Rather than md files, I suggest reading in your data through a JSON file.
Here are some example codes for your reference:
1.
https://github.com/dragondive/masadora/blob/b4fcccb10130b0d5a20325e6ea905fcee45c5c88/plantuml/production_controller_sequence.puml
2.
https://github.com/dragondive/plantuml_demo/blob/main/src/preprocessor/README.rst#generate-multiple-customized-diagrams-from-diagram-template
Here is also a somewhat more "fun" example, that also has more complex JSON data structure:
https://github.com/dragondive/plantuml_demo/blob/main/src/preprocessor/README.rst#compute-and-draw-test-cricket-matches-hosting-data-in-hierarchical-structure
As for updating the diagrams dynamically, as far as I know, PlantUML does not have such a facility to monitor files and update diagrams on the fly. Regardless of that, my suggestion would be to keep separate things separate. Do the dynamic updating outside PlantUML. For example, you could upload your JSON data file to Github (or GHE, at work), then setup a Github Actions workflow (or the equivalent tool you use) to trigger PlantUML at regular intervals. If "real time" data updates are important, you could also trigger the workflow whenever the JSON data file changes. [Note: replace Github/Github Actions with the equivalent tools you're using.]
I think you'll be able to figure out the rest. Please feel free to ask if you have more questions.
