First Workflow

This guide walks through importing a ComfyUI workflow, promoting inputs, and running your first generation.

1) Add a ComfyUI Server

comfyclaw server add \
  --name "Local GPU" \
  --url http://127.0.0.1:8188 \
  --api-key "" \
  --default

You can manage multiple servers and switch per workflow.

2) Import a Workflow JSON

Export a workflow from ComfyUI (API JSON), then import it:

comfyclaw workflow import /path/to/workflow_api.json \
  --server <server-id> \
  --title "FLUX Text to Image" \
  --emoji "🖼️"

ComfyClaw parses the node graph and auto‑classifies input nodes.

3) Promote Primary Inputs

# See detected nodes
comfyclaw workflow inspect <workflow-id>

# Promote the most important inputs
comfyclaw workflow set-primary <workflow-id> <node-id>

Primary inputs appear first in the dashboard and API schema.

4) Run the Workflow

comfyclaw workflow run <workflow-id> \
  --input "13.text:a cat in space" \
  --wait --timeout 120

When --wait is provided, the output is downloaded into the local gallery.

5) Browse Gallery

comfyclaw gallery list
comfyclaw gallery download <output-id> --output ./result.png

The dashboard Gallery tab provides previews, filters, and downloads.