Linkedin Poster Data

LinkedinPosterData is the other "just a caption" node in the MLTask/SocialMan pack, and it's the same shape as TwitterPosterData: one input, one output, no account picker, no options. You get a caption field (multiline, default "linkedin caption"), it gets wrapped into a linkedin_data dict, and SocialManPoster handles everything from there. If you're building a multi-post workflow - the pack's whole reason for existing - this is the node that makes your AI image land on LinkedIn alongside the other networks in a single run.

There's no target_account because the backend posts to your single connected LinkedIn account, same as Twitter and TikTok. There's no privacy toggle because LinkedIn posts don't have one that matters here. There's no thumbnail because LinkedIn doesn't take a cover input in this pack. The single output, linkedin_data, wires into SocialManPoster's linkedin_data input, and that's the entire integration surface.

How it works is refreshingly dull: the node builds {"caption": ...} and passes it along. The poster node then sends the whole bundle - media path, caption, and everything else you connected - to MLTask's API at comfy.api.mltask.com/v1/create-post, uploads your image or video to a pre-signed URL, and the MLTask backend calls LinkedIn's API for you. That's the deal with this entire pack: you never talk to LinkedIn's API directly, and you never see LinkedIn's notoriously fiddly authorization flow. The cost of that convenience is that nothing posts without a token and password set in SocialManPoster - and, for real publishing, a SocialMan subscription. Errors surface as dialog messages from the poster node, not here.

Practical advice is short because there's nothing to configure. Write the caption like a human, because LinkedIn is where "generated by AI" copy reads worst if you're posting professional content. Keep hashtags in the caption - there's no separate field - and keep it under LinkedIn's length limit, since the backend won't truncate. In the example workflow, this node sits right beside TwitterPosterData with a single text input each, which tells you everything about how the author thinks about these two: they're the low-effort posts you include for completeness when the real target is Facebook, Instagram, TikTok, or YouTube.

Installing it

Same pack as all the others - MLTask_ComfyUI. ComfyUI Manager: search MLTask_ComfyUI, install, restart; or:

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

Requirements are just requests and aiohttp; no model downloads, no heavy deps. To actually see it work you need a token from mltask.com/user/comfyui set in SocialManPoster - and when you're just testing, remember prepare_only on the poster stages a post without publishing it. That toggle is the difference between "trying the pack" and "posting to LinkedIn."