The below instructions will allow you to connect your Xano MCP server to Cursor and make them available across any project. The below method does not support authentication. If you need authentication or want to define per-project MCPs, use these instructions instead.
1
Open your Cursor settings
2
Under the Features subsection, scroll down to MCP Servers
3
Click + Add new MCP Server
Give your server a name.
The type should be sse
Paste your server URL in the Server URL section. You can retrieve your server URL by navigating to your server in Xano and clicking Copy Connection URL.
4
You should now see your MCP Server ready in Cursor
5
In your chat window, you can now interact with the tools included in your MCP server that's connected
Make sure the conversation type is set to Agent
In our example, we have a tool that checks if a user is marked as an administrator.
Cursor (per-project)
1
In your project's root directory, create a new folder called .cursor
2
Create a new file inside of that folder called mcp.json
3
Fill out the required details inside of mcp.json
If the file is blank, start with the basic structure and replace the placeholder values with your own.
Claude Desktop currently has a bug related to spaces inside of headers. To mitigate this, change the authentication section of your config file to look like the following:
"Authorization:${AUTH_TOKEN}"
],
"env": {
"AUTH_TOKEN": "Bearer YOUR AUTH TOKEN HERE"
You can add multiple entries if you have multiple MCP servers. See the below example.
Click the Refresh button, and you should see your MCP server(s) available
Methods
Streaming
Streaming connections allow your MCP server to deliver responses in smaller chunks. Great for emulating the typical chatbot experience, where you can provide the experience of the response being 'typed'.
SSE
SSE connections may be more compatible with certain systems, but deliver the entire response as a whole once its ready, instead of in smaller pieces.
Which one should I use?
It depends on your use case and what you're using to connect to the MCP server. Streaming connections allow for a more organic, "chatbot-like" experience and means that your users won't have to wait as long to start seeing the response. However, whatever is connecting to your MCP server needs to support streaming responses for this to work.
FAQ and Troubleshooting
How do I use my MCP server in one of these clients?
Once you're connected following the instructions above, you should be able to converse with the AI like you would any other, asking real-world questions that the tools you built should be able to answer.
For example, if we have a tool that just returns a true or false, we'd probably be asking yes or no questions, such as "Does this user, john@email.com, have administrator access?"
I can't connect to my server from my client of choice.
Check the error message you're receiving for clues. This could be due to one of the following:
An incorrectly formatted configuration file
Your Xano MCP Server is not set to allow connections
You're not providing an authentication token to a server that requires it
For MCP servers with tools that require authentication, after a long waiting period, I get a connection error.
In our current testing, we are finding that running multiple MCP clients causes this issue. Our recommendation is to stick with a single client for the time being.
Try these steps:
Close any open MCP clients
Ensure that your authentication token is not expired