FYI I'm working on a [file ID service in Jupyverse](https://github.com/jupyter-server/jupyverse/pull/244) that is based on [watchfiles](https://github.com/samuelcolvin/watchfiles) and fully async.