yutianxiao6/comfyui-mcp / skills/comfyui-workflow
comfyui-workflow
VerifiedBuild, inspect, modify, validate, run, and troubleshoot ComfyUI workflows through ComfyUI MCP tools. Use for requests involving ComfyUI nodes, models, workflow JSON, graph connections, queue execution, generated outputs, canvas synchronization, LoRA, ControlNet, inpainting, upscaling, image-to-image, video, audio, 3D, or custom nodes.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.8 KB |
| agents/openai.yaml | config | 236 B |
| references/execution-and-files.md | markdown | 2.6 KB |
| references/graph-contract.md | markdown | 2.4 KB |
| references/troubleshooting.md | markdown | 2.7 KB |
| references/workflow-patterns.md | markdown | 2.7 KB |
references/execution-and-files.md
View on GitHub# Execution, files, queue, and canvas
## Run sequence
1. Validate the managed workflow with `live=true`.
2. Submit with `comfyui_run_workflow` and retain the returned `prompt_id`.
3. Query `comfyui_get_run` at reasonable intervals until it reports completed, failed, or unknown with enough evidence to act.
4. Report every returned output descriptor. Download only the files needed by the user or a subsequent operation.
`front=true` changes queue priority. Use it only when the user wants the run placed at the front; it can delay other users' work.
## One-off overrides
Use run overrides to change inputs for one execution without changing the stored graph:
```json
{
"6": {"text": "one-off prompt"},
"3": {"seed": 123456789}
}
```
Keys are node IDs and nested keys are exact input names. Discover or inspect the graph before forming overrides. Use persistent `comfyui_set_input` edits when future runs should inherit the value.
## Queue operations
- `comfyui_get_queue` is safe for inspection. Keep `include_workflows=false` unless full queued graph payloads are necessary.
- `comfyui_interrupt` stops the currently executing job and can discard in-progress computation.
- `comfyui_clear_queue` removes pending jobs, potentially including work submitted by someone else.
Never interrupt or clear as a generic retry step.
## Uploads
`comfyui_upload_image` accepts only local files allowed by the server's configured upload roots. Do not work around a denied path. Move or copy a file only when the user's request authorizes it, then upload from an allowed location.
Use the returned ComfyUI filename and subfolder exactly in `LoadImage`-style node inputs. Do not set `overwrite=true` unless replacement is intended.
## Outputs
Run results can include ComfyUI output descriptors with filename, subfolder, and type. Pass those exact values to `comfyui_download_output`. The server writes downloads into its managed output directory rather than an arbitrary destination.
If a run completes with no files, inspect whether the graph ends in a preview-only node, returns non-file data, or lacks a reachable output node before calling it a failure.
## Canvas synchronization
`comfyui_sync_canvas` sends a managed API graph to an open browser canvas through the optional bundled bridge. It is not required for headless execution.
- Use `mode="replace"` only when the current canvas should be replaced.
- If the bridge is unavailable, keep the managed workflow intact and tell the user how to install the bundled bridge.
- Canvas layout is a presentation concern; execution correctness still comes from API-format validation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |