what is comfyui-lmstudio fr?

This project connects ComfyUI, a visual tool for building AI image generation workflows, to LM Studio, an app that runs AI language and vision models directly on your own computer. It adds two building blocks you can drag into a ComfyUI workflow so you can use a locally running model for tasks like writing image captions or generating text, without sending anything to a cloud service or needing an API key. The first block connects to your running LM Studio server, checking right away that the server is reachable and that the model name you typed matches what is actually loaded, so you get a clear error immediately rather than a failure partway through a batch. The second block sends the actual request: you give it a prompt, and optionally an image if you want the model to describe or caption that picture, along with the usual controls like how creative the response should be. The result comes back as plain text that can be saved to a file or shown directly inside your workflow. The README spends real attention on one specific gotcha: some models write out a long hidden reasoning process before giving their final answer, and if you set the response length limit too low, the model runs out of room mid-thought and you get messy, half-finished text instead of a clean caption. It gives concrete guidance, suggesting a much higher limit for these reasoning-style models than for ordinary ones. Setup involves dropping the project folder into ComfyUI's custom nodes directory, restarting ComfyUI, and starting the server inside LM Studio itself. It only works with vision-capable models when captioning images, and currently only processes the first image in a batch, though the author notes that batch support could be added if requested.