Twitter Poster Data

TwitterPosterData is the minimal node in the MLTask/SocialMan pack, and it's minimal on purpose. Twitter posts have essentially one field you control from the graph - the caption - so this node has exactly one input: caption, a multiline string defaulting to "twitter caption." That's it. No account dropdown (the backend posts to your single connected account), no privacy toggles, no thumbnail, no story option. If you want a post to go out to X/Twitter along with the other networks in a multi-post workflow, this is the whole form.

It outputs twitter_data, which plugs into SocialManPoster's twitter_data input. The node is a pure data bag - it wraps your caption in a dict and passes it along; the poster node sends it to MLTask's API (comfy.api.mltask.com/v1/create-post), uploads your media to a pre-signed URL, and the backend handles the actual Twitter API call. None of that happens on your machine, and none of it happens if you haven't set a token and password in SocialManPoster. See a dialog about a missing token or an expired subscription? That's the poster complaining, and it means this node did its job fine.

The only real decision here is what goes in the caption, and the practical advice is boring but true: put the hashtags in the caption, because there's no separate field for them, and keep the copy within Twitter's character limit - the backend won't trim it for you. If you're posting a video with an eye-catching frame, Twitter doesn't take a cover input in this pack, so the thumbnail field you saw on the TikTok or YouTube forms simply doesn't exist here. What you see is what posts.

Worth knowing: the pack's example workflow wires TwitterPosterData and LinkedinPosterData right next to each other with a single text node each, which is the intended shape - these two are the "just say something" posts, while Facebook, Instagram, Pinterest, TikTok, and YouTube get their own forms. If you don't connect a Twitter node at all, SocialManPoster just skips the platform; there's no requirement to post everywhere.

Installing it

Ships in the MLTask_ComfyUI pack. ComfyUI Manager: search MLTask_ComfyUI, install, restart; or:

cd ComfyUI/custom_nodes
git clone https://github.com/misterjoessef/MLTask_ComfyUI

Dependencies are just requests and aiohttp. The requirements that actually matter - a SocialMan token from mltask.com/user/comfyui, a connected Twitter account, and a subscription for real publishing - live in the service, not in a pip install. If you're testing without spamming followers, leave this node disconnected or skip the finalize by toggling prepare_only on the poster.