Hi everyone,
I’m currently working on a project to build a KNIME MCP integration for Cursor, aiming for something similar to K-AI but with more robust node support.
I’ve hit a roadblock trying to understand how the XML files for each node are generated. I need to understand this structure to automate their inclusion in my tool.
So far, I’ve tried searching for documentation and have used Cursor to parse the knime-core and knime-base GitHub repositories, specifically looking for any readme.md files or code that explains this generation process, but I haven’t found anything concrete.
Could anyone please point me in the right direction? I’m looking for:
-
Any documentation detailing the node XML structure.
-
Information on the process or code that generates these XML files (e.g., is it a wizard, a build script, a template engine?).
Any help or links to relevant docs would be very much appreciated!
Thanks.
Hi @PanteleyShmele1 ,
There are many ways to form an XML structure. I am not sure if there is a templet or wizard. I just worked once on building custom XML tags.
If you know how you want to show the data there are various ways, this being said it is not one node in my case I used String to XML, String Manipulation and others node that helped in my building of the XML file.
The point is you know how you want the data to be displayed in the output and you see the data in hand. You have to build the logic and explore the nodes and your way to use them.
@prashant7526
I have realised that the XML configs of each node is not standardised across nodes and it will be very difficult to manually reverse engineer each node.
This is a missed opportunity. If there were a standardized, documented schema (or a consistent JSON representation) for node configurations, we could easily build AI agents that “ease” users into KNIME by building the skeleton of their workflows for them, lowering the barrier to entry for non-technical users in KNIME