Auto setup, ready to runany ComfyUI workflow
Upload your workflow JSON. RunComfy handles ComfyUI automatic install models that missing, resolves ComfyUI missing python nodes, and opens a ready-to-run ComfyUI cloud session in your browser. You don't need a separate ComfyUI node to auto download models, and you don't need a ComfyUI setup script or terminal.
- No setup scripts
- No missing nodes
- No model hunting
ComfyUI missing models, automatic install
A ComfyUI workflow JSON usually needs five to ten models, including a checkpoint, a few LoRAs, a ControlNet, a VAE, maybe an IP-Adapter or a text encoder. ComfyUI missing models is the most common reason a downloaded workflow won't run.
Upload your workflow JSON. RunComfy handles ComfyUI automatic install models that missing, including every checkpoint, LoRA, ControlNet, VAE, and text encoder. Each file goes into the correct ComfyUI/models/ sub-folder. Gated Hugging Face and CivitAI models work too; attach your token once.
ComfyUI missing python nodes, fixed automatically
Almost every downloaded workflow opens with red nodes and missing node types. ComfyUI missing python nodes is what you see when a custom node pack isn't installed or its Python dependencies don't match: import failed, missing module, or ComfyUI Manager installs the node but it still won't load.
Auto setup resolves every ComfyUI missing python node from your workflow JSON in a clean cloud environment. The graph opens fully wired, every time. No red nodes, no Manager retries, no manual GitHub hunting.
Run any ComfyUI workflow.json online
Upload any custom ComfyUI workflow JSON. After auto setup resolves ComfyUI missing models and ComfyUI missing python nodes, you get a ready-to-run ComfyUI cloud session in your browser, with every node loaded and every model in place. No local install, no GPU on your machine, no ComfyUI setup script to paste anywhere.
Save, share, and one-click deploy your auto-setup workflow
After you fix ComfyUI errors and get the workflow running, hit Cloud Save. RunComfy snapshots the full environment (workflow graph, custom nodes, models, and dependencies) into one reproducible image. Share it with a teammate via a single link; they open the exact same setup with no ComfyUI missing models or missing python nodes on their side.
The same snapshot supports one-click ComfyUI workflow deployment as a serverless ComfyUI API endpoint when you're ready to build a ComfyUI workflow into a product or batch job.
How it works
Upload your workflow JSON
Drag in any ComfyUI workflow.json. We parse the graph and show every missing custom node and missing model we detect.
Auto setup runs
We resolve ComfyUI missing python nodes, run ComfyUI automatic install models that missing, and fix dependencies on our side, not on your rented GPU while the meter runs.
Open the ready-to-run session
Click the link. ComfyUI opens in your browser with every node loaded and every model in place. Queue your prompt.
Stop wasting hours on ComfyUI setup. Start running.
Stop hunting for a ComfyUI node to auto download models or debugging ComfyUI missing python nodes on a rented GPU. Upload your workflow JSON. We handle ComfyUI missing models and missing nodes, then you open the link and run.
ComfyUI auto setup FAQ
Every workflow I download fails when I try to load it. What am I doing wrong?
Usually it’s ComfyUI missing models, ComfyUI missing python nodes, or missing dependency warnings after you load the workflow JSON. ComfyUI Manager often can’t fix them all. Upload the file to RunComfy for ComfyUI automatic install models that missing and a clean install of every custom node, then open a ready-to-run cloud session.
I have a workflow JSON file. How do I actually run it without setting everything up locally?
Upload the ComfyUI workflow JSON to RunComfy. Auto setup resolves ComfyUI missing python nodes, runs ComfyUI automatic install models that missing, and opens a ready-to-run ComfyUI cloud session. No ComfyUI setup script. No terminal.
ComfyUI won’t load my workflow at all because a node is missing. What now?
If a custom node is missing, ComfyUI sometimes won’t load the workflow at all, so you can’t even use Manager’s install-missing button. Upload the workflow JSON to RunComfy. We resolve ComfyUI missing python nodes from the right source and hand you back a session where the workflow actually loads.
Is there a way to fix all the missing nodes without installing everything manually?
You don’t install them yourself. Upload the workflow JSON and RunComfy resolves every ComfyUI missing python node automatically, with no missing node types, no red nodes, and no chasing forks on GitHub.
Why does ComfyUI Manager keep failing when I try to install missing custom nodes?
