agentleFS
badgidsbadgids/comfyui-development-skills/skills/comfyui-workflows/SKILL.md

comfyui-workflows

Skill4 starsChanged 44 days agoMITView on GitHub

Use when creating, modifying, converting, composing, templating, or programmatically generating ComfyUI workflows in frontend/editor format or API/execution format, especially when custom nodes or install-specific model options are involved.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not mix these formats.

    line 21
  2. Distinctive:

    Search official templates/examples for a current comparable graph.

    line 40
  3. Distinctive:

    Do not assume a workflow downloaded from the internet is compatible with the target merely because it is official or recent.

    line 58
  4. Distinctive:

    Treat official templates as examples plus authored metadata.

    line 66
  5. Distinctive:

    Do not run an expensive workflow solely to prove structural validity unless the user asked for execution or the test cost is known to be …

    line 70
  6. Distinctive:

    Before returning a workflow as ready:

    line 74
  7. Distinctive:

    verify every required node and install-specific option from the target when reachable.

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

ComfyUI Workflows

A workflow is not valid merely because its JSON parses.

First identify the artifact

Determine whether the user needs:

  • frontend/editor workflow JSON;
  • API/execution graph;
  • workflow template/package data;
  • subgraph/blueprint content;
  • a conversion between representations.

Do not mix these formats.

Authority order

  1. live target node catalog for installed node definitions and options;
  2. current/version-matched workflow schema in official docs;
  3. official templates/examples for wiring patterns;
  4. target custom-node source when a third-party node requires clarification.

Current CLI guidance

When the task actually uses comfy-cli, inspect its current/version-matched help and bundled agent guidance before choosing CLI workflow commands or construction mechanisms. Treat that guidance as the CLI's operational surface, not as a replacement for the live target node catalog or workflow schema. Do not vendor the CLI skill text into this pack.

Required workflow construction loop

  1. Establish the target ComfyUI installation/version.
  2. Query live node definitions when the target is reachable.
  3. Confirm every required node class exists.
  4. Confirm input names/types, output slots/types, enum choices, and resource names from target data.
  5. Search official templates/examples for a current comparable graph.
  6. Build the requested representation using the current schema.
  7. Preserve unrelated metadata/layout when editing an existing workflow.
  8. Validate against the target using the comfyui-workflow-validation skill.
  9. Fix all deterministic validation errors before returning the workflow.

Hard prohibitions

Do not invent:

  • node types;
  • socket names;
  • output slot indexes;
  • widget order;
  • combo values;
  • model filenames;
  • custom-node package names.

Do not assume a workflow downloaded from the internet is compatible with the target merely because it is official or recent.

When the target is unavailable

You may construct a provisional workflow from current authoritative sources, but state that live compatibility could not be checked. Keep install-specific values easy to identify and replace.

Template use

Treat official templates as examples plus authored metadata. Validate them against the actual install before execution.

Execution

Do not run an expensive workflow solely to prove structural validity unless the user asked for execution or the test cost is known to be acceptable. Prefer preflight validation first.

Acceptance gate

Before returning a workflow as ready:

  • identify the requested workflow representation;
  • verify every required node and install-specific option from the target when reachable;
  • validate the workflow against the current/version-matched schema and live node catalog;
  • fix deterministic validation errors;
  • load, queue, or execute a minimal representative run against the target when practical;
  • preserve unrelated layout and metadata when editing an existing workflow;
  • label the workflow provisional when live target validation was not possible.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.