Im trying the new AI agent to build simple functions and keep getting this message. Couldn't get it to work so far

Thanks for the help!

Omer

But I'm just getting errors in Claude that it can't connect to the MCP Server etc. And from reading a lot of posts on this forum many others have the same issues and most error logs point to the Streaming connection URL / or SSE.

Example here: https://community.xano.com/ask-the-community/post/xano-mcp-server-connection-issue---invalid-content-type-for-sse-endpoint-1KRUQPgrd7dlTYp

And here: https://community.xano.com/ask-the-community/post/how-to-install-xano-mcp-server-in-claude-web-wPXuYQ7W4i4X2ID

I find the official docs to be pretty unclear. How exactly should the url be built up, the docs show:

For the Xano MCP Server or a server you built that requires authentication, you’ll need to embed the token in the server URL, as shown below.
<url>/<token>/streaming

So if the Streaming connection is: https://xxxx-yyyy-xxx.yy.xano.io/x2/mcp/meta/mcp/stream

Should it then be: https://xxxx-yyyy-xxx.yy.xano.io/x2/mcp/meta/mcp/stream/<token>/streaming ?

I've tried every combination there is (I think?) and pasted according to this
For Claude Pro and Max Plans

1. Navigate to Settings > Connectors.

2. Locate the “Connectors” section.

3. Click “Add custom connector” at the bottom of the section.

4. Add your connector’s remote MCP server URL.

5. Finish configuring your connector by clicking “Add.”

Would love to hear from someone on the Xano team as there has been plenty of reports on this.
Thanks in advance,

I created a channel called example, which behaves as a nested channel (example/*). My goal is to have a trigger that captures all messages sent to any subchannel under example/*, so that I can process and store those messages on the backend.

However, the current behavior is that messages are successfully exchanged between clients in subchannels such as example/abc or example/123, but the trigger configured in Xano is never executed for those messages.

In a discussion with the Xano AI agent, I was told that this might be happening because the channel configured in the trigger is set to example instead of example/*. The issue is that the channel appears to be automatically defined by Xano, and even when attempting to modify it via XanoScript, I can only remove the channel name but not change it to a wildcard pattern like example/*.

So my question is whether it is currently possible to configure a Realtime trigger that works with nested channels using a wildcard pattern such as example/*, or if there is a limitation that only allows triggers to be associated with exact channel names. If wildcard support is not available, what would be the recommended approach to capture messages from multiple subchannels under the same parent channel?

FA

Since the latest update, I’ve been experiencing an issue with my triggers. They were working perfectly before, but now they no longer seem to fire at all.

I haven’t made any changes to the related logic, so I’m wondering if this could be linked to the recent update.

Has anyone else encountered the same problem or found a workaround?

Thanks in advance for your help!

Xano automatically appends a Set-Cookie: XNS=...; Path=/socket/; Secure; HttpOnly; SameSite=Strict header to every API response — including read-only, public endpoints with no authentication or Realtime features enabled. This header cannot be removed using Xano's built-in HTTP Header function step, as it is injected at the infrastructure level after the function stack runs.

This causes a critical caching problem: most CDNs (e.g., Cloudflare, Fastly, AWS CloudFront) will not cache any response that contains a Set-Cookie header, since cookies are typically user-specific. As a result, even endpoints with response caching enabled in Xano cannot benefit from CDN-level caching. This defeats the purpose of using a CDN and puts unnecessary load on the Xano instance for every request.

Proposed Solution

Please provide one or more of the following options:

1. A workspace-level toggle to disable the XNS cookie on workspaces that do not use the Realtime feature.

2. 2. A way to suppress or strip the Set-Cookie header on specific API endpoints via the HTTP Header function step.

3. 3. At minimum, official documentation clarifying whether this behavior is intentional or a bug, so users can plan their infrastructure accordingly.

1. Is XNS a Realtime-feature cookie, or is it set by Xano infrastructure (load balancer / sticky session) regardless of feature settings?

2. Is there a way to disable it on a workspace where Realtime isn't used?

3. If not, is this configurable in any other way?

Any clarification would be much appreciated. Thanks!

The new "publish" button on top is just too similar to the regular publish button for the given function/api you are working on.

I published an api version and the saw only the new button

But im used to seeing the green button only when what Im working on is not yet published, so i clicked on the new publish button

And even though you mentioned seomthing about the drafts, I figured its not related and just hit the button

And then my heart dropped as i figured i published 26 drafts all across my workspace! literally my heart dropped. I had to go on to a full QA review for my app to see what got effected as I didnt remember which drafts i even had

Pleae make the ui super different between these two buttons. Publishing a specific api VS publish all of your drafts is a huge difference.. I would make that button say "publish all drafts" and give it a different color....

Thanks!

I enabled Git synchronization yesterday, and while it initially worked, I’ve encountered an issue with new tables.

Currently, whenever I try to add a row of data to a newly created table, I receive an error and the value fails to save. This issue seems isolated to new tables, as existing ones are functioning correctly. I am working exclusively on the live branch due to my legacy plan.

Could you please look into why these new tables aren't syncing or accepting data?