Nice job. I wrote a custom MCP a while back that connects LLM agents to my headless server which runs Comfy so it can generate images using local models (includ

Nice job. I wrote a custom MCP a while back that connects LLM agents to my headless server which runs Comfy so it can generate images using local models (including BFL's Flux Dev models).