RunComfy

### App
RunComfy
https://docs.runcomfy.com/

### Sources
A ComfyUI workflow request completes or fails, using RunComfy deployment webhooks. Image and video generation completion can initially be handled by a Get Request Status action in a scheduled workflow.

### Actions
RunComfy (https://www.runcomfy.com) runs ComfyUI workflows in the cloud and provides AI image generation, AI video generation, and GPU LoRA training. Requested actions: browse AI image models and AI video models; get model input schema; generate an image or video with a supported model such as FLUX, Wan, Seedance, Seedream, or LTX; run a deployed ComfyUI workflow; get request status and output URLs; create and monitor a LoRA training job. Jobs are asynchronous, so submission should return the request or job ID, with status/results available separately. Please add RunComfy as an API-key app with per-user Bearer token authentication. API docs: https://docs.runcomfy.com/. Public MCP implementation: https://github.com/runcomfy-com/runcomfy-mcp. We represent RunComfy and can contribute the native components once the app/auth directory is created.

Start with the RunComfy API documentation and the linked public MCP implementation. Define the app/auth integration around Bearer-token authentication and asynchronous request IDs; done means the requested model, generation, workflow, status/output, and LoRA job actions are available with per-user credentials.

Written by the indexing model from the issue text.