I'm not a super experienced programmer or a deep learning wizard, but I've been wanting to make something like this for a long time. Puffin is a Discord bot that uses NLP and zero-shot learning as well as Google's Toxicity API to flag unsavory messages in servers. Each message is analyzed for sentiment, emotion, and custom zero-shot labels that moderators can provide. If scores for certain emotions are too high, the labels are true, or the message seems to be very toxic, the message is added to a moderation queue hosted in MongoDB. Moderators can view and delete this queue at their will, while not having the bot actually delete anything that shouldn't be deleted. The bot is currently hosted on a Raspberry Pi 5, is written in discord.py, and uses the Hugging Face pipeline.In general, from what I've seen, most auto-moderation solutions that use NLP are either paywalled or non-transparent, and I'm hoping to change that a little bit with this bot. Are there any things that I should change? Are there any features that I should add?