This has everything to do with what I've been trying to achieve with Meshtastic SSB integration! Amazing work @cft.
Not very relevant, but a trick I found to further compress posts, with LoRa in mind, was to use simple string compression (like shorter) on
value.content.text. Doing some test on a random post, I was able to compress 1005 bytes and get 757 bytes. The decompressed string was exactly the same, obviously.