Intercepting requests to Jupyter notebook server

I would like to intercept requests made by JupyterLab to its notebook server without forking, but by configuring the system in a way that the frontend would make its API calls to my intercepting service that would forward its requests to the regular proxy, i.e. I’d like to run the system without forking JupyterLab, the notebook server, JupyterHub nor the Helm chart with which I manage my instance of JupyterHub. Is this doable somehow, or am I mad?

I don’t have a straight forward answer. Could you explain a bit more why you want to intercept calls/what the goal of this exercise is? Maybe there are other options for achieving that goal (like a jupyter extension).

I’d like to store edits made to notebooks at a different (versioned) backend service, allowing for the entire JupyterHub Kubernetes cluster and its storage to be treated as a transient thing. And I’d like to do this transparently enough that we could support unmodified singleuser images.