ComfyUI Node

DeployCash

DeployCash sells your workflow as a mini-app — but your GPU does the serving

By DeployAI365·Created 2 years ago·Updated about a year ago· 0
DeployCash
  • app_img1(optional)
  • app_img2(optional)
  • app_img3(optional)
  • custom_img1(optional)
  • custom_img2(optional)
  • custom_img3(optional)
  • custom_video1(optional)
  • custom_video2(optional)
  • custom_video3(optional)
    app_title这是默认作品标题,请在comfyui中修改
    app_desc这是默认功能介绍,请在comfyui中修改
    app_fee18
    free_times0
    custom_text1(optional)
    custom_text2(optional)
    custom_text3(optional)
    custom_img1_desc请上传图片
    custom_img2_desc请上传图片
    custom_img3_desc请上传图片
    custom_video1_desc请上传视频
    custom_video2_desc请上传视频
    custom_video3_desc请上传视频
    custom_text1_desc请输入文本
    custom_text2_desc请输入文本
    custom_text3_desc请输入文本

    The name promises a one-click cloud deploy. The reality is much more interesting, and much more local: DeployCash is the config node of a Chinese "变现" (monetization) platform that turns your own ComfyUI install into a paid mini-app. Buyers open a WeChat-style app page, run your workflow, pay per use - and the inference happens on your GPU, not some cloud. It's a real business model in the China market, and it's a paperweight (that phones home) if you're not part of that ecosystem. There's no middle ground.

    The DeployCash node itself computes nothing. It has no FUNCTION, no outputs - it's a form. When you queue the workflow, the pack reads this node's inputs straight out of the submitted prompt JSON, verifies them, and uploads your listing to the platform backend at depcash.deployai365.com. You fill it in like you'd fill in an app-store listing, because that's literally what it is.

    The inputs that matter

    • app_title and app_desc - the title and description buyers see on the listing. The defaults are placeholder Chinese ("please change this in ComfyUI"), so you can't accidentally publish with no copy.
    • app_fee - the price, an integer with a default of 18 (presumably the platform's currency unit). 0 to 999999.
    • free_times - how many free runs a buyer gets before the meter starts. Default 0, which is a choice.
    • app_img1 / app_img2 / app_img3 - your gallery. Feed these from a LoadImage or from an image-producing node. The pack's uploader (extract_and_verify_images) refuses to publish if the referenced image files don't actually exist in your input folder - it checks on every queue, so a missing image isn't a silent failure.
    • custom_img1–3, custom_video1–3, custom_text1–3 - the configurable slots buyers interact with. Each has a matching _desc field (default "please upload image" / "please input text") that labels the slot in the UI. The custom_text slots are force-input strings, so wire them from DeployCash_textInput rather than typing.

    How it actually works

    This is where you need to trust the platform, because the pack is built around a tunnel. On import it registers an HTTP route on your ComfyUI server, opens a persistent outbound websocket to the platform's relay, and exposes a local websocket to receive relayed tasks. End-user runs get forwarded to your machine and executed with your VRAM. Your instance is identified by a subdomain hashed from your MAC address and port, and a WeChat QR-code login in the frontend binds your account and stores a token on disk (config/hash/). The pack even scaffolds a bundled sdc relay binary and admin panel in the source - but neither ships in the repo, so treat that plumbing as platform-internal and expect to be talking to their service either way.

    Install

    The README is empty, so this is all you get. Via Manager: search ComfyUI_DeployCash and install, then restart. Or manually:

    cd ComfyUI/custom_nodes
    git clone https://github.com/jacklukai/ComfyUI_DeployCash
    

    Restart ComfyUI. On first import, install.py runs pip install -r requirements.txt for you - just websockets and aiohttp. No model files, no heavy ML dependencies. The weight is on the network side, not your disk.

    The honest troubleshooting section

    • There is no documentation. The README is empty and there's zero community signal in English - you won't find forum threads, because the audience is China-market and the footprint outside it is nil.
    • It phones home on import. Persistent websockets to a third-party server, a persistent client ID derived from your MAC address, tokens written to disk. That's the same category of attack surface the ComfyUI community got burned by in the LLMVISION incident - and this ecosystem still has no code review or sandboxing. Don't install this on a machine you share, and absolutely not on a publicly reachable ComfyUI instance.
    • If their backend is unreachable, the websocket code loops with reconnect delays and logs "连接成功" on success - Chinese log messages, so keep that in mind if you're grepping your terminal.

    Reach for DeployCash if you're actually selling workflows through the platform. If you just saw "deploy" in the name and wanted cloud hosting: close the tab. This node only publishes to one walled garden, and it needs your hardware to stay online to serve.

    CategoryDeployCash

    Inputs (25)

    NameTypeDefaultDescription
    app_titleSTRING这是默认作品标题,请在comfyui中修改
    app_descSTRING这是默认功能介绍,请在comfyui中修改
    app_feeINT180–999999
    free_timesINT00–999999
    app_img1(optional)optIMAGE
    app_img2(optional)optIMAGE
    app_img3(optional)optIMAGE
    custom_img1(optional)optIMAGE
    custom_img2(optional)optIMAGE
    custom_img3(optional)optIMAGE
    custom_video1(optional)optIMAGE
    custom_video2(optional)optIMAGE
    custom_video3(optional)optIMAGE
    custom_text1(optional)optSTRING
    custom_text2(optional)optSTRING
    custom_text3(optional)optSTRING
    custom_img1_descoptSTRING请上传图片
    custom_img2_descoptSTRING请上传图片
    custom_img3_descoptSTRING请上传图片
    custom_video1_descoptSTRING请上传视频
    custom_video2_descoptSTRING请上传视频
    custom_video3_descoptSTRING请上传视频
    custom_text1_descoptSTRING请输入文本
    custom_text2_descoptSTRING请输入文本
    custom_text3_descoptSTRING请输入文本

    Outputs (0)

    No outputs