ComfyUI Manager often shows import failed, missing module, or installs a node that still won’t load because of a Python dependency issue or torch version conflict. It also doesn’t handle ComfyUI missing models. RunComfy builds the environment from your workflow JSON in a clean container instead.
Manager’s “Install Missing Custom Nodes” didn’t fix anything. What else should I try?
Here are the main options people use, and where each one falls short:
ComfyUI Manager (for an existing local install): click Install Missing Custom Nodes, and it scans your workflow JSON for missing custom nodes. Works for simple cases, but often fails on import failed errors, doesn’t download missing models, and can’t load the workflow if a node is missing in the first place.
ComfyUI-Launcher (local, zero-setup): exports workflows to a special launcher.json format and builds an isolated local environment. You still need to install and maintain the launcher on your machine.
ComfyRunner (headless / API): automates model and node downloads for API-format JSON. Built for backend scripting, not for opening a workflow in the ComfyUI UI.
RunComfy auto setup (cloud): upload a normal workflow JSON. We handle ComfyUI automatic install models that missing, resolve ComfyUI missing python nodes, and open a ready-to-run ComfyUI cloud session in your browser. No separate ComfyUI node to auto download models, no local setup ComfyUI workflow steps, no terminal.
Downloaded a workflow from CivitAI and I’m getting missing node types everywhere. How do I fix it?
CivitAI workflows often rely on custom node packs that aren’t in your local install. You’ll see red nodes or missing node types even after Manager tries. Upload the CivitAI workflow JSON to RunComfy. Auto setup finds ComfyUI missing python nodes and installs them, then opens the workflow fully wired in your browser.
Can I upload a workflow I grabbed from CivitAI, or does it need to be a plain export?
Yes. CivitAI workflows work the same way. Any ComfyUI workflow JSON is supported, whether it came from CivitAI, Reddit, Discord, or your own export.
I load the JSON and get a wall of missing dependency warnings. What does that even mean?
It means the workflow references custom nodes or Python packages your current ComfyUI install doesn’t have. Fixing it manually means hunting GitHub repos, matching torch/CUDA versions, and often hitting dependency hell on top of ComfyUI missing models and ComfyUI missing python nodes. Upload the workflow JSON to RunComfy and auto setup resolves those dependencies in an isolated cloud environment.
Will RunComfy download the missing models for me, or do I still need to go find every file?
Yes. RunComfy reads your ComfyUI workflow JSON and handles ComfyUI automatic install models that missing, including checkpoints, LoRAs, ControlNet, VAEs, IP-Adapter, upscalers, and text encoders, into the correct ComfyUI/models/ sub-folder. You don’t need to hunt for a separate ComfyUI node to auto download models; it’s built into auto setup when you upload the workflow.
Does auto setup handle LoRAs, ControlNet, VAEs, and all that, or just basic checkpoints?
Yes, all of them. Every standard ComfyUI model class is recognized and placed in the correct sub-folder so the workflow loads without path errors.
Some models need a Hugging Face login. Can I still run workflows that use gated models?
Add your Hugging Face token or CivitAI API key once. Auto setup uses it to fetch gated and private models for you.
I don’t want to install a bunch of custom nodes and break my local ComfyUI. Is there another way?
Custom nodes often disagree about Python, torch, xformers, or CUDA versions, the classic ComfyUI dependency hell. Instead of installing everything on your machine, upload the workflow JSON to RunComfy. Auto setup runs in a clean cloud environment with versions matched to the workflow, so your local ComfyUI install stays untouched.
I keep running into onnxruntime CUDA errors when I add certain nodes. Does this help with that?
RunComfy handles ComfyUI CUDA fixes automatically. Nodes that depend on onnxruntime CUDA ComfyUI builds get the matching CUDA toolkit in the cloud environment, so common onnxruntime needs the right CUDA errors don’t surface in your session.
Once the workflow is running, can I save it and share the whole setup with someone else?
Yes. After the workflow runs, use Cloud Save to snapshot the full environment (custom nodes, models, dependencies, and the graph) into one reproducible image. Share it with a link, or use one-click ComfyUI workflow deployment to ship it as a serverless ComfyUI API.
RunComfy es la principal ComfyUI plataforma, ofreciendo ComfyUI en línea entorno y servicios, junto con flujos de trabajo de ComfyUI con impresionantes imágenes. RunComfy también ofrece AI Models, permitiendo a los artistas aprovechar las últimas herramientas de AI para crear arte increíble.