Does anyone has any experience with this Umbraco MCP package? Im currently testing it out but would be really intersted to set it up in a hosted Staging environment for testing. But the open SSE Path which can not be secured through and API Key gives me some headaches. Can anyone or the package developer help me out here? https://www.nuget.org/packages/Axendo.ModelContextProtocol.Umbraco

Haven't tried it, nor this one I'm about to link (just wanted to give you an alternative): https://github.com/Matthew-Wise/umbraco-mcp

i'll give it a shot, are there allready some articles about it?

this is all very very new, so no, not that I know of 😅

Not articles just a few linkedin posts showing it off 😄 Phil and I are still working on it

I've had a look at it and i really looks great.

Little crazy ATM so can't do a call, but happy to chat here or on the issue tracker. Be interested to see how you get on 🙂 Contributions are welcome, we just need to workout the best way right now give AI's nature 😄

I'll let you know how far i get. Currently trying to get it to work with my n8n workflow.

You need to configure the MCP with an API user, the MCP server then uses that user and the Manament API to communicate

ive already did that 🙂

We have been using Claude and vscode and they just find it :F

hmm okey

npm run build

mhhhh

Out of curiosity, do you all have any ideas yet of what you'd want to do with this at all?

A client of ours recieves PDF's via E-Mail of Newsletter Content. Through N8N this pdf gets processed to markdown and then rewritten to their own writing style. (This process is already done in n8n and working) Now is the goal to bring it directly to an unpublished Umbraco Document with the use of the umbraco MCP server and to use and approval process to publish it. hope you get the idea of what we're thinking about: Currently seems a bit futuristic but the possibilities are endless...

It's a cool usecase! I am really looking forward to hearing more about this endless stream of ideas! 😁 Not trying to be pessimistic, just saying that my imagination is so limited, I have no idea what good use cases would be!

jup probably things like this will happen but there are also solutions for that i think like scraping the web for news and looking if it is already old coffe or uninteresting. its quite hard guessing where the whole thing will lead us to

Claude desktop on a paid sub at least handles them all, but we do plan on allowing a bit more control on which are active

I'm interested, always looking around for GOOD ideas (that are not just: "I told the AI to write content and publish it" - which is just.. "spam" basically). Anything that can actually do something interesting and novel I want to hear more about!

Love it! 🙂 I'll tell my boss to buy me the pro subscription then 😄

I have written a couple of MCP Servers, one maps to the Teamwork.com API which has over 100 end points. I have grouped the end points into logical groups and when you setup your connection you can include and exclude both the groups and the individual end points which gives quite a bit of flexibility.

Yeah that is the plan, we are also checking the user permissions (in progress) to get the number of them down. The challenge is the cross overs so we might need to do more of a persona type grouping (data types and content are a great example of crossing over logic groups)

@Chris Houston good input, sounds like something really useful!

Thanks, if anyone is interested, my Teamwork-MCP source is here: https://github.com/Vizioz/Teamwork-MCP but the easiest way to run it is using NPX straight from NPM ( https://www.npmjs.com/package/@vizioz/teamwork-mcp )