This sound interesting. What makes it superior to something that already exists like bukkit or spigot? I've never made a plugin for a completely server side api, but I would probably like to at some point.
Although Spigot optimizes Bukkit, Spigot is not multi-threaded. Since, Spigot/Bukkit is not multi-threaded, that means that it doesn't use all of the resources available in the CPU. It's just wasting there. With Torch, you get a faster server by using all of the resources and threads in the CPU.
Let me compare this with Sponge.
Sponge currently has a implementation in Forge, and Forge takes time to update because of the release of MCP and updating their API. Thus, making Sponge's implementation slower to update.
Our server modification to the default Minecraft Server allows the server to use multiple threads to make use of the resources available in the CPU and making tasks concurrent inside of the threads, thus making the server faster. (unlike Sponge, Bukkit, Spigot)
Let me compare this with Trident and TridentSDK.
The Trident project is more mature, and introduces multi-threading, we still have advantages to Trident as well too.
Injects into NMS (net.minecraft.server)
Torch injects into NMS, thus making it easier and faster to update than Trident because Trident is made from scratch and needs some networking changes for their server software to upgrade to the next version of Minecraft.
[TORCHSRV-4] Implemented a plugin command (/pl, /plugins, /plugin)
[TORCHAPI-11] Torch class has been added methods!
[TORCHAPI-12] ChatColor class has been implemented! (ChatColor.GOLD, etc)
Build 3 will be released today, however we're almost done adding the features for the first full alpha release. So, don't expect tons of more features; because after this build; we'll start optimizing and fixing bugs on the software.