — not just create new ones. Modify tables, add steps, change transformations.
  • Any other Power Query transformation — the skill supports the full Power Query transformation library, not just merges and date tables

The Bigger Picture — A Fully No-Code Power BI and Fabric Solution (Video: 9:25)

The Dataflow is one piece. But this same approach — GitHub Copilot + Fabric Skills — can build the entire Power BI and Fabric solution with no code:

  1. Connect to the data source → Dataflow Gen 2 via Fabric Skills (this video)
  2. Build the Dataflow transformations → Fabric Skills Dataflow Authoring skill (this video)
  3. Load results into a Lakehouse → set data destinations via prompt
  4. Build a semantic model → Power BI Modeling MCP Server (covered in my dedicated video: No-Code Power BI using Claude AI and the Power BI Modeling MCP Server)
  5. Write DAX measures → Power BI Modeling MCP Server
  6. Build the report → Report Authoring Skill for PBIR (coming in a future video)

Steps 1 through 6. End to end. No code. No Power Query editor. No Power BI Desktop for modeling. No manual DAX writing. No report canvas clicking.

This is not a concept for the future. Steps 1 through 5 are available today. Step 6 — the report authoring skill — is the newest addition to Fabric Skills and is available now. I will cover that in a dedicated video.

For a deeper understanding of what Fabric Skills are and how they connect to the broader Fabric MCP ecosystem — including the Fabric CLI and Core MCP Server — I covered this in detail in Fabric Insider Ep. 8 with Hasan Abo-Shally from Microsoft.

And for everything you need to know about Dataflow Gen 2 itself — including the differences between Dataflow Gen 1, Gen 2, and Gen 2 CI/CD — I have a full article here: Different Versions of Dataflow in Power BI and Microsoft Fabric.

Key Takeaways

  • GitHub Copilot + Fabric Skills = a complete no-code authoring experience for Dataflow Gen 2 in Microsoft Fabric
  • Fabric Skills is published by the Microsoft Fabric team — free, open source, and actively maintained at github.com/microsoft/skills-for-fabric
  • The Dataflow Authoring skill handles creation, editing, merge transformations, date tables, destinations, and more — all from natural language prompts
  • You are not limited to GitHub Copilot — Claude, ChatGPT, or any LLM that supports agentic experiences can use Fabric Skills
  • The full no-code solution is available today: Dataflow → Lakehouse → Semantic Model → Report — all from prompts
  • Always use the Microsoft-official skills — they are maintained by the product team and auto-update

Reza Rad is a Microsoft Regional Director, Data Platform MVP, Author, and Trainer. He is the co-founder of RADACAD and the author of multiple books on Power BI, Power Query, and Microsoft Fabric. You can follow him on LinkedIn and subscribe to the RADACAD YouTube channel.

Reza Rad on FacebookReza Rad on LinkedinReza Rad on TwitterReza Rad on Youtube
Reza Rad
Trainer, Consultant, Mentor
Reza Rad is a Microsoft Regional Director, an Author, Trainer, Speaker and Consultant. He has a BSc in Computer engineering; he has more than 20 years’ experience in data analysis, BI, databases, programming, and development mostly on Microsoft technologies. He is a Microsoft Data Platform MVP for 12 continuous years (from 2011 till now) for his dedication in Microsoft BI. Reza is an active blogger and co-founder of RADACAD. Reza is also co-founder and co-organizer of Difinity conference in New Zealand, Power BI Summit, and Data Insight Summit.
Reza is author of more than 14 books on Microsoft Business Intelligence, most of these books are published under Power BI category. Among these are books such as Power BI DAX Simplified, Pro Power BI Architecture, Power BI from Rookie to Rock Star, Power Query books series, Row-Level Security in Power BI and etc.
He is an International Speaker in Microsoft Ignite, Microsoft Business Applications Summit, Data Insight Summit, PASS Summit, SQL Saturday and SQL user groups. And He is a Microsoft Certified Trainer.
Reza’s passion is to help you find the best data solution, he is Data enthusiast.
His articles on different aspects of technologies, especially on MS BI, can be found on his blog: https://radacad.com/blog.

Leave a Reply

Your email address will not be published. Required fields are marked *