Tableau Public MCP Server

Inspiration

The inspiration for this project was to draw on all the amazing talent in the community. It’s so hard to keep up with and tap into at times. I wanted to make this much more accessible to everybody.

What it does

The Tableau Public MCP server connects to the public APIs available on Tableau Public. These do great things, like helping you understand your profile, find your workbooks, and search for different visualisations.

One feature I discovered was downloading workbooks via an API call, making it much easier to understand how a workbook was built. Now, instead of manually downloading and reverse-engineering it, you can have an AI explain how it was made.

How we built it

It was inspired by the current Tableau MCP server. I followed similar design principles, updating existing API calls to my own. Since these APIs are publicly available, authentication isn't an issue.

Challenges we ran into

The API calls change over time. I had been documenting them for about six years. Some of those API calls were old, incorrect, or have been deprecated.

What I’ve done to fix that is, on the actual Tableau Public API document page, I’ve created a status page to check whether these API calls still work. That way, I can be assured the tool calls in MCP will still work as well. https://github.com/wjsutton/tableau_public_api?tab=readme-ov-file#-api-status

Accomplishments that we're proud of

I’m genuinely amazed that I’ve been able to build an MCP server that works and actually does some really cool things. I believe it would be incredibly useful for the community.

There are so many times when I think, “Oh, I’d really like to make this chart,” but I can’t quite remember how to do it, and I have to go googling around for a blog post or something. But now I’ve been able to say, “Hey, I want a radial bar chart.” It will go and search Tableau Public, find some relevant workbooks, download them, explain how they made it, and tell me how to do it for my own data set.

I think it’s just incredibly empowering for any creator out there.

What we learned

It’s not easy to make an MCP server, particularly one that works well—and works well with the client applications too. I’ve spent a lot of time going back and forth, testing how the tool calls work with applications like Claude.

Understanding how to get it to not just view the API calls, but actually do things—like unpack workbooks in its own local file system—has been a learning curve to get over and understand.

What's next for Tableau Public MCP Server

Now that Tableau Public MCP exists, I think we’re in a really good place to start working out how to connect it with the other MCP servers in the Tableau ecosystem.

I know there’s been a lot of talk around the Authoring Extensions API and how that’s able to create Tableau content. If that system could have access to millions of visualisations made by the community, that’s a really good starting point for creating good visualisations through an AI model, which is really exciting to me.