Are you familiar with Jellyfin?

https://jellyfin.org/

If you can set up a server of this then its as easy as making the library folder the output folder for yt-dlp

https://github.com/yt-dlp/yt-dlp

I didnt even have to write me own script. I gave chatgpt a notepad with channel urls and just told it to write me code to load these urls one by one and download the Last 2 videos. (Trust me you dont want to accidentally download a whole channel). Yt dlp can maintain a log of sort so videos aren't downloaded more then once.

I run this script on a schedule and delete the video when i am done with it. Nice and clean. I can also recommend trying to run an invididious instance for general video browsing but mine took some twiddling to setup right.

https://invidious.io/