---
title: Use Cases
description: Common use cases for ComfyUI Skills for OpenClaw, including local image generation, multi-server routing, schema-based workflow reuse, and agent-managed setup.
permalink: /use-cases/
---

<section class="hero">
  <p class="eyebrow">Use Cases</p>
  <h1>Where this repository fits in a real ComfyUI and agent workflow.</h1>
  <p class="lede">
    These are the main user intents this project serves well. They also map
    directly to the keywords and search patterns people use to discover the repository.
  </p>
</section>

<div class="mini-grid">
  <section class="mini-card">
    <h3>OpenClaw image generation</h3>
    <p>Trigger existing ComfyUI workflows from chat instead of opening the graph every time.</p>
  </section>
  <section class="mini-card">
    <h3>Schema-based reuse</h3>
    <p>Expose a clean parameter surface instead of making the agent reason about every node.</p>
  </section>
  <section class="mini-card">
    <h3>Multi-server routing</h3>
    <p>Keep local and remote ComfyUI machines under one workflow namespace.</p>
  </section>
</div>

<div class="content-stack">
  <section class="section-card">
    <p class="eyebrow-label">Use Case 1</p>
    <h2>Turn a local ComfyUI workflow into a callable skill</h2>
    <p>
      The most direct use case is taking a workflow you already exported from ComfyUI,
      wrapping it with <code>schema.json</code>, and letting OpenClaw call it with
      natural-language-derived arguments. This is useful when the workflow logic is stable
      but the prompts or seeds change from request to request.
    </p>
  </section>

  <section class="section-card">
    <p class="eyebrow-label">Use Case 2</p>
    <h2>Route jobs across different ComfyUI servers</h2>
    <p>
      Many users have more than one generation target: a local GPU box, a remote workstation,
      or a higher-memory machine for heavier checkpoints. This repository lets you keep those
      workflows under a single skill system with server-level and workflow-level enable switches.
    </p>
    <p>
      The result is a cleaner namespace such as <code>local/sdxl-base</code> and <code>remote-a100/sdxl-base</code>
      without losing the original workflow identity.
    </p>
  </section>

  <section class="section-card">
    <p class="eyebrow-label">Use Case 3</p>
    <h2>Standardize what the agent is allowed to change</h2>
    <p>
      Raw ComfyUI graphs are powerful but noisy. The schema layer gives you a stable contract:
      which parameters are visible, which are required, which types are expected, and how they map
      into node fields. That makes the repository useful as a safety layer between an LLM and a graph.
    </p>
  </section>

  <section class="section-card">
    <p class="eyebrow-label">Use Case 4</p>
    <h2>Let an agent manage configuration for you</h2>
    <p>
      Because the project is structured around predictable files and a single CLI tool (<code>comfyui-skill</code>),
      any coding agent can add servers, import workflows, check dependencies, and verify the setup.
      This is especially helpful when moving the same skill pack across machines using <code>config export</code> and <code>config import</code>.
    </p>
    <div class="card-cta">
      <a class="button primary" href="{{ '/getting-started/' | relative_url }}">Start Setup</a>
      <a class="button secondary" href="{{ '/faq/' | relative_url }}">Read FAQ</a>
    </div>
  </section>
</div>