I think that anything in Plan 9 that involves a filesystem makes total sense. It's just too easy to just make a thin layer that works with files.

For instance, I was once building a GUI for a project in Plan 9 C, that communicated with the remote server using REST requests. All I had to do is use the webfs filesystem interface. It works somewhat like this:

• You open /mnt/web/clone for reading and keep it open. Reading that file gives you a handle number for your connections.
• Open /mnt/web/%d/ctl (%d being the handle number) for writing. Then you write a few textual information for your request, e.g. request get for the method, url https://my-website.com/ for URL, headers blabla for each header parameter, etc.
• If you're sending data, set the contenttype through the ctl file as above, then write the request body to /mnt/web/%d/postbody.
• Open /mnt/web/%d/body for reading. This action performs the actual request, and may take some time. Nevertheless, for your program, it is simply a blocking file operation.

I think this illustrates how simple Plan 9 can be, so I think it is possible to port most, if not every part of 9fans.net libraries to native Plan 9 by reimplementing a few functions to do file operations instead of using the plan9port binaries.